alhosn university journal of engineering & applied sciences
alhosn university journal of engineering & applied sciences
alhosn university journal of engineering & applied sciences
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
ISSN 2076-8516<br />
www.<strong>alhosn</strong>u.ae<br />
ALHOSN UNIVERSITY<br />
JOURNAL OF<br />
ENGINEERING<br />
& APPLIED<br />
SCIENCES<br />
CONTENTS<br />
Power Control-based Channel Assignment<br />
Algorithm for Wireless Mesh networks<br />
Aizaz U. Chaudhry, Roshdy H.M. Hafez,<br />
Osama Aboul-Magd<br />
Characterization <strong>of</strong> Elastic Buckling <strong>of</strong><br />
Cylindrical Liners Using Finite Element<br />
Approach<br />
A.M. Sweedan, K.M. El-Sawy<br />
Optimized 3D Mesh for the Finite Element<br />
Analysis <strong>of</strong> Laminated Composite Plate<br />
with Central Hole<br />
Khaled M. El-Sawy, Abdul Hameed<br />
Arwani<br />
Affordability, Enablement and Sustainability<br />
<strong>of</strong> Low-income Housing<br />
Mahbubur Rahman<br />
Drilling process Robust Optimization for<br />
6061 Aluminum Using Desirability<br />
Functions<br />
Mahmoud I. Awad, Elsayed Orady,<br />
Chenmaya Dandekar<br />
Neighbourhood Design and Walkability: A<br />
synthesis from Planning, Design, Transportation<br />
and Environmental Health fields<br />
Mohammed Kashef<br />
Qos Improvements in Mobile WiMAX<br />
Networks<br />
P. Rengaraju, C.H. Lung, A. Srinivasan,<br />
R.H.M. Hafez<br />
The Solar-Wind-EV Paradigm: A Strategic<br />
Policy for a Sustainable Energy Future<br />
Zuwei Yu, Tarik Aouam
ALHOSN UNIVERSITY JOURNAL<br />
OF ENGINEERING AND APPLIED SCIENCES<br />
ADVISORY BOARD<br />
(in alphabetical order)<br />
Pr<strong>of</strong>. Ghassan Aouad<br />
Salford University, UK<br />
Pr<strong>of</strong>. Goodarz Ahmadi<br />
Clarkson University, USA<br />
Pr<strong>of</strong>. Hisham Elkadi<br />
University <strong>of</strong> Ulster, UK<br />
Pr<strong>of</strong>. Jamal A. Abdalla<br />
American University <strong>of</strong> Sharjah, UAE<br />
Dr. Khaled El-Sawy<br />
United Arab Emirates University, UAE<br />
Dr. Mohamed Lachemi<br />
Ryerson University, Canada<br />
Pr<strong>of</strong>. Mufid Abdul Wahab Samarai<br />
Sharjah University, UAE<br />
Pr<strong>of</strong>. Nizar Al-Holou<br />
University <strong>of</strong> Detroit Mercy, USA<br />
Pr<strong>of</strong>. Riadh Al-Mahaidi<br />
Monash University, Austrialia<br />
Pr<strong>of</strong>. Sadik Dost<br />
University <strong>of</strong> Victoria, Canada<br />
Pr<strong>of</strong>. Ziad Saghir<br />
Ryerson University, Canada<br />
2
ISSN 2076-8516<br />
ALHOSN UNIVERSITY<br />
JOURNAL OF<br />
ENGINEERING<br />
& APPLIED<br />
SCIENCES<br />
A bi annual, refereed <strong>journal</strong> published by<br />
ALHOSN University - Abu Dhabi - UAE<br />
Volume 3 Number 1 July 2010<br />
Chairman<br />
Dr. Nasser Bin Saif Al Mansoori<br />
Editor<br />
Pr<strong>of</strong>. Abdul Rahim Sabouni<br />
Associate Editor<br />
Dr. Hamdi Sheibani<br />
Members<br />
Dr. Adel Khelifi<br />
Dr. Adnan Husnein<br />
Dr. Abdelaziz Soufyane<br />
Dr. Naima Benkari<br />
Managing Editor<br />
Dr. Al Haj Salim Mustafa<br />
Address:<br />
P.O. Box : 38772<br />
Abu Dhabi - UAE<br />
Tel. : +971 2 4070700<br />
Fax : +971 2 4070799<br />
E-mail : eng<strong>journal</strong>@<strong>alhosn</strong>u.ae<br />
Website : www.<strong>alhosn</strong>u.ae<br />
3
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) 2010<br />
© 2009 ALHOSN University<br />
CONTENTS<br />
Power Control-based Channel Assignment Algorithm for Wireless Mesh networks 7<br />
Aizaz U. Chaudhry, Roshdy H.M. Hafez, Osama Aboul-Magd<br />
Characterization <strong>of</strong> Elastic Buckling <strong>of</strong> Cylindrical Liners Using Finite Element Approach 17<br />
A.M. Sweedan, K.M. El-Sawy<br />
Optimized 3D Mesh for the Finite Element Analysis <strong>of</strong> Laminated Composite Plate with 31<br />
Central Hole<br />
Khaled M. El-Sawy, Abdul Hameed Arwani<br />
Affordability, Enablement and Sustainability <strong>of</strong> Low-income Housing 51<br />
Mahbubur Rahman<br />
Drilling process Robust Optimization for 6061 Aluminum Using Desirability Functions 75<br />
Mahmoud I. Awad, Elsayed Orady, Chenmaya Dandekar<br />
Neighbourhood Design and Walkability: A synthesis from Planning, Design, 87<br />
Transportation and Environmental Health fields<br />
Mohammed Kashef<br />
Qos Improvements in Mobile WiMAX Networks 107<br />
P. Rengaraju, C.H. Lung, A. Srinivasan, R.H.M. Hafez<br />
The Solar-Wind-EV Paradigm: A Strategic Policy for a Sustainable Energy Future 119<br />
Zuwei Yu, Tarik Aouam<br />
5
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 7-15 (2010)<br />
© 2009 ALHOSN University<br />
POWER CONTROL-BASED CHANNEL ASSIGNMENT<br />
ALGORITHM FOR WIRELESS MESH NETWORKS<br />
Aizaz U. Chaudhry 1 , Roshdy H. M. Hafez 1* , Osama Aboul-Magd 1<br />
1 Department <strong>of</strong> Systems and Computer Engineering, Carleton University, Ottawa, Canada<br />
ABSTRACT: The assignment <strong>of</strong> channels to radios in a multi-radio mesh network is a challenging task. In this<br />
paper, we propose a channel assignment algorithm, TICA (Topology-controlled Interference-aware Channelassignment<br />
Algorithm), which significantly improves network throughput by minimizing interference within the<br />
mesh network using a novel approach <strong>of</strong> controlling the network topology based on power control before<br />
intelligently assigning the channels to the multi radio mesh routers as well as guaranteeing network connectivity.<br />
Key words: Channel assignment algorithm; multi-radio multi-channel; power control; topology control algorithm.<br />
1. INTRODUCTION<br />
In traditional Wireless Local Area Networks (WLANs), each Access Point (AP) is connected to<br />
the wired network whereas in Wireless Mesh Networks (WMNs), only a subset <strong>of</strong> APs is<br />
connected to the wired network. An AP that is connected to the wired network is called gateway<br />
(GW). APs that do not have wired connections are called mesh routers (MRs) and they connect to<br />
the gateway through multiple hops. Similar to routers in wired networks, mesh routers in a WMN<br />
forward each other’s traffic in order to establish and maintain their connectivity [1].<br />
The IEEE 802.11a standard provides 12 non-overlapping frequency channels. A frequency reuse<br />
pattern similar to those used in cellular networks can be used to assign channels to MRs.<br />
Additionally, if a MR is equipped with more than one transceiver, we can assign multiple channels<br />
to the same router within the interference range. In a large WMN, the total number <strong>of</strong> radios<br />
(transceivers) is higher than the number <strong>of</strong> available channels. So, many links between the MRs in<br />
the WMN operate on the same set <strong>of</strong> channels and interference among transmissions on these<br />
channels decreases their utilization. Therefore, minimizing the effect <strong>of</strong> interference is required for<br />
the efficient reuse <strong>of</strong> the scarce radio spectrum. Therefore, the key issue in a Multi-Radio Multi-<br />
Channel (MRMC) WMN architecture is the Channel Assignment (CA) problem which involves<br />
assigning each radio to a channel in such a way that minimizes interference on any given channel<br />
as well as guarantees network connectivity [2].<br />
This paper addresses the channel assignment problem by proposing a centralized CA algorithm.<br />
The innovative element in the proposed CA is controlling the network topology using a power<br />
control design. The main trade <strong>of</strong>f is as follows: If all radios transmit at maximum power , then<br />
the network will have high connectivity but it will also have high interference and low throughput,<br />
on the other hand, if each radio controls its power, then the network connectivity will be less with<br />
lower interference and better throughout. Therefore, power control can give us a mechanism to<br />
trade <strong>of</strong>f connectivity with interference as illustrated in this paper. Specifically, the contributions<br />
<strong>of</strong> this work are as follows: “A new Topology-controlled Interference-aware Channel-assignment<br />
Algorithm (TICA) which intelligently assigns the available non-overlapping 802.11a frequency<br />
_________________________<br />
* Corresponding Author.<br />
E-mail : hafez@sce.carleton.ca<br />
7
AIZAZ U. CHAUDHRY, ROSHDY H.M. HAFEZ AND OSAMA ABOUL-MAGD<br />
channels to the mesh nodes with the objective <strong>of</strong> minimizing interference and, thereby, improving<br />
network throughput”.<br />
The rest <strong>of</strong> the paper is organized as follows. Section II presents the related work. Network<br />
architecture for the proposed model is presented in section III. Section IV explains the channel<br />
assignment problem and sections V and VI discuss the proposed topology control and channel<br />
assignment algorithms. Performance evaluation and results are given in section VII. The paper<br />
finally concludes in section VIII.<br />
2. BACKGROUND<br />
A neighbor-based topology control scheme in [3] is based on a Local Minimum Spanning Tree<br />
Algorithm. A location-based topology control scheme, proposed in [4], is based on an Enhanced<br />
Local Minimum Shortest-Path Tree algorithm. In centralized CA schemes such as Centralized<br />
Hyacinth (C-HYA) [5] and Traffic and Interference Aware Channel Assignment Scheme<br />
(MesTiC) [2], the traffic load is required to be known before assigning channels whereas our<br />
proposed algorithm requires no such knowledge. The Hybrid Multiple Channel Protocol (HMCP)<br />
proposed in [6] requires radios to switch between channels on a per-packet basis. In such cases,<br />
time synchronization and coordination between mesh nodes is required which is not needed in our<br />
proposed algorithm. The Breadth First Search - Channel Assignment (BFS-CA) scheme proposed<br />
in [7] requires certain number <strong>of</strong> MRs with certain number <strong>of</strong> radio interfaces to be placed at<br />
certain hops from gateway whereas our proposed algorithm simply requires all MRs to have four<br />
data radios, does not require any careful router placement strategy and works with any placement<br />
<strong>of</strong> routers as verified by the performance evaluation.<br />
3. NETWORK ARCHITECTURE<br />
We propose a model where each MR is equipped with five radios (i.e. each MR can<br />
communicate over 5 independent channels using five miniature transceivers). The model assumes<br />
that all radios conform to the IEEE802.11a standard and they work in the 5 GHz band. The MR<br />
dedicates one <strong>of</strong> the 5 radios for control and uses the remaining four radios for communications.<br />
IEEE902.11a has 12 non-overlapping channels. We dedicate channel # 12 to the control<br />
functions at all nodes. We define the term: Maximum Node Degree (MND) to denote the<br />
maximum number <strong>of</strong> independent radios dedicated for data communications with neighboring<br />
nodes. In our model the MND is 4. According to the work reported in [5], having a total <strong>of</strong> 11<br />
channels, the use <strong>of</strong> MND=4 would result in the highest possible throughput and that higher MND<br />
will not increase the throughput.<br />
4. CHANNEL ASSIGNMENT PROBLEM<br />
Given the connectivity graph, the main challenge for CAA is how to assign a channel to each<br />
radio in a way to minimize interference between MRs and ensure network connectivity. In order to<br />
achieve this, the CAA should satisfy the following constraints:<br />
In order to communicate, two nodes within TR <strong>of</strong> each other need to have a common channel<br />
assigned to their end-point radios<br />
8
POWER CONTROL-BASED CHANNEL ASSIGNMENT ALGORITHM FOR WIRELESS MESH NETWORKS<br />
<br />
<br />
<br />
<br />
Links in direct IR <strong>of</strong> each other should be assigned non-overlapping channels<br />
The number <strong>of</strong> distinct channels that can be assigned to a MR is bounded by the number <strong>of</strong><br />
radios it has<br />
The total number <strong>of</strong> non-overlapping channels is fixed<br />
Since traffic in a WMN is directed to and from the gateway, so traffic flows are likely to<br />
aggregate at routers close to gateway. Links that are expected to support heavy traffic should<br />
be given more bandwidth than others. So, these links should use a channel that is shared by a<br />
fewer number <strong>of</strong> nodes. Hence, priority in channel assignment should be given to links starting<br />
from the gateway based on number <strong>of</strong> nodes that use a link to reach the gateway.<br />
5. TOPOLOGY CONTROL ALGORITHM (TCA)<br />
Initially, the gateway broadcasts a “Hello” message on the control channel, announcing itself as<br />
the gateway. Each MR that receives this hello message on the control channel over its control<br />
radio broadcasts it again and it is flooded throughout the network. The hello message contains a<br />
hop-count field that is incremented at each hop during its broadcast. So, a MR may receive<br />
multiple copies <strong>of</strong> this message over its control radio but distance <strong>of</strong> a mesh router from gateway<br />
is the shortest path length (shortest hop count) <strong>of</strong> hello message received by MR through its<br />
control radio over different paths. In this way, each MR knows the next hop to reach the gateway<br />
using its control radio.<br />
Our proposed TCA controls the network topology by selecting the nearest neighbors for each<br />
node in the network with the objective <strong>of</strong> minimizing interference among mesh routers. The<br />
proposed TCA is based on the following assumptions:<br />
A. Assumptions<br />
All nodes start with the maximum transmission power.<br />
Each node has its location information.<br />
Each node uses an omni-directional antenna for both transmission and reception.<br />
Each node is able to adjust its own transmission power.<br />
The maximum transmission power is the same for all nodes.<br />
The maximum TR for any two nodes to communicate directly is also the same.<br />
The initial topology graph created, when every node transmits with maximum power, is<br />
strongly connected.<br />
B. Phases <strong>of</strong> TCA<br />
1) Exchange <strong>of</strong> information between nodes<br />
In the first exchange, each node broadcasts a Hello message at maximum power containing its<br />
node id and position.<br />
2) Building the Maximum Power Neighbor Table (MPNT)<br />
From the information in the received Hello messages, each node arranges its neighboring nodes<br />
in ascending order <strong>of</strong> their distance. The result is MPNT. Then, each node sends its MPNT along<br />
with its position and node id to the gateway using its control radio.<br />
9
AIZAZ U. CHAUDHRY, ROSHDY H.M. HAFEZ AND OSAMA ABOUL-MAGD<br />
3) Building the Direct Neighbor Table (DNT)<br />
For each node in the network, the gateway builds a DNT. Based on information in the<br />
neighbor’s MPNT and the total number <strong>of</strong> nodes in the MPNT <strong>of</strong> a node v, gateway eliminates a<br />
node w from MPNT <strong>of</strong> node v if node w is closer to any other node y in the MPNT <strong>of</strong> node w than<br />
MRs send MPNT to GW<br />
GW runs TCA & builds connectivity graph<br />
Connected<br />
GW checks resulting<br />
network for connectivity<br />
Not connected<br />
GW builds SPT based on minimum power<br />
GW moves to a<br />
higher TCA<br />
For each link in SPT, GW builds link ranking<br />
GW assigns 11 available<br />
channels to 11 highest ranked<br />
links<br />
For 12 th ranked link and onwards,<br />
GW checks channel assignment<br />
within its IR neighborhood<br />
If any node in the SPT<br />
has MND > 4<br />
Connected<br />
No<br />
Yes<br />
GW keeps four links with the<br />
minimum weight<br />
Yes<br />
If GW finds channels that are not<br />
assigned to any link within IR <strong>of</strong> a link<br />
No<br />
Not connected<br />
GW checks resulting<br />
network for connectivity<br />
GW assigns channel with<br />
highest channel number<br />
GW selects a LIC<br />
GW moves to a<br />
higher TCA<br />
Similarly, GW assigns<br />
channels to all links<br />
Figure 1. Topology-controlled Interference-aware Channel-assignment Algorithm (TICA)<br />
to node v.<br />
If after removing nodes from MPNT <strong>of</strong> node v, the remaining number <strong>of</strong> nodes in MPNT <strong>of</strong><br />
node v is equal to “x-1”, then gateway selects “x” nearest nodes as neighbors <strong>of</strong> node v which<br />
results in the DNT. However, after removing nodes from MPNT <strong>of</strong> node v, if the remaining<br />
number <strong>of</strong> nodes is greater than or equal to “x”, the result is DNT. We call the above algorithm as<br />
select x for less than x TCA where x is a positive integer.<br />
4) Converting into bi-directional links<br />
For each node in the network, the gateway converts the uni-directional links in the DNT <strong>of</strong> a<br />
node into bi-directional links. For each uni-directional link, this is done by adding a reverse link in<br />
the DNT <strong>of</strong> the neighboring node. This converts the DNT into Bi-directional DNT. This results in<br />
the Final Neighbor Table (FNT).<br />
5) Calculating the minimum power required<br />
For each node in the network, gateway calculates minimum power required to reach each <strong>of</strong> the<br />
nodes in FNT <strong>of</strong> a node, using appropriate propagation model formulas. If distance between two<br />
nodes u and v is less than cross over distance, i.e. d( u,v ) < Cross_over_dist, Free Space model is<br />
used whereas if d( u,v ) > Cross_over_dist, Two-ray model is used. Cross over distance is<br />
calculated using following expression [8]:<br />
4<br />
hth r<br />
Cross _ over _ dist <br />
<br />
<br />
10
POWER CONTROL-BASED CHANNEL ASSIGNMENT ALGORITHM FOR WIRELESS MESH NETWORKS<br />
where h t and h r are the antenna heights <strong>of</strong> the transmitter and receiver respectively.<br />
The free-space propagation model is given by following expression [8]:<br />
2<br />
RxThresh(4d<br />
)<br />
(2)<br />
P <br />
min<br />
2<br />
G t<br />
G r<br />
<br />
The two-ray model is given by the following expression [8]:<br />
4<br />
RxThresh(<br />
d )<br />
(3)<br />
Pmin<br />
<br />
2 2<br />
GtGrht<br />
hr<br />
where G t and G r are transmitter and receiver antenna gains respectively. RxThresh is the power<br />
required by radio interface <strong>of</strong> the receiving node to correctly receive the message.<br />
6. CHANNEL ASSIGNMENT ALGORITHM (CAA)<br />
The problem <strong>of</strong> assigning channels to links in a mesh network is NP-hard based on its mapping<br />
to graph-coloring problem [5] and standard graph-coloring algorithms can not satisfy all <strong>of</strong> its<br />
constraints. Also, the CA problem for mesh networks looks similar to the list coloring problem<br />
which is NP-complete [9]. Therefore, we propose an approximate algorithm for channel<br />
assignment. Our proposed algorithm, TICA, as shown in Figure 1, has following phases:<br />
A. Topology Control<br />
In order to create the network connectivity graph with the aim <strong>of</strong> reducing the interference<br />
between MRs, network topology is controlled using power control at each MR. All nodes send<br />
their MPNTs to gateway using their control radio. Note that in order to send its MPNT to gateway,<br />
each MR knows the next hop to reach gateway using its control radio via gateway advertisement<br />
process. Gateway starts with the select 1 for less than 1 TCA and builds FNTs for all nodes.<br />
B. Connectivity Graph<br />
Based on the information <strong>of</strong> the FNTs <strong>of</strong> all nodes, gateway builds the connectivity graph. The<br />
gateway checks the resulting network for connectivity to ensure that it can reach any node in the<br />
SPT directly or through intermediate hops. If resulting network is not connected, gateway moves<br />
to a higher TCA by incrementing x in the select x for less than x TCA.<br />
C. Minimum Power based SPT with a MND <strong>of</strong> 4<br />
After ensuring that the connectivity graph is connected, the gateway builds the SPT based on<br />
the connectivity graph. The metric for path selection is minimum power. While building the SPT,<br />
gateway ensures that each node can have only 4 TR neighbors and builds a SPT with a MND <strong>of</strong> 4<br />
per node. If any node in the SPT has more than four links, gateway selects those four links for that<br />
node which have the minimum weight and sets weight <strong>of</strong> all other links to infinity. The gateway<br />
checks the resulting SPT graph for connectivity. If resulting SPT graph is not connected, gateway<br />
moves to a higher TCA.<br />
D. Link Ranking<br />
To assign channels to links <strong>of</strong> SPT, gateway calculates the rank <strong>of</strong> each link. If l is the link and<br />
x is node using link l, then rank <strong>of</strong> link l, i.e. r l , is given by:<br />
r<br />
l<br />
I x , l<br />
(4) I x,l is 1 if node x is using link l and 0<br />
x<br />
otherwise. In case <strong>of</strong> links with same rank, link whose power <strong>of</strong> farthest node to gateway is<br />
11
AIZAZ U. CHAUDHRY, ROSHDY H.M. HAFEZ AND OSAMA ABOUL-MAGD<br />
smallest is given highest rank. If there are still links with same rank, link with smallest node ids is<br />
given highest rank.<br />
E. Channel Assignment<br />
The algorithm then assigns a channel to each link according to its rank and it begins with<br />
assigning the 11 available channels to the 11 highest ranked links such that channel 1 is assigned<br />
to the first ranked link.<br />
For the 12th ranked link and onwards, gateway checks the channel assignment <strong>of</strong> all the links<br />
within IR <strong>of</strong> both nodes that constitute that link. If gateway finds one or more channels among the<br />
11 available channels that are not assigned to any link within IR neighborhood <strong>of</strong> the 12th ranked<br />
link, it assigns that channel from the unassigned channels to the 12th ranked link which has the<br />
highest channel number.<br />
1) Least Interfering Channel (LIC)<br />
If gateway cannot find any channel, among the 11 available channels, that is not assigned to<br />
any link within IR neighborhood <strong>of</strong> the 12th ranked link, it selects a LIC and assigns it to that link.<br />
2) Interference Level(IL)<br />
To find out LIC, gateway builds IL for all 11 channels. LIC is the channel with minimum IL<br />
which means that assigning this channel to the 12th ranked link results in minimum interference in<br />
the network.<br />
In order to build IL for channel one, gateway finds all links within IR <strong>of</strong> either <strong>of</strong> the two nodes<br />
that constitute the 12th ranked link that use channel one and calculates IL <strong>of</strong> each link based on its<br />
rank and distance from the nodes that constitute the 12th ranked link. It sums up the individual ILs<br />
<strong>of</strong> all links, that use channel one within the IR <strong>of</strong> either <strong>of</strong> the two nodes that constitute the 12th<br />
ranked link, to find out the total IL for channel one. This is done by using the following equation:<br />
( IL)<br />
i<br />
<br />
<br />
<br />
r <br />
1 <br />
<br />
<br />
<br />
R d<br />
<br />
<br />
<br />
1<br />
<br />
<br />
<br />
r <br />
1 <br />
<br />
<br />
<br />
R d<br />
<br />
<br />
<br />
2<br />
<br />
r <br />
1 <br />
....... <br />
<br />
<br />
<br />
R d<br />
<br />
<br />
<br />
<br />
n<br />
(5) where IL is interference level <strong>of</strong> channel i, r<br />
is rank <strong>of</strong> the link using channel i, R is maximum rank assigned to a link in SPT, d is distance <strong>of</strong><br />
either <strong>of</strong> nodes on link using channel i to either <strong>of</strong> the two nodes that constitute 12th ranked link, <br />
is 2 or 4 depending on cross over distance, i is the channel and has value between 1 and 11, n is<br />
link using channel i which is within IR <strong>of</strong> either <strong>of</strong> the two nodes that constitute 12th ranked link.<br />
If a link is emanating from either <strong>of</strong> the two nodes that constitute the 12th ranked link and a<br />
channel has been assigned to that link, then IL for this channel is set to infinity. LIC is selected<br />
using following equation:<br />
IL LIC<br />
minIL1 , IL2<br />
,........, IL11<br />
(6)<br />
Similarly, the gateway assigns channels to all the links in the SPT. Using its control radio,<br />
gateway then sends each mesh node the CA message. For each channel assigned to a MR, channel<br />
assignment contains the channel number and the neighbor node to communicate with using this<br />
channel. The CA message also contains the next hop to reach gateway. Based on channel assigned<br />
to a MR to communicate with a neighbor and its distance to that neighbor, MR applies power<br />
control and adjusts its transmission power accordingly.<br />
12
POWER CONTROL-BASED CHANNEL ASSIGNMENT ALGORITHM FOR WIRELESS MESH NETWORKS<br />
7. PERFORMANCE EVALUATION<br />
The performance <strong>of</strong> our proposed algorithm, TICA, for MRMC WMNs is compared against a<br />
“Single-Radio Single-Channel” (SRSC) scheme and a “Common Channel Assignment” (CCA)<br />
scheme for multi-radio mesh nodes. In CCA scheme [2], all MRs have four radio interfaces. The<br />
first radio on each MR is tuned to first non-overlapping channel; the second radio is tuned to<br />
second non-overlapping channel and so on. Note that in CCA and SRSC schemes, all nodes<br />
transmit with same the maximum power and use AODV (Ad hoc On-Demand Distance Vector)<br />
routing protocol [10].<br />
F. Simulation Environment<br />
For the performance evaluation via throughput analysis, NS2 (version 2.30) [11] simulation<br />
tool is used. Multi-interface wireless mesh nodes are created in NS2 by modifying the original<br />
node model in NS2 using the procedure given in [12].<br />
All radios are IEEE 802.11a radios and support 12 channels. As per the IEEE 802.11 standard<br />
[13], RxThresh is set equal to -65 dBm in order to achieve a maximum data rate <strong>of</strong> 54 Mbps<br />
supported by IEEE 802.11a. In order to achieve a strongly connected topology, the maximum<br />
transmission power for all the radios is set to 27 dBm. RTS/CTS is disabled.<br />
G. Network Topologies<br />
Three topologies are used in the evaluation and each consists <strong>of</strong> 36 MRs distributed in an area<br />
<strong>of</strong> 500 m × 500 m. Topology 1 is a grid and topology 2 is randomly generated while in topology 3,<br />
called controlled random, physical terrain is divided into cells and a MR is placed randomly in<br />
each cell.<br />
H. Simulation Parameters in NS2<br />
Based on CA by gateway, IEEE 802.11a channels are assigned to links between MRs and<br />
transmission power for each radio <strong>of</strong> each MR is set accordingly. The physical (PHY) and MAC<br />
(Medium Access Control) layer settings used for simulation are shown in Table I and Table II,<br />
respectively.<br />
All MRs at the periphery <strong>of</strong> network send traffic to gateway simultaneously. So, this represents<br />
scenario in which multiple flows within mesh network interfere with each other. Each <strong>of</strong> these<br />
nodes generates an 8 Mbps Constant Bit Rate UDP traffic stream consisting <strong>of</strong> 1024 byte packets.<br />
They start at 1 second and do not stop until the end <strong>of</strong> simulation (600 seconds).<br />
Table I. PHY Layer Configuration in NS2<br />
Physical Layer Parameters<br />
Settings<br />
TX/RX Antenna Height (m) 3<br />
Gain <strong>of</strong> TX/RX Antenna (dB) 1<br />
Packet Capture Threshold (dB) 10<br />
Packet Reception Threshold (watts)<br />
3.16227e-10<br />
Carrier Sense Threshold (watts)<br />
7.91359e-11<br />
Table II. MAC layer Configuration in NS2<br />
13
AIZAZ U. CHAUDHRY, ROSHDY H.M. HAFEZ AND OSAMA ABOUL-MAGD<br />
MAC Layer Parameters<br />
Settings<br />
Minimum Contention Window 15<br />
Maximum Contention Window 1023<br />
Slot Time (micro seconds) 9<br />
SIFS period (micro seconds) 16<br />
Preamble Length (bits) 96<br />
PLCP Header Length (bits) 24<br />
Basic Rate (Mbps) 6<br />
Data Rate (Mbps) 54<br />
80<br />
70<br />
TICA<br />
CCA<br />
SRSC<br />
60<br />
Average Throughput (Mbps)<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
0 Random Controlled Random Grid<br />
TOPOLOGY<br />
Figure 2. Average Throughput <strong>of</strong> all schemes for the three topologies<br />
80<br />
70<br />
TICA<br />
CCA<br />
SRSC<br />
60<br />
Average Throughput (Mbps)<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
1 2 3 4 5 6 7 8 9 10<br />
Topology Number<br />
Figure 3. Average Throughput for ten random topologies<br />
80<br />
70<br />
TICA<br />
CCA<br />
SRSC<br />
60<br />
Average Throughput (Mbps)<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
1 2 3 4 5 6 7 8 9 10<br />
Topology Number<br />
Figure 4. Average Throughput for ten controlled random topologies<br />
I. Simulation Results<br />
Figure 2 shows a graphical comparison <strong>of</strong> average throughput <strong>of</strong> all schemes for the three topologies.<br />
As compared to CCA scheme, throughput improvement with TICA is 76% for random topology, 1244%<br />
for controlled random topology and 1341% for grid topology. In comparison to SRSC scheme, throughput<br />
improvement with TICA is 319% for random topology, 10863% for controlled random topology and<br />
13941% for grid topology.<br />
14
POWER CONTROL-BASED CHANNEL ASSIGNMENT ALGORITHM FOR WIRELESS MESH NETWORKS<br />
Figures 3 and 4 show average throughput <strong>of</strong> all schemes for ten different random and controlled random<br />
topologies, respectively. These results clearly indicate that our proposed algorithm, TICA, significantly<br />
outperforms other schemes for all topologies.<br />
8. CONCLUSION<br />
This paper addresses the trade-<strong>of</strong>f between connectivity and throughput <strong>of</strong> wireless mesh networks. If<br />
nodes are allowed to transmit at maximum power, the connectivity will be high but the excessive<br />
interference will reduce the overall network throughput. This paper proposes and evaluates algorithms that<br />
allow the designer to choose the appropriate balance between connectivity and throughput. The algorithm<br />
assumes that each node has multiple independent radios, and calls for careful calculation <strong>of</strong> transmitted<br />
power to achieve the required connectivity with minimum interference. The new algorithm is termed TICA<br />
(Topology-controlled Interference-aware Channel Assignment Algorithm). The algorithm performance<br />
was tested on three types <strong>of</strong> network topologies, Random topology, Controlled Random and Regularly<br />
spaced nodes. The improvement in the overall network throughput due to the proposed TICA is<br />
significantly higher than the conventional CCA scheme and is much higher than the SRSC scheme in all<br />
three topologies. This is due to the fact that power control results in an efficient frequency reuse which<br />
leads to improvement in the network throughput.<br />
9. REFERENCES<br />
[1.] N. Nandiraju et al., “Wireless Mesh Networks: Current Challenges and Future Directions <strong>of</strong> Web-In-<br />
The-Sky,” IEEE Wireless Communications 14(4), pp. 79-89, August 2007.<br />
[2.] H. Skalli, S. Ghosh, S. K. Das, L. Lenzini, and M. Conti, “Channel assignment strategies for multiradio<br />
wireless mesh networks: Issues and solutions,” IEEE Comm. Magazine, 45(11), pp. 86-93,<br />
November 2007.<br />
[3.] N. Li, J. Hou, and L. Sha, “Design and analysis <strong>of</strong> an MST-based topology control algorithm,” IEEE<br />
Transactions on Wireless Communications , 4(3), pp. 1195- 1206, May 2005<br />
[4.] F. O. Aron, T. O. Olwal, A. Kurien and Y. Hamam, “Network preservation through a topology<br />
control algorithm for wireless mesh networks,” in Proc. 2nd IASTED Africa Conference on Modelling<br />
and Simulation, September 2008.<br />
[5.] A. Raniwala, K. Gopalan, and T. Chiueh, “Centralized channel assignment and routing algorithms for<br />
multi-channel wireless mesh networks,” ACM MC2R 8(2), pp. 50-65, April 2004.<br />
[6.] P. Kyasanur and N. Vaidya., “Routing and Interface Assignment in Multi-Channel Multi-Interface<br />
Wireless Networks,” in Proc. IEEE WCNC, 4, pp. 2051-2056, March 2005.<br />
[7.] K. Ramachandran, E. Belding, K. Almeroth, and M. Buddhikot, “Interference-Aware Channel<br />
Assignment in Multi-Radio Wireless Mesh Networks,” in Proc. IEEE INFOCOM, April 2006.<br />
[8.] T. Rappaport, “Wireless Communications: Principles and Practice,” 2nd Ed. Prentice Hall, Upper<br />
Saddle River, NJ, 2002.<br />
[9.] K. Rosen, “Discrete Mathematics and its Applications,” McGraw Hill, 1999.<br />
[10.] Ad Hoc On-Demand Distance Vector (AODV) Routing Ptotocol, RFC3561,<br />
http://www.ietf.org/rfc/rfc3561.txt<br />
[11.] The VINT Project, “Network Simulator – NS2,” http://www.isi.edu/nsnam/ns/<br />
[12.] R. A. Calvo and J. P. Campo, “Adding Multiple Interface Support in NS-2,” University <strong>of</strong> Cantabria,<br />
January 2007.<br />
[13.] "IEEE Standard for IT-Telecom. and Info. Exchange Between Systems-Local and Metropolitan Area<br />
Networks-Specific Requirements - Part 11: Wireless LAN MAC and PHY Specifications," IEEE Std<br />
802.11-2007 (Revision <strong>of</strong> IEEE Std 802.11-1999) , June 2007<br />
15
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 17-30 (2010)<br />
© 2009 ALHOSN University<br />
CHARACTERIZATION OF ELASTIC BUCKLING OF<br />
CYLINDRICAL LINERS USING FINITE ELEMENT<br />
APPROACH<br />
A. M. Sweedan 1* K. M. El-Sawy 1<br />
1 Civil and Environmental Eng. Faculty <strong>of</strong> Eng., UAE University, P.O.Box: 17755, Al Ain, U.A.E.<br />
ABSTRACT: The results <strong>of</strong> a numerical study to evaluate the elastic buckling pressure, thrusts and bending<br />
moments <strong>of</strong> slender liners are reported. The adopted finite element model accounts for the non-linearity effects <strong>of</strong><br />
large deformations and geometric imperfections on the stability <strong>of</strong> analyzed cylindrical liners. Results <strong>of</strong> finite<br />
element analysis are used to identify the critical buckling pressure and the associated maximum thrust and moment.<br />
In addition, several analytical and numerical solutions for the same problem are critically reviewed; namely Lo and<br />
Zhang, Thépot, and Madryas and Szot solutions. An extensive finite element analyses are conducted to evaluate the<br />
elastic buckling pressure and the associated thrust and bending moment at buckling <strong>of</strong> cylindrical liners with various<br />
dimensions. Obtained numerical results are then compared to relevant predictions provided by other solutions to<br />
investigate the influence <strong>of</strong> approximations adopted by such solutions on the accuracy <strong>of</strong> the predicted values.<br />
Comparison reveals the adequacy <strong>of</strong> some <strong>of</strong> these methods while others are shown to be on the non-conservative<br />
side and, therefore, may be used with caution for design/check purposes.<br />
Key words: Elastic Buckling, Liners, Finite Element Method, Non-Linear Analysis.<br />
1. INTRODUCTION<br />
Despite the fact that most <strong>of</strong> the damaged pipelines are partially deteriorated, they require<br />
renovation to prevent potential leakage. One solution is to apply an internal flexible polymeric<br />
liner to the pipeline. These liners are <strong>of</strong>ten installed in pipelines that lie below the water table,<br />
and as such they are subjected to external hydrostatic pressure which is transmitted through the<br />
leaking cracked pipeline. In this case, the liner carries only the pressures from the contained fluid<br />
and the external groundwater pressure. In addition, due to some installation considerations, the<br />
liner is temporarily subjected to an additional pressure when grout is pumped around the liner to<br />
fill any annular gaps between the liner and its host pipeline.It is experimentally evident that the<br />
encased liner subject to external pressure may buckle and cannot withstand any additional<br />
pressure at a critical pressure, which is well above the one for the unsupported (free) liner<br />
[1,2,3,4 and 5]. For slender liners commonly used in practice it is assumed, and confirmed<br />
experimentally, that geometrical instability (i.e. elastic buckling) occurs first before any material<br />
failure. The study presented herein is restricted to the case <strong>of</strong> elastic slender liners only. The<br />
typical fundamental buckling mode shape is shown in Fig. 1(b) along with the initial liner’s<br />
geometry depicted by Fig. 1(a). Real conditions in which liners are installed (i.e. non-uniform<br />
groundwater pressure and any possible debris at the bottom <strong>of</strong> the host pipe) hint that the liner<br />
most probably will buckle at the invert in a one-lobe buckling mode (Fig. 1(b)). In addition,<br />
theoretical potential energy balance <strong>of</strong> the liner system indicates that the one-lobe mode<br />
corresponds to the minimum critical pressure.<br />
_______________________________________<br />
* Corresponding Author. Tel.: +971-3-762 3154<br />
E-mail : amr.sweedan@uaeu.ac.ae<br />
17
A.M. SWEEDAN, K.M. EL-SAWY<br />
Pressure<br />
Load<br />
Pressure<br />
Load<br />
Soil<br />
R o<br />
R<br />
Liner<br />
Soil<br />
Single Lobe<br />
Liner<br />
gap = d<br />
Rigid Host Pipe<br />
(a) Initial Liner Geometry<br />
Rigid Host Pipe<br />
(b) Single-Lobe Buckling Mode<br />
Fig. 1. Initial geometry and fundamental buckling mode <strong>of</strong> cylindrical liners.<br />
In the current study, finite element method is utilized to investigate elastic buckling <strong>of</strong> slender<br />
liners. The induced critical thrust and bending moment at buckling are also evaluated<br />
numerically.<br />
2. PREDICTION OF CRITICAL BUCKLING LOAD<br />
The buckling <strong>of</strong> an unsupported circular ring subjected to an external pressure was first studied<br />
by Levy [6]. The buckling pressure, P , for a ring <strong>of</strong> radius, R<br />
cr<br />
o , cross sectional moment <strong>of</strong> inertia,<br />
I, and modulus <strong>of</strong> elasticity, E, is defined by<br />
3 EI<br />
(1)<br />
P<br />
cr<br />
=<br />
R<br />
The elastic solution for a pipe with wall thickness, t, is adopted from the ring case by simply<br />
2<br />
adjusting the elastic Young’s modulus E, to be E ' E /(1 ) (where is the Poisson’s ratio for<br />
the liner material) and the moment <strong>of</strong> inertia I, to be t 3 /12. Thus, P cr<br />
for the pipe case can be<br />
defined as<br />
3<br />
o<br />
3 EI<br />
Pcr<br />
=<br />
2<br />
(1 -<br />
) R<br />
3<br />
o<br />
E'<br />
<br />
<br />
4<br />
<br />
<br />
The encased liner stability problem appeared in the literature in many different forms according<br />
to the <strong>engineering</strong> application being examined. The most common stability problem types are<br />
encased liner subjected to thermal expansion, or under external pressure loading. The two types<br />
differ significantly in their boundary conditions. For the thermal expansion problem, the detached<br />
part <strong>of</strong> the liner due to buckling has no longer any contact pressure acting on it. On the other<br />
hand, for the external pressure case, the detached portion <strong>of</strong> the encased liner is still subjected to<br />
the <strong>applied</strong> pressure. It is therefore important to distinguish between these two types in the<br />
literature review <strong>of</strong> the stability problem.<br />
Due to the installation techniques <strong>of</strong> liners and the nature <strong>of</strong> the lined element, some<br />
imperfections may develop. These imperfections lead to deviation <strong>of</strong> the encased liner problem<br />
t<br />
R<br />
o<br />
<br />
<br />
<br />
3<br />
(2)<br />
18
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
from the ideal case <strong>of</strong> tightly fitted cylindrical liner. Three major imperfection types are<br />
considered in literature and are listed as follows:<br />
gap between the liner and the host element,<br />
ovality in the host element which is consequently reflected to the tightly fitted liner, and<br />
longitudinal wrinkle (or wavy imperfection) in the liner itself.<br />
These imperfections are responsible for the reduction <strong>of</strong> the liner stability and should be<br />
accounted for in the assessment <strong>of</strong> the liner critical pressure. The work presented herein deals<br />
with the first imperfection type (i.e. gap imperfection or the case <strong>of</strong> liner).<br />
As a result <strong>of</strong> this imperfection and as the external pressure increases, the initially loose (or<br />
unsupported) liner deforms until it contacts its surrounding rigid host. Then the behavior <strong>of</strong> the<br />
liner changes abruptly from the case <strong>of</strong> an unsupported loose liner to the case <strong>of</strong> the constrained<br />
one. The constrained liner continues to deform until it becomes geometrically unstable (i.e.<br />
buckles) or material failure occurs. As mentioned earlier, only elastic buckling is considered<br />
herein and therefore this limits the results <strong>of</strong> this study to the case <strong>of</strong> slender liners where no<br />
material failure occurs. This would raise a question about the validity <strong>of</strong> the thin liner<br />
assumption, which will be answered later at the end <strong>of</strong> this study.<br />
Researchers and code developers have employed the concept <strong>of</strong> comparing the buckling<br />
pressures for unsupported and encased circular liners. For the case <strong>of</strong> circular liner the idea has<br />
been to determine an enhancement factor which depends on the liner material, geometry and the<br />
size <strong>of</strong> the gap between the liner and its rigid host. Now the critical pressure for circular liner,<br />
P cr , can be defined as<br />
P<br />
cr<br />
= K P cr<br />
(3)<br />
where P and<br />
cr<br />
P are the buckling pressures for encased and unsupported liners respectively, and<br />
cr<br />
K is the enhancement factor, which will be discussed later in section 4.<br />
Different solutions have been developed for estimating the critical pressure <strong>of</strong> the slender liner.<br />
None <strong>of</strong> these methods have provided a validation <strong>of</strong> the thin liner assumption they are based<br />
upon. All such methods assumed smooth contact between the liner and the rigid host element (the<br />
most conservative case). They differed mainly in their basic assumptions for strain (or<br />
deformations) and geometry <strong>of</strong> the detached portion <strong>of</strong> the liner. Most <strong>of</strong> the available solutions<br />
have adopted simplifying assumptions that may or may not provide a good approximation for the<br />
liner buckling pressure.<br />
Equations for modelling single-lobe buckling <strong>of</strong> a cylindrical encased liner have been derived<br />
by Jacobsen [7] in which the single-lobe buckled portion <strong>of</strong> the liner is approximated by a cosine<br />
function representing three half-waves. The geometry <strong>of</strong> the buckled shape is shown in Fig. 2(a).<br />
Basic assumptions used by Jacobsen are listed as follows:<br />
the detached part <strong>of</strong> the liner follows a cosine function (three half-waves) super-imposed on a<br />
circle <strong>of</strong> radius ,<br />
the thrust in the detached portion <strong>of</strong> the liner is constant and equals P, and<br />
the thrust in the other portion <strong>of</strong> the liner (i.e. portion sliding on the walls <strong>of</strong> the rigid host) is<br />
constant and equals PR.<br />
The equations developed by Jacobsen’s are given as follows:<br />
19
A.M. SWEEDAN, K.M. EL-SAWY<br />
Attached Segment <strong>of</strong><br />
the Liner<br />
Pressure<br />
Load<br />
<br />
<br />
<br />
R<br />
Detached Segment<br />
<strong>of</strong> the Liner<br />
Rigid Circular<br />
Cavity<br />
(a) Jacobsen (1974)<br />
Attached Segment<br />
<strong>of</strong> the Liner<br />
Pressure<br />
Load<br />
<br />
R<br />
<br />
<br />
Detached Segment<br />
<strong>of</strong> the Liner<br />
(b) Lo and Zhang (1994)<br />
Rigid Circular<br />
Cavity<br />
Pressure<br />
Load<br />
Attached Segment <strong>of</strong><br />
the Liner<br />
Rigid Circular<br />
Cavity<br />
Deformation w 1<br />
Detached Segment<br />
<strong>of</strong> the Liner<br />
R <br />
Point O<br />
(c) Thépot (2000)<br />
,<br />
n w<br />
,<br />
s u<br />
Initial<br />
gap, d<br />
Fig. 2. Geometry <strong>of</strong> buckled liner after Jacobsen (1974), Lo and Zhang (1994) and Thépot (2000)<br />
2<br />
9<br />
<br />
2<br />
sin( )<br />
<br />
1<br />
<br />
<br />
<br />
2<br />
R 4<br />
<br />
sin( )<br />
<br />
(4)<br />
<br />
3<br />
t<br />
2<br />
sin( )<br />
d<br />
sin( )<br />
tan <br />
<br />
<br />
12<br />
<br />
<br />
1<br />
<br />
sin( )<br />
2R<br />
sin( )<br />
<br />
4<br />
<br />
<br />
20
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
2<br />
9<br />
<br />
<br />
<br />
3<br />
E ' R <br />
12 <br />
t <br />
<br />
1<br />
<br />
<br />
2<br />
P 4<br />
(5)<br />
sin( ) <br />
<br />
sin( )<br />
<br />
<br />
y t sin( )<br />
<br />
<br />
<br />
<br />
PRsin(<br />
)<br />
4R<br />
sin( )<br />
tan( )<br />
1 <br />
1<br />
<br />
E'<br />
2R<br />
<br />
sin( )<br />
E'<br />
t sin( )<br />
t<br />
sin( )<br />
<br />
(6)<br />
where R is radius <strong>of</strong> the rigid host pipe, R o is initial radius <strong>of</strong> the liner, t is thickness <strong>of</strong> the liner, d<br />
is the gap between the liner and the rigid host walls, P is the external liner pressure, E’ is<br />
2<br />
E /(1 ) where E is the Young’s modulus <strong>of</strong> the liner material, is the Poisson’s ratio <strong>of</strong> the<br />
liner material, y is the yield stress <strong>of</strong> the liner material, and and are geometrical quantities<br />
and are shown in Fig. 2(a).<br />
3<br />
The major criticism for Jacobsen solution is that he mixed the solution <strong>of</strong> the elastic buckling<br />
problem represented, by (4) and (5), with the condition for onset <strong>of</strong> yielding at the outermost<br />
fibres <strong>of</strong> the most heavily stressed cross-section represented by (6). Thus the solution provided a<br />
value for the pressure that corresponds to first material yielding and, <strong>of</strong> course, this happens at a<br />
value for the pressure different from the elastic buckling pressure. It is also clear from Jacobsen<br />
solution, that there is no compatibility between thrusts and bending moments in the attached and<br />
detached portions <strong>of</strong> the liner at their junction point (i.e. at the ends <strong>of</strong> the detached portion <strong>of</strong> the<br />
liner). It may also be noted that the two basic assumptions for the thrust distribution along the<br />
liner and deformed shape at buckling should be assessed.<br />
Lo and Zhang [3] studied the effect <strong>of</strong> the liner’s lack <strong>of</strong> fit on the critical pressure for curedin-place<br />
polymer liners. Figure 2(b) shows the initial configuration <strong>of</strong> the problem. The buckling<br />
pressure was derived using simple linear theory. Their basic assumptions are listed as follows.<br />
the detached part <strong>of</strong> the liner is circular with radius ,<br />
buckling <strong>of</strong> the detached portion <strong>of</strong> the liner is analogous to buckling <strong>of</strong> fixed-ended circular<br />
arch,<br />
thrust in the whole liner is constant and equals PR o where R o is the initial radius <strong>of</strong> the liner,<br />
and<br />
shortening in liner is due to thrust only, and first order terms are considered in calculating<br />
deformations.<br />
Their equations were based on satisfying the geometrical boundary conditions as follows:<br />
perimeter <strong>of</strong> deformed liner equals the original undeformed perimeter less the membrane<br />
deformations which can be expressed as a function <strong>of</strong> the liner pressure, and<br />
liner buckling pressure is analogous to the buckling pressure <strong>of</strong> a fixed-ended circular arch <strong>of</strong><br />
radius (radius <strong>of</strong> the detached part <strong>of</strong> the liner).<br />
Lo and Zhang’s solution has been reformulated earlier by El-Sawy and Moore [8] to permit<br />
direct comparison with other solutions. Accordingly, the solution for the liner critical pressure,<br />
P cr , can be represented by the following equations.<br />
21
A.M. SWEEDAN, K.M. EL-SAWY<br />
<br />
P = E'<br />
<br />
<br />
t<br />
R<br />
3<br />
<br />
<br />
<br />
<br />
<br />
<br />
sin( ) t<br />
<br />
sin( )<br />
2R<br />
-3<br />
<br />
<br />
<br />
(7)<br />
2<br />
k 1<br />
12<br />
2<br />
2<br />
-3<br />
2<br />
d<br />
sin( ) t d sin( ) t 1<br />
1 <br />
k<br />
<br />
sin( )<br />
2R<br />
R 2R<br />
sin( )<br />
2R<br />
12<br />
k tan( ) tan( k<br />
)<br />
(9)<br />
(8)<br />
where all symbols are as defined previously and in Fig. 2(b). The solution strategy starts with the<br />
assumption <strong>of</strong> a value for in (8) followed by the calculation <strong>of</strong> the corresponding and k<br />
values from (8) and (9). The pressure, P, can then be calculated from (7). The critical pressure,<br />
P cr , corresponds to the value <strong>of</strong> which maximizes the value <strong>of</strong> the pressure, P.<br />
The major criticism for Lo and Zhang’s solution is due to their inappropriate analogy between<br />
buckling <strong>of</strong> the detached portion <strong>of</strong> the liner and the buckling <strong>of</strong> the fixed-ended circular arch.<br />
The boundary conditions <strong>of</strong> the fixed-ended circular arch are inappropriate as they are associated<br />
with unknown fixed end moments at the arch ends. In the case <strong>of</strong> the liner problem, these end<br />
moments and bending moments along the attached part <strong>of</strong> the liner have a constant value defined<br />
by the constant change in curvature in the liner after deformation, where the constant change <strong>of</strong><br />
curvature equals [( 1/ R o<br />
) (1/ R)]<br />
. This moment distribution in the liner is not consistent with an<br />
arch model featuring fixed ends. Therefore, Lo and Zhang solution will be considered only in<br />
calculating the critical pressure <strong>of</strong> the liner and will not be used for any thrust or bending moment<br />
calculations.<br />
Recently, Thépot [9] has developed an analytical solution for non-circular sewer liners which,<br />
as reported by Thépot, is a general solution that can be adopted for the case <strong>of</strong> circular and noncircular<br />
liners with single-lobe and double-lobe buckling modes as well. The solution calculates<br />
the maximum pressure, P cr (i.e., buckling pressure), and thrust, N cr , and maximum bending<br />
moment, M cr , at the mid-point <strong>of</strong> the detached segment <strong>of</strong> the liner (i.e., =0) just before buckling<br />
(Fig. 2(c)). Furthermore, Thépot provided an approximate solution which can be written as:<br />
Pcr<br />
<br />
E '<br />
0.455 k<br />
2<br />
5<br />
11<br />
5<br />
t<br />
Ro<br />
N<br />
cr<br />
Pcr<br />
1.26 ,<br />
'<br />
' <br />
<br />
<br />
E Ro<br />
E <br />
3<br />
cr<br />
t <br />
<br />
2<br />
'<br />
<br />
<br />
E Ro<br />
Ro<br />
<br />
2<br />
9<br />
1.0<br />
0.41<br />
0.006<br />
2 5 5<br />
and g Ro<br />
11.65<br />
10. 0 (10)<br />
2 3 6<br />
5 5 5<br />
k t<br />
2<br />
0.1<br />
0.053<br />
0.0012<br />
<br />
2<br />
1.0<br />
0.17<br />
0.007<br />
2<br />
M (11)<br />
where g is the annular gap between the liner and the host pipe (i.e., g=d/2 for the single-lobe<br />
case), k is the number <strong>of</strong> buckled lobes (i.e., k=1 for one-lobe buckling model), and is the<br />
perimeter <strong>of</strong> the liner (i.e., =2R o for the circular liner). Thépot recommended an approximate<br />
solution based on data fitting <strong>of</strong> his analytical solution for the cases <strong>of</strong> liner gaps defined by<br />
10. Values for P cr and (Eq. 10) can be rewritten for the case <strong>of</strong> circular liner with a singlelobe<br />
as:<br />
Pcr<br />
<br />
E '<br />
0.218<br />
2<br />
1.0 0.41<br />
0.006<br />
<br />
<br />
<br />
<br />
<br />
t<br />
R<br />
o<br />
<br />
<br />
2.2<br />
and d Ro<br />
<br />
1.465<br />
<br />
10. 0 (12)<br />
Ro<br />
t <br />
4<br />
5<br />
1.2<br />
22
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
More recently, Madryas and Szot [10] have preformed Finite Element analyses and provided a<br />
formula for the critical pressure, P cr , <strong>of</strong> liner which is based on regression analysis for their<br />
numerical results. It is worth mentioning that their formula was part <strong>of</strong> finding another formula<br />
for the combined effect <strong>of</strong> three liner imperfection types, including the gap, ovality, and wavy<br />
imperfections. Madryas and Szot reported an approximate solution for the critical pressure, P cr ,<br />
<strong>of</strong> the liner as:<br />
<br />
<br />
2.2<br />
P cr <br />
1<br />
t<br />
(13)<br />
<br />
E '<br />
<br />
<br />
1 <br />
<br />
d <br />
0.84176 <br />
2R<br />
<br />
0.89415<br />
<br />
<br />
<br />
t<br />
R<br />
<br />
<br />
<br />
1.16189<br />
<br />
<br />
<br />
<br />
<br />
<br />
<br />
2R<br />
<br />
It is important to mention that the above formula satisfies the case <strong>of</strong> no gap and provides the<br />
well established critical pressure <strong>of</strong> the tightly fitted circular liner (Glock solution [11]) with onelobe<br />
buckling as<br />
4. FINITE ELEMENT MODELING AND ANALYSIS<br />
2.2<br />
P Glock t <br />
<br />
(14)<br />
E ' 2R<br />
<br />
The Finite Element Method is employed to solve the liner buckling problem. A strip <strong>of</strong> unit<br />
length <strong>of</strong> the liner is only considered because <strong>of</strong> the long cylindrical geometry <strong>of</strong> the liner (i.e.<br />
plane strain conditions prevail). Node-to-surface contact elements are used to simulate the<br />
contact problem while 3-noded shell elements are used to model the liner. The liner material is<br />
considered linear and elastic and the liner geometry is approximated by an ellipse with a ratio <strong>of</strong><br />
Normalized Thrust N/(E'R)*10 5<br />
for d/R=0.00 and t/R=0.01<br />
2.0<br />
Normalized Thrust N/(E'R)*10 6<br />
for d/R=0.05 and t/R=0.01<br />
2.0<br />
Normalized Thrust N/(E'R)*10 6<br />
for d/R=0.10 and t/R=0.01<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
Normalized Thrust N/(E'R)*10 3<br />
for d/R=0.00 and t/R=0.10<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
Normalized Thrust N/(E'R)*10 3<br />
for d/R=0.05 and t/R=0.10<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
Normalized Thrust N/(E'R)*10 3<br />
for d/R=0.10 and t/R=0.10<br />
2.0<br />
1.0<br />
0.0<br />
-1.0<br />
-2.0<br />
-1.0<br />
0.0<br />
1.0<br />
2.0<br />
(a) Initial Liner Geometry<br />
(b) Single-Lobe Buckling Mode<br />
Fig. 3. Distribution <strong>of</strong> Thrust in liner with d/R=0.0, 0.05, and 0.1<br />
23
A.M. SWEEDAN, K.M. EL-SAWY<br />
1.0005 between the ellipse axes. This initial elliptical imperfection is used to trigger the onelobe<br />
buckling mode. The Finite Element analysis used accounts for non-linearity effects due to<br />
large change <strong>of</strong> geometry and non-conservative follower pressure load.<br />
A parametric study is conducted to investigate the variation <strong>of</strong> the buckling load, P cr , with the<br />
liner’s geometry (t/R) and (d/R) and material properties E’. The adopted finite element model is a<br />
revised and extended version <strong>of</strong> the one that was used earlier by El-Sawy and Moore [8]. The<br />
extended model predicts the critical buckling pressure and the corresponding thrust, N o , and<br />
bending moment, M o , at the mid-point <strong>of</strong> the detached liner (i.e., the most heavily stressed point<br />
in the liner). This information is essential for conducting more comprehensive <strong>engineering</strong> design<br />
<strong>of</strong> such special structure. It is important to mention that the FE solution is the only consistent<br />
solution among all the previously reviewed ones since it does not assume any approximation<br />
(other than the assumption <strong>of</strong> elastic and smooth thin liner) in modelling the liner behaviour.<br />
5. DISCUSSION OF RESULTS<br />
Figure 3 shows the distribution <strong>of</strong> the normalized thrust, N/E’R, based on the FEM solution for<br />
different (d/R) and (t/R) ratios. In this figure, it can be noticed that the thrust is almost constant<br />
for all values <strong>of</strong> (d/R) and (t/R) except for a very localized rise in the thrust magnitude at the<br />
points where the liner starts to contact, and gets supported on, the rigid host. This increase in the<br />
thrust is noticeable for large values <strong>of</strong> (d/R) where the liner is freer to deform. The maximum rise<br />
in the thrust magnitude is about 23% and occurs for (d/R)=0.1 and (t/R)=0.1. This would show<br />
that the constant thrust along the whole liner is an approximate one, which in general may not<br />
affect the estimation <strong>of</strong> the critical pressure, but may affect the estimations <strong>of</strong> the accompanying<br />
thrusts and bending moments.<br />
Normalized Buckling Pressure (P cr / E') x 10 4<br />
20.0<br />
15.0<br />
10.0<br />
5.0<br />
Tightly Fitted Liner (d / R) = 0.0<br />
FEM<br />
Thepot<br />
Lo and Zhang<br />
Madryas and Szot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
24
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
Normalized Buckling Pressure (P cr / E') x 10 4<br />
15.0<br />
10.0<br />
5.0<br />
Loosely Fitted Liner (d / R) = 0.05<br />
FEM<br />
Thepot<br />
Lo and Zhang<br />
Madryas and Szot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
Normalized Buckling Pressure (P cr / E') x 10 4<br />
15.0<br />
10.0<br />
5.0<br />
Loosely Fitted Liner (d / R) = 0.10<br />
FEM<br />
Thepot<br />
Madryas and Szot<br />
Lo and Zhang<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
Fig. 4. Critical pressure for liner with d/R=0.0, 0.05 and 0.10<br />
Figure 4 shows a comparison between the normalized critical pressure, P cr /E’, based on the<br />
different solutions reviewed earlier (i.e., solutions provided by FEM, Lo and Zhang [3], Thépot<br />
[9] and Madryas and Szot [10]) versus the liner thickness-to-radius ratio (t/R) at different radial<br />
gap ratios (d/R=0.0, 0.05, and 0.10). It is clear from the figure that the buckling pressure<br />
decreases as the gap size increases. This decrease is more pronounced for thin liners where the<br />
flexural stiffness-to-axial stiffness ratio is very small. For this case, flexural deformations<br />
dominate the behavior <strong>of</strong> the liner and the liner depends mainly on the support <strong>of</strong> its rigid host<br />
element. For thick liners, the axial deformations dominate the behavior and the liner elastic<br />
stability depends mainly on the liner axial stiffness. The comparison shows also the following:<br />
The solution <strong>of</strong> Lo and Zhang provides the highest estimate for the buckling pressure<br />
(i.e., non-conservative pressure), which may be due to the inconsistent assumption that the<br />
detached segment <strong>of</strong> the liner behaves similarly to a fixed-ended circular arch. The<br />
maximum difference between Lo and Zhang solution and the results <strong>of</strong> the FEM reaches<br />
about 35% for thin liners.<br />
On the contrary, the solutions <strong>of</strong> Thépot and Madryas and Szot provide the most<br />
conservative estimate for the buckling pressure. However, pressure predictions become<br />
highly over-conservative for liners with bigger gaps (i.e., 10). This is attributed to the<br />
fact that the equations provided by Thépot provide reasonable fitting <strong>of</strong> the finite element<br />
25
A.M. SWEEDAN, K.M. EL-SAWY<br />
results related to gap liners not exceeding a threshold value <strong>of</strong> 10. The difference<br />
between Thépot solution and the obtained finite element results ranges from -24.7% to<br />
+2.7%, where the -24.7% corresponds to (t/R)=0.04 and (d/R)=0.1.<br />
12.0<br />
Tightly Fitted Liner (d / R) = 0.0<br />
Normalized Bending Moment<br />
[M cr / (E' R 2 )] x 10 5<br />
10.0<br />
8.0<br />
6.0<br />
4.0<br />
2.0<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
14.0<br />
Loosely Fitted Liner (d / R) = 0.05<br />
Normalized Bending Moment<br />
[M cr / (E' R 2 )] x 10 5<br />
12.0<br />
10.0<br />
8.0<br />
6.0<br />
4.0<br />
2.0<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
16.0<br />
Loosely Fitted Liner (d / R) = 0.10<br />
Normalized Bending Moment<br />
[M cr / (E' R 2 )] x 10 5<br />
14.0<br />
12.0<br />
10.0<br />
8.0<br />
6.0<br />
4.0<br />
2.0<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
Fig. 6. Bending moment at the point <strong>of</strong> maximum stress in liner with d/R=0.0, 0.05 and 0.1<br />
Figure 5 shows the thrusts associated with the buckling pressure (based on FEM and Thépot<br />
[9]) at the point <strong>of</strong> the maximum stress in the liner. This point corresponds to the mid-point <strong>of</strong><br />
the detached part <strong>of</strong> the liner. One may conclude the following from Figure 5:<br />
26
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
20.0<br />
Tightly Fitted Liner (d / R) = 0.0<br />
Normalized Thrust [N cr / (E' R)] x 10 4<br />
15.0<br />
10.0<br />
5.0<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
15.0<br />
Loosely Fitted Liner (d / R) = 0.05<br />
Normalized Thrust [N cr / (E' R)] x 10 4<br />
12.5<br />
10.0<br />
7.5<br />
5.0<br />
2.5<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
12.5<br />
Loosely Fitted Liner (d / R) = 0.10<br />
Normalized Thrust [N cr / (E' R)] x 10 4<br />
10.0<br />
7.5<br />
5.0<br />
2.5<br />
FEM<br />
Thepot<br />
0.0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
Fig. 5. Thrust at the point <strong>of</strong> maximum stress in liner with d/R=0.0, 0.05 and 0.10.<br />
<br />
<br />
By inspection, it is clear that the ratio No/(PcrR) is greater than unity in all <strong>of</strong> the<br />
considered solutions, where P cr /E’ is shown in Fig. 4 and N o /(E’R) is shown in Fig. 5.<br />
The figure reveals also a very good agreement bet-ween Teapot’s solution and the finite element<br />
results.<br />
The bending moment associated with the buckling pressure (at the point <strong>of</strong> the maximum<br />
stresses) is shown in Figure 6. In general, the figure shows very good agreement between<br />
Thépot’s solution and the finite element results. For relatively thicker liners, Thépot’s solution<br />
provides more conservative estimates for the bending moment at the point <strong>of</strong> maximum stresses.<br />
27
A.M. SWEEDAN, K.M. EL-SAWY<br />
Figure 7 shows the enhancement factor, K, versus the liner (t/R) for different values for (d/R),<br />
where the enhancement factor is defined according to (3) and can be written as<br />
3<br />
Pcr Pcr<br />
Pcr<br />
1 d<br />
R<br />
<br />
K <br />
4 <br />
3<br />
' 4 '<br />
<br />
2<br />
<br />
(15)<br />
Pcr<br />
E t R E t R t R <br />
o<br />
It is clear that the enhancement in the liner stability is maximized at zero or small values <strong>of</strong><br />
(d/R) and small values <strong>of</strong> (t/R) (i.e. for the case <strong>of</strong> tightly installed thin liners). For thick liners<br />
and large gaps (i.e. (d/R) = 0.07 to 0.10), the liner enhancement becomes almost constant.<br />
50<br />
Tightly Fitted Liner (d / R) = 0.0<br />
Enhancement Factor K<br />
40<br />
30<br />
20<br />
10<br />
FEM<br />
Thepot<br />
Lo and Zhang<br />
Madryas and Szot<br />
0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
7<br />
Loosely Fitted Liner (d / R) = 0.05<br />
6<br />
Enhancement Factor K<br />
5<br />
4<br />
3<br />
2<br />
1<br />
FEM<br />
Thepot<br />
Lo and Zhang<br />
Madryas and Szot<br />
0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
5<br />
Loosely Fitted Liner (d / R) = 0.10<br />
4<br />
Enhancement Factor K<br />
3<br />
2<br />
1<br />
FEM<br />
Thepot<br />
Lo and Zhang<br />
Madryas and Szot<br />
0<br />
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
(t / R)<br />
Fig. 7. Enhancement factor for liner with d/R=0.0, 0.05 and 0.10<br />
28
CHARACTERIZATION OF ELASTIC BUCKLING OF CYLINDRICAL LINERS USING FINITE ELEMENT APPROACH<br />
In view <strong>of</strong> Fig. 7, it can be shown that for the cases <strong>of</strong> thin liners with large gaps (i.e., small<br />
(t/R) and (d/R)=0.05 and 0.10):<br />
Solution <strong>of</strong> Lo and Zhang is the most non-conservative solution and may lead to unsafe<br />
estimation <strong>of</strong> K.<br />
On the other hand, the solution <strong>of</strong> Madryas and Szot, although based on Finite Element<br />
Analysis, provides the most conservative K, especially at small (t/R) values,<br />
Thépot’s solution is also in a very close agreement to the finite element analysis results.<br />
5. CONCLUSION<br />
The finite element method is used to study elastic stability <strong>of</strong> liners subjected to uniform<br />
pressure. The adopted model accounts for the non-linearity effects <strong>of</strong> large deformations and<br />
geometric imperfections on the stability <strong>of</strong> analyzed liners. Results <strong>of</strong> finite element analysis are<br />
used to identify the critical buckling pressure and the associated maximum thrust and moment.<br />
A critical review is also conducted to evaluate the available methods for estimating the elastic<br />
critical pressure <strong>of</strong> slender liners. Comparison is held between finite element results and<br />
predictions that are based on such methods from which the following conclusions may be drawn:<br />
Liner’s buckling pressure decreases as the gap size increases. This decrease is more<br />
pronounced for thin liners where the flexural stiffness-to-axial stiffness ratio is very small and<br />
the behaviour is controlled by flexural deformations <strong>of</strong> the liner.<br />
On the contrary, for thick liners, the axial deformations dominate the behavior and the liner<br />
elastic stability depends mainly on its axial stiffness.<br />
The solution proposed by Lo and Zhang provides non-conservative prediction <strong>of</strong> the critical<br />
pressures especially for small values <strong>of</strong> (d/R). On the contrary, the solution <strong>of</strong> Thépot<br />
provides more conservative estimate for the buckling pressure. However, pressure predictions<br />
become highly over-conservative for liners with bigger gaps (i.e., 10).<br />
Thepot’s method is shown to be in very good agreement with obtained finite element results<br />
and provides more conservative estimates for the thrust induced in liners with small gaps (i.e.,<br />
10 ).<br />
For thin liners with large gaps the solution <strong>of</strong> Lo and Zhang is the most non-conservative<br />
solution and may lead to unsafe estimation <strong>of</strong> the enhancement factor K. Meanwhile, the<br />
solution <strong>of</strong> Madryas and Szot, although based on Finite Element Analysis, provides the most<br />
conservative K, especially at small (t/R) values.<br />
6. ACKNOWLEDGMENT<br />
Support <strong>of</strong> this research was provided through research grants No. 03-7-11/99 and 02-7-11/00<br />
from the Research Council <strong>of</strong> the United Arab Emirates University.<br />
29
A.M. SWEEDAN, K.M. EL-SAWY<br />
7. REFERENCES<br />
[1.] Aggarwal, S. C., and Cooper, M. J. (1984). “External pressure testing <strong>of</strong> Insituform lining.”<br />
Internal Rep., Coventry (Lanchester) Polytechnic, Coventry, U.K.<br />
[2.] Kyriakides S., Babcock C. D. (1983). Buckle propagation phenomena in pipelines collapse:<br />
the buckling <strong>of</strong> structures in theory and practice, Proc. IUTAM Symp. on Collapse, London,<br />
Aug. 1982 (edited by J. M. T. Thomson and G. W. Hunt), Cambridge University Press,<br />
75-91.<br />
[3.] Lo, K. H. and Zhang, J. Q. (1994). “Collapse resistance modelling <strong>of</strong> encased pipes.” Buried<br />
Plastic Pipe Technology, vol. 2, ASTM STP1222, Dave Eckstein, Ed., American Society for<br />
Testing and Materials, Philadelphia, Pa., 97-110.<br />
[4.] Guice L. K., Straughan T., Norris C. R., Bennet R. D. (1994). Long-term structural behavior<br />
<strong>of</strong> pipeline rehabilitation systems, Louisiana Technical University, Trenchless Technology<br />
Center, Technical Report #302.<br />
[5.] Bakeer, R. M., Barber, M. E., Pechon, S. E., Taylor, J. E., and Chunduru, S. (1999).<br />
“Buckling <strong>of</strong> HDPE liners under external uniform pressure.” J. Mat. in Civ. Engrg., ASCE,<br />
11(4), 353-361.<br />
[6.] Levy, M. (1884). “Me’memoire sur un nouveau cas inte’grable du proble’m de l’elastique et<br />
l’une de ses applications.” English translation is “Memoir on a new integrable case <strong>of</strong> the<br />
problem <strong>of</strong> elasticity and one <strong>of</strong> its applications.” J. Math. Pure et Appl., Lioville, France,<br />
Series 3, Vol. 10, 5-42.<br />
[7.] Jacobsen, S. (1974). “Buckling <strong>of</strong> circular rings and cylindrical tubes under external<br />
pressure.” Water Power, 26 (Dec.), 400-407.<br />
[8.] El-Sawy, K., Moore, I. D. (1998). “Stability <strong>of</strong> liners used to rehabilitate rigid pipes.” J.<br />
Struct. Engrg., ASCE, 124(11), 1350-1358.<br />
[9.] Thepot, O. (2000). “A new design method for non-circular sewer lining.” Trenchless<br />
Technology Res., Vol. 15, No. 1, 25-41.<br />
[10.] C. Madryas, A. Szot (2003). Structural sensitivity <strong>of</strong> circular sewer liners to geometrical<br />
imperfections, Tunnelling and Underground Space Technology, 18, 421–434.<br />
[11.] Glock, D., 1977. U¨berkritisches verhalten eines starr ummantelten kreisrohres bei<br />
wasserdruck von aussen und temperaturerho¨hung [English translation: Post-critical<br />
behaviour <strong>of</strong> a rigidly encased circular pipe subject to external water pressure and<br />
temperature rise]. Der Stahlbau 46 (7), 212–217.<br />
30
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 31-49 (2010)<br />
© 2009 ALHOSN University<br />
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS<br />
OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
*<br />
Khaled M. El-Sawy, 1 Abdul Hameed Arwani 2<br />
1 Civil & Environmental Engineering Department, Faculty <strong>of</strong> Engineering, United Arab Emirates University,<br />
P.O. Box: 17755, Al Ain, UAE<br />
2 Principal Engineer, Parsons, Abu Dhabi, UAE<br />
ABSTRACT: The stress field around a hole in a laminated composite plate under uni-axial tension exhibits a<br />
very complex pattern. Usually, the Finite Element Method (FEM) is used to calculate such stress field. The<br />
accuracy <strong>of</strong> the Finite Element (FE) analysis depends mainly on the appropriateness <strong>of</strong> the mesh used in the<br />
analysis. Adaptive meshing is one <strong>of</strong> the good options to search for that optimum mesh. Unfortunately, the<br />
adaptive meshing technique involves iterative solutions <strong>of</strong> the problem with gradual refining <strong>of</strong> an initial coarse<br />
mesh until an optimum mesh (that satisfies the requested user tolerances) is reached. These iterative solutions<br />
may add an extra cost to the analysis if the initial mesh is very coarse. In addition to this, some problems still<br />
exist with the error estimation used in the adaptive meshing technique especially when analyzing a problem with<br />
multiple materials (which is the case for laminated composite plates). This initiated a parametric study that tries<br />
to find some guidelines for the choice <strong>of</strong> the optimum mesh without the need to do costly adaptive meshing.<br />
The sub-modeling method has been used to increase the accuracy <strong>of</strong> the solution around the hole (zones <strong>of</strong> high<br />
stress concentrations). Different parameters that control the shape <strong>of</strong> the sub-model mesh have been studied and<br />
recommendations for their optimum values are reported. These parameters are used later in the analysis <strong>of</strong> a<br />
laminated plate with real sequence <strong>of</strong> plies and the problem size and run-time is proved to be affordable with<br />
current PC configurations.<br />
Key words: Finite Element Method, Laminated Composite Plate, Perforated Plate<br />
1. INTRODUCTION<br />
Bolts and rivets are usually necessary components <strong>of</strong> some structural joints. Despite careful<br />
design, practically every structure contains stress concentrations due to bolt or rivet holes. In the<br />
case <strong>of</strong> laminated structures, the presence <strong>of</strong> a hole in the laminate introduces significant stress<br />
contributions in the third dimension, which creates a very complicated three-dimensional (3D)<br />
stress field in the vicinity <strong>of</strong> the hole. This complex state <strong>of</strong> stresses depends on the stacking<br />
sequence <strong>of</strong> the laminate, the fiber orientation <strong>of</strong> each lamina (or ply) as well as the material<br />
properties <strong>of</strong> the fiber and the matrix [1, 5]. Ultimately, these stress concentrations form a primary<br />
source <strong>of</strong> damage initiation especially in the form <strong>of</strong> delamination.<br />
Delamination along the interface <strong>of</strong> two adjacent laminas has long been recognized as one<br />
<strong>of</strong> the most important failure modes in laminated composite structures. The growth <strong>of</strong><br />
delamination may result in a substantial reduction <strong>of</strong> strength and stiffness <strong>of</strong> the laminate. A<br />
primary requirement for the study <strong>of</strong> delamination in composite plates is to identify clearly<br />
the stress field in the regions where delamination is most likely to happen.<br />
Over the past century, many investigators have considered the problem <strong>of</strong> having a circular<br />
hole in a composite plate where delamination happens in the vicinity <strong>of</strong> the hole. Since the<br />
complex stress field around the hole is <strong>of</strong> great interest to correctly recognize delamination,<br />
many researchers have considered several methods <strong>of</strong> determining the stresses in an axially<br />
loaded rectangular plate with a centrally located circular hole [1-9]. Experimental studies<br />
have shown that the interlaminar stresses at a free-edge, such as in the case <strong>of</strong> a hole, can<br />
produce delamination under cyclic loading [1.]. Solution <strong>of</strong> such a problem (e.g. a uni-axially<br />
loaded rectangular plate with centrally located circular hole) could be analytical or numerical.<br />
______________________________________<br />
* Corresponding Author. Tel.: +971-3-7133516<br />
E-mail : k.elsawy@uaeu.ac.ae<br />
31
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
Analytical or closed-form solutions are quite few in number [2.]. Alternately, much<br />
advancement has occurred in using numerical solutions especially the use <strong>of</strong> Finite Element<br />
Method (FEM). The presences <strong>of</strong> complex three-dimensional (3D) stress field in the vicinity<br />
<strong>of</strong> the hole and the complex structural geometry have turned researchers to the use <strong>of</strong> FEM in<br />
the last two decades. Two dimensional (2D) solutions are available in the literature [3.] and<br />
provide some insight into the stress field in the neighborhood <strong>of</strong> the hole edge. Several<br />
authors have recognized the importance <strong>of</strong> 3D analysis in dealing with such problems. Since<br />
3D analysis is not cheap and needs more resources (time and computer hardware), several<br />
researchers have tried to use hybrid methods, for example, the combination <strong>of</strong> the 2D<br />
Classical Laminated Theory (CLT) with 3D FEM [4.] or the combination <strong>of</strong> 2D and 3D FEM<br />
[5.]. Treating the problem in full 3D analysis had encountered difficulties in the past due to<br />
complexity <strong>of</strong> the problem and lack <strong>of</strong> computer memory. For this reason the concept <strong>of</strong><br />
finding a FE strategy for the 3D analysis that optimizes the solution and reduces the huge<br />
amount <strong>of</strong> resources required have been emphasized. Experience tells us that far away from<br />
the hole, stresses are relatively smaller and their rates <strong>of</strong> changing with distance (stress<br />
gradients) are small. However, in the vicinity <strong>of</strong> the hole, the stresses are relatively higher and<br />
their gradients are very high. Generally, the FE mesh should be fine at stress concentration<br />
regions and coarse far from these regions.<br />
Recently, various approaches <strong>of</strong> optimizing 3D FE analysis <strong>of</strong> composite plates having<br />
circular holes have been considered. Adaptive refinement is one <strong>of</strong> these approaches [6.].<br />
Although adaptive meshing technique is a good candidate, unfortunately the error estimation<br />
due to multiple materials still needs further investigation and verification. Problems with the<br />
error estimation happen most likely at material interfaces (i.e. at laminas interfaces), which<br />
are the regions <strong>of</strong> interest. In addition, adaptive meshing is not cheap since it involves<br />
iterative process <strong>of</strong> error estimation, refinement, and mesh regeneration. The adaptive<br />
meshing process usually terminates based on some tolerances, set initially by the analyst, to<br />
ensure that the solution has converged to the required accuracy.<br />
In this study we introduce a FE strategy that optimizes the modeling and analysis <strong>of</strong><br />
laminated composite plates with centrally located circular holes. The strategy is based on<br />
detailed parametric analysis conducted to study the effect <strong>of</strong> different mesh parameters on the<br />
desired results. The final outcome is some recommendations for the optimum mesh<br />
parameters to be used. The conducted parametric analysis is expected to help in putting<br />
guidelines for FE modeling <strong>of</strong> composite plates with holes and in reducing the costly<br />
modeling by reducing solution time. Moreover, when used with adaptive procedure it could<br />
help in reaching convergence faster. The following sections discuss the briefly the different<br />
aspects <strong>of</strong> the FE technique and the detailed parametric study.<br />
2. SUB-MODELING TECHNIQUE<br />
To get accurate stress estimations, it is a common practice to use finer element sizes in<br />
regions <strong>of</strong> the FE model that exhibits high stress concentration and allow for larger element<br />
sizes away from these regions. For the case <strong>of</strong> laminated composite plate, this implies using a<br />
very fine mesh around the hole while allowing larger element sizes away from the hole. The<br />
use <strong>of</strong> a fine mesh around the hole would add more cost to the 3D analysis <strong>of</strong> the laminated<br />
composite plate. A cheaper method <strong>of</strong> achieving the same goal (i.e., accurate results around<br />
the hole) is the use <strong>of</strong> the sub-modeling technique.<br />
The sub-modeling technique provides the solution and stress estimations in two stages. The<br />
32
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
first stage is to use a model with a coarse mesh (termed as global model, Fig. 1a) sufficient to<br />
get accurate stresses away from the hole (i.e. away from the localized zone <strong>of</strong> high stress<br />
concentration). The second stage uses a model with a very fine mesh (termed as sub-model,<br />
Fig. 1b) for the zone around the hole (i.e., zone <strong>of</strong> interest) and then get interpolated values<br />
for the displacements and rotations from the previously solved global model and prescribe<br />
them as displacement and rotation loads in the fine sub-model. Figure 1 shows the concept <strong>of</strong><br />
sub-modeling.<br />
(1a) Global coarse Model with Superimposed High Stress Regions<br />
Sub-models<br />
(1b) Fine sub-models superimposed in the coarse model<br />
Fig 1. Models Used in the Finite Element Sub-modeling Technique<br />
1 Spacing ratio is defined as the ratio <strong>of</strong> the size <strong>of</strong> first element to the size <strong>of</strong> the last element i.e. if the spacing ratio was 14 in certain<br />
direction, then the ratio <strong>of</strong> the first element length in that direction to the last element length in the same direction is 14.<br />
33
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
r<br />
<br />
2a<br />
2W<br />
2L<br />
Fig 2. Problem geometry<br />
3. PROBLEM DEFINITION AND THE FINITE ELEMENT MODEL<br />
Figure 2 shows a laminated composite plate having a central circular hole and being axially<br />
loaded at both ends. The plate is composed <strong>of</strong> four plies with symmetrical stacking sequence<br />
(typically 0/30/30/0, 0/45/45/0, 0/60/60/0 or 0/90/90/0). The material <strong>of</strong> each ply is assumed<br />
to be homogenous, linear elastic, and orthotropic. The direction <strong>of</strong> orthotropy coincides with<br />
the direction <strong>of</strong> the fibers (i.e. the material is stiff in the direction <strong>of</strong> fibers and less stiff in the<br />
other perpendicular directions). Geometry is defined by the plate length L, plate width W,<br />
radius <strong>of</strong> the central hole, a, and the thickness <strong>of</strong> each ply, h (i.e. total plate thickness is 4h).<br />
The mesh parameters are defined as follows:<br />
N z is the number <strong>of</strong> elements through the thickness <strong>of</strong> each ply,<br />
R z is the spacing ratio 1 <strong>of</strong> elements through the thickness <strong>of</strong> each ply,<br />
N is the number <strong>of</strong> elements around the edge <strong>of</strong> the hole and no spacing ratio is used in this<br />
direction,<br />
N r is the number <strong>of</strong> elements extending radially out from the edged <strong>of</strong> the hole edge to the<br />
end <strong>of</strong> the sub-model, and<br />
is the spacing ratio <strong>of</strong> the elements in the radial direction.<br />
R r<br />
The ratio <strong>of</strong> the radius <strong>of</strong> the hole to the ply thickness, a/h, has been taken to be 4, 20, and<br />
100 to examine three schemes <strong>of</strong> composite plates; thick, medium, and thin plates.<br />
L = 12a<br />
Uniform displacement<br />
Ux = 0.0125h<br />
W = 6a<br />
N r , R r<br />
N <br />
N z , R z<br />
Sub-model<br />
2h<br />
Material Properties<br />
Ex=139.0 GPa Gxy=5.52 GPa xy=0.3<br />
a S=6h Ey=10.43 GPa Gxz=5.52 GPa xz=0.3<br />
Ez=10.43 GPa Gyz=3.31 GPa yz=0.55<br />
Fig 3. Finite Element model<br />
34
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
Because <strong>of</strong> the symmetry <strong>of</strong> the problem with respect to x, y, and z directions, only oneeighth<br />
<strong>of</strong> the laminate is analyzed. Figure 3 shows the finite element model, prescribed<br />
displacement boundary conditions, mesh variables, and the orthotropic material properties. It<br />
has been shown by [4.] that the plate dimensions L and W are essentially infinite with respect<br />
to the diameter <strong>of</strong> the hole and that the far-field stresses are reached within a distance <strong>of</strong> 6<br />
times the ply thickness (i.e. 6h) away form the hole edge. Thus a region <strong>of</strong> radial distance <strong>of</strong><br />
at least S=6h (measured from the edge <strong>of</strong> the hole) has been considered as the sub-model<br />
where more refinement has been done to depict the complex stress field.<br />
High order isoparametric 20-noded brick elements were utilized in the analysis. These<br />
elements are more tolerant <strong>of</strong> curved boundaries than standard 8-node brick elements and<br />
avoid “locking” due to high aspect ratios.<br />
Figure 3 shows the mesh parameters. The default values for these parameters are as follows<br />
unless otherwise noted: N z =6, R z =14, N r =10, R r =40, and N =30.<br />
4. VERIFICATION FOR THE SUB-MODEL METHOD<br />
It is essential to check the accuracy <strong>of</strong> the FE analysis using the sub-modeling technique<br />
before using it in performing the parametric analysis. This would ensure that stresses are not<br />
affected by the location <strong>of</strong> the sub-model boundaries.<br />
Two FE models, for the problem geometry shown in Fig. 3 with a/h=20, were used for<br />
comparison purposes (Fig. 4). The first one, MODEL-1, has a very fine mesh over the entire<br />
model while the second, MODEL-2, has used the sub-modeling technique which involves the<br />
use <strong>of</strong> two meshes for the global model and the sub-model. The mesh for the zone <strong>of</strong> interest<br />
in MODEL-1 (i.e. in the neighborhood <strong>of</strong> the hole defined by dimension S in Fig. 3) matches<br />
exactly the mesh <strong>of</strong> the sub-model <strong>of</strong> MODEL-2 and has been based on the parameters N z =6,<br />
R z =14, N r =10, R r =40, and N =30. Figure 4 shows the different meshes used for both models.<br />
(4a) Mesh for MODEL-1<br />
(4b) Mesh for MODEL-2<br />
35
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
(4c) Mesh for sub-model <strong>of</strong> MODEL-2<br />
Fig 4. Finite Element models to test the reliability <strong>of</strong> sub-modeling technique<br />
Since the interlaminar stresses, which are most likely, the source <strong>of</strong> delamination, are very<br />
sensitive to the shape <strong>of</strong> the mesh, a comparison between the stresses resulting from the two<br />
models would indicate how accurate MODEL-2 relative to MODEL-1 is. From comparison,<br />
it is noticed that results from MODEL-2 show excellent agreement with the ones <strong>of</strong> MODEL-<br />
1 with reduction in solution time up to 80%. Figures 5a through 5c show the comparison<br />
between the results for MODEL-1 and MODEL-2. Interlaminar stresses are shown along<br />
paths extend radially form the edge <strong>of</strong> the hole at 0, 45, and 90 degrees where the 0-degree<br />
path coincides with the x-axis and the 90-degree path coincides with the y-axis. All results are<br />
normalized against the far field stress, avg , due to <strong>applied</strong> displacement ( avg = 74.1 MPa).<br />
It is also essential to verify that the chosen boundaries for the sub-model are far enough<br />
from the stress concentrations around the hole (i.e. distance S is sufficient). We can do this by<br />
comparing stresses along the cut boundaries with those along the corresponding locations on<br />
the global model mesh <strong>of</strong> MODEL-2. If the results are in good agreement, this indicates that<br />
proper boundaries have been chosen. Figure 6 shows excellent agreement between x and y<br />
stresses in the sub-model and the corresponding stresses in the coarse model. Stresses are<br />
compared at the material interface between the two plies.<br />
DESCRIPTION OF THE PARAMETRIC ANALYSIS<br />
A parametric study for the problem (shown in Fig. 3) is performed to identify the optimum<br />
values for the parameters defining the FE mesh <strong>of</strong> the sub-model (i.e. N z , R z , N r , R r , and N ).<br />
The problem geometry is defined by the plate length L, width W, radius <strong>of</strong> the central hole, a,<br />
and the ply thickness, h. The default values for the mesh parameters are N z =6, R z =14, N r =10,<br />
R r =40, and N =30 (unless otherwise noted).<br />
To verify that the FE results (for any mesh parameters) have reached the required accuracy,<br />
there should be a measure for that. The rate <strong>of</strong> change <strong>of</strong> the maximum interlaminar stress<br />
(i.e. the maximum <strong>of</strong> rz , zz , or z at the circular boundary <strong>of</strong> the hole) with any mesh<br />
parameter is selected to estimate the accuracy <strong>of</strong> the results <strong>of</strong> any mesh. The maximum<br />
interlaminar stress has be chosen for this purpose since it is, generally, responsible for the<br />
delamination happening in most <strong>of</strong> the laminates composite plates and since other stresses are<br />
small and change abruptly, which consequently leads to fluctuation in the measurements <strong>of</strong><br />
the rate <strong>of</strong> change <strong>of</strong> the stress between any two consecutive runs.<br />
Figure 7 shows different interlaminar stress components along the boundary <strong>of</strong> the hole for<br />
different stacking sequences. The maximum interlaminar stress is found to be the component<br />
36
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
<strong>of</strong> the stress z along the boundary <strong>of</strong> the hole and between =70 o and =80 o , where is<br />
measured from the x-axis counterclockwise. The previous observation is valid for all the<br />
examined stacking sequences and hole sizes.<br />
0.06<br />
zz / avg<br />
0.04<br />
0.02<br />
0.00<br />
0-degree (MODEL-1)<br />
45-degree (MODEL-1)<br />
90-degree (MODEL-1)<br />
0-degree (MODEL-2)<br />
45-degree (MODEL-2)<br />
90-degree (MODEL-2)<br />
-0.02<br />
-0.04<br />
0 1 2 3 4 5<br />
(r-a)/h<br />
(5a) Stress zz (Comparison between MODEL-1 and MODEL-2)<br />
0.4<br />
0.3<br />
45-degree (MODEL-1)<br />
45-degree (MODEL-2)<br />
z / avg<br />
0.2<br />
0.1<br />
0.0<br />
0 1 2 3 4 5<br />
(r-a)/h<br />
(5b) Stress z (Comparison between MODEL-1 and MODEL-2)<br />
0.06<br />
0.04<br />
0.02<br />
rz / avg<br />
0.00<br />
-0.02<br />
-0.04<br />
-0.06<br />
0-degree (MODEL-1)<br />
45-degree (MODEL-1)<br />
90-degree (MODEL-1)<br />
0-degree (MODEL-2)<br />
45-degree (MODEL-2)<br />
90-degree (MODEL-2)<br />
-0.08<br />
0 1 2 3 4 5 6<br />
(r-a)/h<br />
(5c) Stress rz (Comparison between MODEL-1 and MODEL-2)<br />
Fig 5. Comparisons between MODEL-1 and MODEL-2<br />
37
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
2.0<br />
1.5<br />
Normalized Stress<br />
1.0<br />
0.5<br />
x - MODEL-1<br />
y - MODEL-1<br />
x - MODEl-2<br />
y - MODEL-2<br />
0.0<br />
-0.5<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0<br />
(6a) (Distance Along the Cut-Boundary) / a<br />
Fig 6: Comparisons between Stresses in MODEL-1 and MODEL-2 at the Boundaries<br />
/ avg<br />
2.0<br />
1.8<br />
1.6<br />
1.4<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
-0.2<br />
-0.4<br />
z [0/90]<br />
z [0/60]<br />
z [0/45]<br />
z [0/30]<br />
zz<br />
[0/90]<br />
zz<br />
[0/60]<br />
zz<br />
[0/45]<br />
zz<br />
[0/30]<br />
rz<br />
[0/90]<br />
rz<br />
[0/60]<br />
rz<br />
[0/45]<br />
rz<br />
[0/30]<br />
a/h=20, N z<br />
=8, R z<br />
=14, N r<br />
=10, R r<br />
=40, and N =30<br />
-0.6<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 7. Interlaminar stresses along the hole boundary for different stacking sequences<br />
Figure 8 shows z , the maximum stress, for different plate thickness (thick with a/h=4,<br />
medium with a/h=10, and thin with a/h=100) with [0/90/90/0] stacking sequence. It is worth<br />
mentioning that maximum stress appears to be more sensitive to the change in the a/h ratio<br />
than the change in the stacking sequence. Later in this study, we will find that increasing the<br />
number <strong>of</strong> elements through the thickness N z from 8 to 12 for the case <strong>of</strong> a/h=4 will improve<br />
the accuracy <strong>of</strong> the results to reach the same accuracy for the cases <strong>of</strong> a/h=20 and 100.<br />
The parametric analysis is performed by varying one mesh parameter at a time while fixing<br />
all the others. Comparison between the maximum interlaminar stress z <strong>of</strong> any two<br />
consecutive solutions (i.e. consecutive meshes based on the considered mesh parameter)<br />
provided a good measure <strong>of</strong> accuracy <strong>of</strong> the mesh. The FE solution is assumed to converge to<br />
the accurate value whenever the difference in the maximum interlaminar stress (i.e. z at <br />
angle between =70 o and =80 o ) between any two consecutive computer runs becomes less<br />
than 1.0%, and the value <strong>of</strong> the parameter at the last run is taken as the optimum value for this<br />
case. At this stage, further refinement <strong>of</strong> the mesh would enhance the results by less than<br />
1.0%, which is not economical compared to the cost <strong>of</strong> the solution time <strong>of</strong> the refined mesh.<br />
38
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
1.4<br />
1.2<br />
[0/90], N z =8, R z =14, N r =10, R r =40, and N =30<br />
1.0<br />
0.8<br />
z<br />
/ avg<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
-0.2<br />
-0.4<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
a/h=100<br />
a/h=4<br />
a/h=20<br />
Fig 8. Stress z along the hole boundary for different hole sizes<br />
6. RESULTS OF THE FE PAREMETRIC ANALYSIS<br />
In the parametric analysis each mesh parameter was examined alone while fixing other parameters<br />
at their default values until convergence is obtained. The following sections report the effect <strong>of</strong><br />
changing these parameters on the accuracy <strong>of</strong> the solution. All the forthcoming results (reported<br />
in figures 9-21) show the variation <strong>of</strong> the %difference between the interlaminar stress z at the<br />
hole boundary <strong>of</strong> any two successive meshes with the angle (at which the difference in stress is<br />
calculated). The shaded areas shown in Figs. 9-21 correspond to the location <strong>of</strong> the maximum<br />
interlaminar stress z (i.e. at between 70 o and 80 o ).<br />
Symbols shown in the figure legend refer to the varying mesh parameter and its value for<br />
the two successive meshes. For example, in Fig. 9 the varying mesh parameter is N z and the<br />
figure shows two cases; difference in z between meshes with N z =6 and N z =7, and difference<br />
in z between meshes with N z =7 and N z =8. One may notice that the symbols for the<br />
difference between any two successive runs show a discontinuity at =20º-30º (+ to -<br />
discontinuity). This is because values <strong>of</strong> the stress z at these locations are very small and<br />
approach zeros (Fig. 8) and therefore any slight stress difference will represent a huge<br />
%difference.<br />
A. Analysis <strong>of</strong> Parameters N z and R z<br />
Since we are concerned with the interlaminar stresses, N z and R z are expected to be the most<br />
important parameters since they are participating considerably in controlling the element size<br />
at the material interface. The parametric analysis started with initial values <strong>of</strong> N z and R z as 6<br />
and 14, respectively. The analysis has been performed for different stacking sequences and<br />
different plate thickness (i.e. different a/h ratios). Figures 9-15 show the results <strong>of</strong> such<br />
parametric analysis for N z and R z .<br />
It is clear from Figs. 9-12 for the case <strong>of</strong> R z =14, N r =10, R r =40, N =30, S=6h and different<br />
stacking sequences, that N z converges (i.e. produce %difference less than 1%) at number <strong>of</strong><br />
elements N z between 8 and 10. We can notice also that for thick plates (i.e. a/h=4) more<br />
elements are needed through the ply thickness while for thin plates (i.e. a/h=100) fewer<br />
elements are needed. This conclusion is general because it is observed that the interlaminar<br />
stresses increase in the neighborhood <strong>of</strong> the hole (i.e. more stress concentration) as a/h ratio<br />
decreases, which agrees with Choi and Folias 1994. In general we could conclude that for<br />
thick plates (i.e. a/h=4) N z =10 could be enough, for medium ones (i.e. a/h=20) N z =8 is<br />
39
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
enough, and for thin ones (i.e. a/h=100) less number could be reliable (N z =6).<br />
For R z parameter it is noticed that changing this parameter does not affect the results<br />
significantly (Fig. 15). In addition, it does not affect the solution time and cost since the<br />
number <strong>of</strong> DOFs does not change. Therefore, this parameter could take any reasonable value<br />
between 12 and 16.<br />
%Difference between two consecutive runs<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
[0/90] s<br />
& a/h=20<br />
R z<br />
=14, N r<br />
=10, R r<br />
=40<br />
N <br />
=30, S=6h<br />
Nz=7 & Nz=6<br />
Nz=8 & Nz=7<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 9. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
%Difference between two consecutive runs<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
[0/45] s<br />
& a/h=20<br />
R z<br />
=14, N r<br />
=10, R r<br />
=40<br />
N <br />
=30, S=6h<br />
Nz=10 & Nz=9<br />
Nz=9 & Nz=8<br />
Nz=8 & Nz=7<br />
Nz=7 & Nz=6<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 10. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
40
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
%Difference between two consecutive runs<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
[0/30] s & a/h=20<br />
R z =14, N r =10, R r =40<br />
N =30, S=6h<br />
0.5<br />
Nz=10 & Nz=9<br />
Nz=9 & Nz=8<br />
Nz=8 & Nz=7<br />
Nz=7 & Nz=6<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 11. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
%Difference between two consecutive runs<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
[0/60] s & a/h=20<br />
R z =14, N r =10, R r =40<br />
N =30, S=6h<br />
0.5<br />
Nz=10 & Nz=9<br />
Nz=9 & Nz=8<br />
Nz=8 & Nz=7<br />
Nz=7 & Nz=6<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 12. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
Error between two consecutive runs<br />
4.0<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
[0/90] s & a/h=4<br />
R z =14, N r =10, R r =40<br />
N =30, S=6h<br />
Nz=10 & Nz=9<br />
Nz=9 & Nz=8<br />
Nz=8 & Nz=7<br />
Nz=7 & Nz=6<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 13. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
41
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
%Difference between two consecutive runs<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
[0/90] s & a/h=100<br />
R z =14, N r =10, R r =40<br />
N =30, S=6h<br />
Nz=8 & Nz=7<br />
Nz=7 & Nz=6<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 14. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
%Difference between two consecutive runs<br />
1.5<br />
1.0<br />
0.5<br />
[0/90] s & a/h=20<br />
N z =8, N r =10, R r =40<br />
N =30, S=6h<br />
Rz=14 & Rz=12<br />
Rz=16 & Rz=14<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 15. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N z parameter<br />
B. Analysis <strong>of</strong> Parameters N r and R r<br />
Based on the discussion in the previous section, N z is kept at value <strong>of</strong> 8 while R z is kept at 14<br />
for all the forthcoming analyses. The effect <strong>of</strong> N r on the results was examined by assuming<br />
different values for N r between 10 and 13 (Figs. 16-20). It is found that increasing N r would<br />
enhance the results for all the considered stacking sequences and a/h ratios. This is because N r<br />
participates significantly in the element size (in the plane <strong>of</strong> the plate) at the hole<br />
neighborhood, which is a major factor affecting the accuracy <strong>of</strong> the results. In general we can<br />
state that for thin and thick plies (a/h=100 and a/h=4) high number <strong>of</strong> elements and high N r<br />
are required for convergence, while for moderately thick ones (a/h=20) fewer element and<br />
low N r are needed. As an example, for a/h ratio equals 20, N r could be taken as 13 (Fig. 16),<br />
while for a/h=4, N r equals 14 or 15 could be reasonable (Figs. 17 and 18).<br />
The effect <strong>of</strong> the stacking sequence on Nr is shown in Figs. 16,19, and 20. The figures show<br />
that the less the difference between the orientations (e.g. 0/45) <strong>of</strong> the two plies the less the value<br />
<strong>of</strong> N r .<br />
______________________________________<br />
* Corresponding Author. Tel.: +971-3-7133516<br />
E-mail : k.elsawy@uaeu.ac.ae<br />
42
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
%Difference between two consecutive runs<br />
4<br />
3<br />
2<br />
1<br />
[0/90] s & a/h=4<br />
N z =8, R z =14, R r =40<br />
N =30, S=6h<br />
0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Nr=11 & Nr=10<br />
Nr=12 & Nr=11<br />
Nr=13 & Nr=12<br />
Fig 16. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N r parameter<br />
%Difference between two consecutive runs<br />
1.8<br />
1.6<br />
1.4<br />
1.2<br />
1.0<br />
0.8<br />
[0/90] s & a/h=20<br />
N z =8, R z =14, R r =40<br />
N =30, S=6h<br />
0.6<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Nr=11 & Nr=10<br />
Nr=12 & Nr=11<br />
Nr=13 & Nr=12<br />
Fig 17. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N r parameter<br />
%Difference between two consecutive runs<br />
2.0<br />
1.5<br />
1.0<br />
[0/90] s & a/h=100<br />
N z =8, R z =14, R r =40<br />
N =30, S=6h<br />
0.5<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Nr=11 & Nr=10<br />
Nr=12 & Nr=11<br />
Nr=13 & Nr=12<br />
Fig 18. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N r parameter<br />
43
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
%Difference between two consecutive runs<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
[0/45] s & a/h=20<br />
N z =8, R z =14, R r =40<br />
N =30, S=6h<br />
Nr=11 & Nr=10<br />
Nr=12 & Nr=11<br />
0.2<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 20. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N r parameter<br />
%Difference between two consecutive runs<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
[0/60] s & a/h=20<br />
N z =8, R z =14, R r =40<br />
N =30, S=6h<br />
Nr=11 & Nr=10<br />
Nr=12 & Nr=11<br />
Nr=13 & Nr=13<br />
0.2<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
Fig 20. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N r parameter<br />
Based on his discussion, we may conclude that for thick and thin plates N r is advised to be<br />
15, for medium ones N r =13 could enough. As the difference between the plies orientations<br />
increases, one should add 2 to 4 elements to the above numbers.<br />
As stated earlier in the analysis <strong>of</strong> R z , R r does not affect the solution speed and cost, and<br />
then it can take any reasonable value between 40 and 60.<br />
C. Analysis <strong>of</strong> Parameter N <br />
The number <strong>of</strong> elements in the circumferential direction is usually dictated by the geometry <strong>of</strong><br />
the circular hole. This dictated number <strong>of</strong> elements is usually enough to produce accurate<br />
stress results. The initial value <strong>of</strong> N was set to 30 elements. Increasing the value <strong>of</strong> N from<br />
30 elements to 40 affects the results with %difference more than 1%, while increasing this<br />
value from 40 to 50 elements affects the results slightly (Fig. 21) although it increases the<br />
DOFs and solution time considerably. This applies for all stacking sequences and a/h ratios.<br />
Therefore, we may conclude that N =40 is a reasonable choice that would provide accurate<br />
stress results.<br />
44
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
%Difference between two consecutive runs<br />
1.0<br />
0.5<br />
0.0<br />
-0.5<br />
-1.0<br />
-1.5<br />
-2.0<br />
-2.5<br />
[0/90] s & a/h=20<br />
N z =8, R z =14, R r =40<br />
N r =10, S=6h<br />
-3.0<br />
0 10 20 30 40 50 60 70 80 90<br />
degree<br />
N =40 & N =30<br />
N =50 & N =40<br />
Fig 21. Difference in stress z between two consecutive computer runs along the hole<br />
boundary for the analysis <strong>of</strong> N parameter<br />
D. Analysis <strong>of</strong> a Plate with Real Ply Sequence<br />
The previous parametric study has been conducted for symmetrical four-ply laminated<br />
composites having different stacking sequences and hole size to thickness ratio. The study<br />
presented in this section targets the stress analysis <strong>of</strong> the same laminated composite plate with<br />
the same material, but with a real number <strong>of</strong> plies and a real ply-sequence. The newly<br />
examined symmetrical stacking sequence <strong>of</strong> the 18 plies is defined by 0/0/0/ 45/45/45/<br />
90/90/90/ 90/90/90/ 45/45/45/ 0/0/0. The ratio <strong>of</strong> the hole size to the ply thickness (a/h) is<br />
chosen to be 20.<br />
Fig 22. Geometry <strong>of</strong> the Plate and the General Details <strong>of</strong> the Global Mesh (a/h=20)<br />
As before, the sub-modeling technique is used and two FE models and meshes (i.e., global<br />
and sub-models) have been used. The parameters used for the mesh for the zone <strong>of</strong> interest in<br />
the global model are defined as follows:<br />
N z = 2, R z . = 1<br />
N = 10<br />
N r = 5, R r = 15<br />
45
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
Figure 22 shows the geometry and FE mesh <strong>of</strong> the zone <strong>of</strong> interest with the previously defined<br />
parameters and the mesh for the rest <strong>of</strong> the global model.<br />
Fig 23. FE mesh <strong>of</strong> the sub-model based on the chosen mesh parameters (a/h=20)<br />
Figure 23 shows the geometry and the fine mesh for the sub-model where the following<br />
parameters are used:<br />
N z = 6, R z . = 10 (i.e. 3 elements on each half-thickness <strong>of</strong> the ply)<br />
N = 40<br />
N r = 15, R r = 50<br />
An Intel Core2 Duo PC with 2.33 GHz processor has been used to perform all the<br />
calculations. A hard disk space <strong>of</strong> about 160 MB was used to store all the vital files for the<br />
analysis. The time required to analyze the global model for plate with 18 symmetrical plies<br />
was less than one minute while the time required to analyze the sub-model was slightly less<br />
than 10 minutes and required about 2.4 Giga Bytes <strong>of</strong> storage. This computation power<br />
promises that a PC can handle a plate with number <strong>of</strong> symmetric plies more than 18, but with<br />
more demand on the storage space.<br />
E. Effect <strong>of</strong> Less Detailing <strong>of</strong> the Model through the Thickness Direction<br />
Fig 24. The FE mesh <strong>of</strong> the sub-model with fine mesh around all plies (a/h=20)<br />
46
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
Fig 25. The FE mesh <strong>of</strong> the sub-model with fine mesh around three plies only (a/h=20)<br />
Fig 26. Stress zz along the ply thickness for a model with fine mesh for all plies (a/h=20)<br />
Fig 27. Stress zz along the ply thickness for a model with fine mesh for the ply under<br />
consideration and its sandwiching plies (a/h=20)<br />
To reduce the problem size and the run time, the ply under consideration and its<br />
sandwiching plies can be finely meshed while the rest <strong>of</strong> the other plies can be coarsely<br />
meshed. Two analyses have been performed; the first for a sub-model mesh with N z =6 for all<br />
plies (Fig. 24) and the second for a mesh with N z =6 for the ply under consideration and its<br />
sandwiching ones and N z =2 for the rest <strong>of</strong> the plies (Fig. 25).<br />
47
KHALED M. EL-SAWY, ABDUL HAMEED ARWANI<br />
Figures 26 and 27 show the distribution <strong>of</strong> zz along the thickness <strong>of</strong> the considered ply. It<br />
is clear that there is no considerable difference in the results from both analyses which proves<br />
that fine meshing through the ply thickness is only required for the ply under consideration<br />
and its sandwiching plies.<br />
7. CONCLUSION<br />
A study is performed to select the optimum values for the parameters controlling the mesh<br />
used in the Finite Element analysis <strong>of</strong> laminated composite plates with holes. The mesh<br />
parameters considered are N z , R z , N r , R r , and N . The recommended values for the mesh size,<br />
arrangement, and parameters are as follows:<br />
1) The number <strong>of</strong> elements through the ply thickness N z depends mainly on the ply<br />
thickness. For thick plies (i.e. hole radius to ply thickness ratio a/h=4) N z can be taken as<br />
10. For medium ones (i.e. a/h=20) N z =8 is enough, while for thin ones (i.e. a/h=100) N z =6<br />
is sufficient.<br />
2) The spacing ratio R z for the elements through the ply thickness could take any reasonable<br />
value between 12 and 16.<br />
3) The number <strong>of</strong> elements in the sub-model along the radial direction N r depends on the<br />
hole radius to ply thickness ratio a/h. For a/h=20, N r could be taken as 13, while for<br />
a/h=100 and a/h=4, N r may be reasonably selected between 14 and 15.<br />
4) The spacing ratio for the element size in the radial direction R r can take any value<br />
between 40 and 60 since it does not affect the size <strong>of</strong> the problem.<br />
5) The number <strong>of</strong> elements along the hole circumference N may be reasonably chosen as 40.<br />
6) To reduce the size and run-time <strong>of</strong> the composite plate problem, the ply under<br />
consideration and its sandwiching plies can be finely meshed along the thickness direction<br />
while the rest <strong>of</strong> the other plies can be coarsely meshed with 2 elements along the ply<br />
thickness..<br />
8. ACKNOWLEDGMENT<br />
Support <strong>of</strong> this research was provided through a research project funded by the Dassault<br />
Aviation, France, and the United Arab Emirates University.<br />
9. REFERENCES<br />
[1.] R. B. Pipes & N. J. Pagano. “Interlaminar Stresses in Composite Laminates Under<br />
Uniform Axial Extension,” J. Composite Materials, Vol. 4, p. 204, 1970.<br />
[2.] D. Pullman & J. R. Schaff. “Three Dimensional Spline Variational and Finite Element<br />
Theory: A Comparison for a Laminate with a Circular Hole,” Structural Design and<br />
Analysis Methods, 1996.<br />
[3.] S. G. Lekhnitskii. “Theory <strong>of</strong> Elasticity <strong>of</strong> an Anisotropic Body,” Mir Publishers,<br />
Moscow, 1963.<br />
[4.] I. S. Raju & J. H. Crews, Jr. “Three Dimensional Analysis <strong>of</strong> [0/90]s and [90/0]s<br />
Laminates with a Central Circular Hole,” NASA-TM-83300, April 1982.<br />
48
OPTIMIZED 3D MESH FOR THE FINITE ELEMENT ANALYSIS OF LAMINATED COMPOSITE PLATE WITH CENTRAL HOLE<br />
[5.] W. E. Lucking, S. V. Hoa & T. S. Sankar. “The Effect <strong>of</strong> Geometry on Interlaminar<br />
Stresses <strong>of</strong> [0/90]s Composite Laminates with Circular Holes,” Journal <strong>of</strong> Composite<br />
Materials, Vol. 17, pp. 188-198, 1984.<br />
[6.] Y. Choi & E. S. Folias. “The 3D Stress Field in a Laminated Composite Plate with a Hole<br />
Based on an h-r Finite Element Method,” AMD-Vol. 196. Failure Mechanics in<br />
Advanced Polymeric Composites, ASME, 1994.<br />
[7.] J. Y. Kim & C. S. Hong. “Three-Dimensional Finite Element Analysis <strong>of</strong> Interlaminar<br />
Stresses in Thick Composite Laminates,” Computers & Structures, Vol. 40, No. 6, pp.<br />
1395-1404, 1991.<br />
[8.] T. Nishioka & S. N. Atlri. “Stress Analysis <strong>of</strong> Holes in Angle-Ply Laminates: An Efficient<br />
Assumed Stress ‘Special-Hole-Element’ Approach and a Simple Estimation Method,”<br />
Computers & Structures, Vol. 15, No. 2, pp. 135-147, 1982.<br />
[9.] W. C. Hwang & T. C. Sun . “A Finite Element Iterative Approach for Analysis <strong>of</strong><br />
Laminated Composite Structural Elements,” Computer & Structures, Vol. 31, No. 1, pp.<br />
55-62, 1989.<br />
49
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 51-73 (2010)<br />
© 2009 ALHOSN University<br />
AFFORDABILITY, ENABLEMENT, AND SUSTAINABILITY OF<br />
LOW-INCOME HOUSING<br />
Mahbubur Rahman *<br />
Universiti Teknologi Malaysia<br />
International Campus, 54100 Jalan Semarak, Kuala Lumpur, Malaysia<br />
ABSTRACT: Squatter settlements providing housing to 30-70% <strong>of</strong> the urban population in many developing countries<br />
have grown due to widespread poverty and inadequate housing finance and land development systems. Governments<br />
mostly assisted by the international aid agencies have improved environment, tenure security, income and resources in<br />
many settlements, but could not eradicate the problems as benefits did not multiply due to lack in institutional<br />
development, policy implementation, governance, participation etc. Moreover, the squatters’ capability to bring<br />
affordable and sustainable solutions was ignored. This paper discusses the changing approaches to the issues <strong>of</strong> lowincome<br />
groups housing worldwide in the above context, and examine the role <strong>of</strong> the World Bank in setting the trend.<br />
Keywords: Affordability, Enablement, Housing Policy, Self-Help, Squatter Settlement, Sustainability, Urban<br />
Development, World Bank.<br />
1. INTRODUCTION<br />
The emergence <strong>of</strong> cities as the unchallenged site <strong>of</strong> human development for the future and the goal<br />
<strong>of</strong> sustainable development have pushed hundreds <strong>of</strong> them to seek better means to encourage a<br />
synoptic view, act as levers for strategic change, and facilitate sustainable development (Holden<br />
2006). Rapid urbanisation taking place in the developing world's cities needs to be critically<br />
managed. There has been greater appreciation <strong>of</strong> their growing importance in the national<br />
economies recently as development became dependent on the ability <strong>of</strong> the urban centres to meet<br />
the essentials like housing. A rising standard <strong>of</strong> living and political ideologies have increased the<br />
awareness <strong>of</strong> human needs and social values (UKGP 1998), putting the onus to provide them,<br />
endorsed by the Habitat Conferences, on the government.<br />
The developing world’s urban population is going to be doubled by 2030; this will be accompanied<br />
by dramatic growth in the number <strong>of</strong> slums where a sixth <strong>of</strong> humanity lives today (UNCHS 2003).<br />
Given the social, economic and political situation <strong>of</strong> these people, these ill-built and ill-served<br />
houses were what most <strong>of</strong> them could afford (World Bank 1993), subject to exploitation due to<br />
scarcity (Tipple 1994). The global community was falling short <strong>of</strong> the Millennium Development<br />
Goal’s (MDG) target <strong>of</strong> significantly improving the lives <strong>of</strong> at least 100 million slum dwellers by<br />
2020 (UN 2000). 1<br />
This paper attempts to infer sustainable housing out <strong>of</strong> the concepts <strong>of</strong> sustainable development,<br />
compare the low-income group’s housing in the developing world against that, and establish the<br />
changing approaches to the low-income groups housing worldwide starting from the 1950s. It<br />
particularly examines the role <strong>of</strong> the World Bank, and other international bodies in setting the core<br />
1 Although slums are defined in a broader manner than inadequate housing, the UNHCS (2006) refers to it as the ‘shelter<br />
dimension <strong>of</strong> urban poverty’. According to it, the number <strong>of</strong> slum dwellers in the world has increased from 715 million in<br />
1991 to 913 million in 2001 and to 998 million in 2005, adding another 50 million in 2005-7. UNHCS (2007) projects that<br />
the world will have 1.4 billion slum dwellers in 2020.<br />
_____________________________________________________<br />
* Corresponding Author. Tel.: +6017683404، Fax: +60326934844<br />
E-mail : mmrahman417@gmail.com<br />
51
MAHBUBUR RAHMAN<br />
themes. It also highlights the advantages <strong>of</strong> self-built incremental in situ upgrading. The paper<br />
builds the case supporting such developments as an affordable means <strong>of</strong> providing sustainable<br />
housing to the low-income groups in the developing countries.<br />
2. SUSTAINABLE DEVELOPMENT<br />
According to the WCED (1987), sustainable development means meeting "the needs <strong>of</strong> the present<br />
without compromising the ability <strong>of</strong> future generations to meet their own needs". The idea emerged<br />
in the 1980s to bridge the gap between environmental concerns about the ecological consequences<br />
<strong>of</strong> human actions and socio-political concerns about development issues. While the<br />
conservation/preservation debate went on, the issues <strong>of</strong> population growth, pollution, and nonrenewable<br />
resource depletion got prominence (Robinson 2004). 2 Government and private bodies<br />
adopted the term sustainable development; academics used 'sustainability' reflecting the more<br />
managerial and incremental but less radical approach <strong>of</strong> the Brundtland report. Concerned that<br />
development was seen synonymous to growth, they focussed on the role <strong>of</strong> institutions,<br />
governance, and social capital in the process (Lehtonen 2004). Following will be a brief resume <strong>of</strong><br />
approaches to sustainable development and its implication in housing.<br />
2.1 Three Pillars:<br />
Initially not distinct, later sustainable development became a combination <strong>of</strong> ecological, economic,<br />
and social dimensions. This ‘triple bottom line’ constituted the basis for the generally accepted<br />
definitions <strong>of</strong> sustainable development (Lehtonen 2004). The international organisations endorsed<br />
the hierarchically equal, mutually interacting dimensions. While the importance <strong>of</strong> each pillar may<br />
vary from situation to situation, the model did not attribute any priority; it rather placed them as<br />
independent elements that can be analysed separately. Characteristics and logic specific to each <strong>of</strong><br />
them may conflict with others, with no basis to arbitrate between the conflicting objectives <strong>of</strong><br />
economic rationality, social justice and ecological equilibrium (Lehtonen 2004).<br />
While the three-pillar concept has been criticised for legitimising the current goals <strong>of</strong> the society,<br />
government agencies found favourable objectives to corroborate it. It also perpetuated the<br />
‘economism’ and ‘productivism’ characters <strong>of</strong> modern societies. Yet this articulated a new set <strong>of</strong><br />
checks and balances beyond the basic efficiency-equity, and cost- benefit binaries <strong>of</strong> traditional<br />
policy analysis. However, maintaining a positive balance in all three areas directly implies<br />
privileging limits and precaution over growth and accumulation.<br />
Philosophically, Norton (1999) defined sustainability as a social imperative, not ecological with<br />
social and economic implications. Until recently, sustainable development perceived as<br />
environmental issue not integrated into economic decision-making. The importance <strong>of</strong> social<br />
dimension in the overall economy— considered the weakest ‘pillar’ <strong>of</strong> sustainable development in<br />
its analytical and theoretical underpinnings, is only seen now (Forster-Kraus et al. 2009). Woolcock<br />
(2001) attributed the social interest to the fall <strong>of</strong> communism, the ostensible difficulties <strong>of</strong> creating<br />
2 Rather than a consensus-based understanding, the Brundtland report presents the term as a language truce about a set <strong>of</strong><br />
ideas, like democracy, freedom, or justice, over which wars could be waged (Mebratu 1998; Pezzoli 1997; Guha &<br />
Martinez-Alier 1997). For international development scholars and practitioners in environmental <strong>sciences</strong> and policy<br />
fields, sustainable development has become a universally integrative term, elevating the idea beyond urban planning and<br />
policy domains.<br />
52
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
market institutions in transitional economies, the financial crises in Latin America, East Asia, and<br />
Russia, and the persistent problems <strong>of</strong> unemployment and social marginalisation.<br />
The three-pillar model strengthens the idea that the economy can be detached from the social<br />
context that embodies all human activities, separating the ‘social’ from the ‘economic’ and denying<br />
the relations between human societies and their environment (Passet 1996). Moreover, though<br />
always prevalent, these are not qualitatively equal; disagreements over their hierarchy have<br />
prevented a model from being widely adopted. However, the social dimension that is <strong>of</strong>ten difficult<br />
to quantify cannot be analysed through the same framework and tools as the ecological or<br />
economic one (Empacher 2002; Dubois & Mahieu 2002). Passet (1996) suggested an alternative<br />
model where economic activities should serve all while also safeguarding the biophysical systems<br />
for their existence; the social would be in the command <strong>of</strong> the economic, and also to the ultimate<br />
environmental constraints.<br />
The hierarchy does not mean that the environment would always be the most important and<br />
relevant dimension. In a given situation, social or economic aspects may be more relevant as long<br />
as the operation <strong>of</strong> socioeconomic systems does not conflict with the environmental framework<br />
(Norgaard 1994).<br />
2.2 Sustainable Housing:<br />
Sustainability focussed on social and economic conditions in developing countries, their connection<br />
to environmental degradation, and the people’s coping ability. Policies advocated for sustainable<br />
development see urban development as that promotes economic growth, maintains social inclusion,<br />
and minimises environmental impact. Some adopters <strong>of</strong> sustainability frameworks, e.g. the European<br />
Commission (2001), created more directed approaches for building pr<strong>of</strong>essionals: “economic growth<br />
[that] supports social progress and respects the environment, social policy [that] underpins economic<br />
performance, and environmental policy [that] is cost-effective” (Roseland 2000).<br />
Sustainability is a political act based on human decisions and ways <strong>of</strong> life, not a scientific concept<br />
(Robinson 2004). According to Greider (1997), it “carries revolutionary implications” for urban<br />
planning and management, “but sounds so wholesome that almost everybody can endorse it”. Initially<br />
conceived as a term related to macro economic development, sustainability has been <strong>applied</strong> to<br />
housing only recently (Choguill 1999). Given the phenomena <strong>of</strong> urbanisation and severe housing<br />
problems in these cities where the greatest resource use occurs and which generate the most waste<br />
products and pollution, the concept has relevance to urban housing with economic, environmental<br />
and social implications.<br />
The complexity accompanying this shift is a reason for this delay. To remain meaningful, the concept<br />
<strong>of</strong> sustainable human settlements must stay within the absorptive capacity <strong>of</strong> local and global waste<br />
(Foy & Daly 1992), the achievement <strong>of</strong> the sustainable use <strong>of</strong> renewable (Daly 1992) and replenishable<br />
(Rees 1996) resources, the minimisation in the use <strong>of</strong> non-renewable resources (El Sarafy 1989),<br />
and meeting basic human needs (Hardoy et al. 1992). The last, distinguishing the definition from the<br />
more general environmental approaches to sustainability, guides the housing issues.<br />
The relationship between development and urban social conditions is complex. That social aspects<br />
<strong>of</strong> housing were as important as environment and economic dimensions <strong>of</strong> sustainability is<br />
increasingly realised (Forster-Kraus et al. 2009). The economic sector addressing the financial<br />
aspects <strong>of</strong> social justice, accompanying the environmental sustainability, is an important element <strong>of</strong><br />
53
MAHBUBUR RAHMAN<br />
it, while the environmental limits are an important constraining influence on economic growth.<br />
Social sustainability refers to “policies and institutions that can integrate diverse groups and<br />
cultural practices in a just and equitable fashion. As cities develop, the degree <strong>of</strong> social inequality,<br />
cultural conflict and political fragmentation experienced within urban boundaries has increased”<br />
(Stren & Polese 2000).<br />
The “free-market” city attempts to overcome both environmental and social challenges as their<br />
wealth increases. Thus growth not only increases the general wealth <strong>of</strong> society, it also increases<br />
scope for protecting environment by trickling down. In order to be sustainable, housing initiatives<br />
must be economically viable, socially acceptable, technically feasible and environmentally<br />
compatible; housing policies should target to achieve these aims. The ‘brown agenda’ laid down<br />
guidelines for sustainable urban development, followed by the UN calling the local governments to<br />
mobilise broad-based, participatory, and sustainable environmental improvement. Implementing<br />
the agreements reached at the 1992 Rio Summit required a concerted action at the local level<br />
(Agyeman & Evans 1996), focussing on community, participation, partnership, accountability, etc.<br />
3. HOUSING POLICY AND PRACTICE<br />
In the decades after the WWII, public housing became a major mode in the developing countries<br />
where meagre resources were spent on housing mostly for the government employees (Keivani &<br />
Werna 2001). Aimed at eliminating the 'ugly ' informal settlements, these could not overcome the<br />
escalating urbanisation, overcrowding, poverty, and informal settlement growth (Pugh 1995). Many<br />
<strong>of</strong> them did not meet the low income populations’ demand; 90% <strong>of</strong> the urban population depended<br />
on the private and informal sectors (Baross & van der Linden 1990; Drakakis-Smith 1981; Kevani<br />
& Werna 2001). For example, Grimes (1976) found that families in Ahmedabad, Bogota, Hong<br />
Kong, Madras, Mexico City and Nairobi could not afford to buy a ‘cheap’ subsidised government<br />
house.<br />
Critics <strong>of</strong> public housing, urban renewal, and modernist visions <strong>of</strong> urban development were<br />
growing during the 1960s. Abrams (1964) and Turner (1968, 1972, 1976a, 1976b), who influenced<br />
the theories and policies <strong>of</strong> low-income housing for decades, drew attention to the housing<br />
production process inherent in informal settlements (Choguill 1999). They brought the gross<br />
housing shortages and huge squatter settlements lacking in basic utilities in the cities in these<br />
countries into the world’s notice in the 1950s. By focussing on incremental housing <strong>of</strong> the lowincome<br />
groups, they initiated in situ upgrading <strong>of</strong> such settlements that was the main form for<br />
improving the living conditions <strong>of</strong> these people (Abbot 2002).<br />
Turner advocated sites-and-services and slum improvement schemes identifying the aspects <strong>of</strong> selffulfilment<br />
<strong>of</strong> the slum-dwellers and their commitment to housing shown in things they value (Pugh,<br />
2000). He defined an extended process by ‘freedom to build’ or as a ‘verb’. Furthermore, Turner<br />
(1976b) supported owner-built homes, however modest, to well-built public housing because <strong>of</strong><br />
their differing “structure[s] <strong>of</strong> authority and control”, i.e. capability to participate (Harris 2003) as<br />
they could best decide their needs (Choguill 2007). Such autonomy was fundamental to Turner<br />
(1968), who argued that squatter settlements improved over time within means, fitting their<br />
circumstances, is more affordable to both the dwellers and the government.<br />
Therefore, squatter upgrading and self-help would be more sustainable compared to mostly<br />
unaffordable typical public housing schemes that did not reach the target groups (Rahman 1999,<br />
54
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
2004). These were advocated consistently by experts and international agencies since the late 1940s<br />
(Harris 2003). It meant not only the investment <strong>of</strong> sweat equity, but also the processes <strong>of</strong> ownerdesign<br />
and management. Turner saw these as solutions, not problems (Mangin 1967; Drakakis-<br />
Smith 1981). Watts (1997) credited him for the “reorientation <strong>of</strong> shelter programs, away from slum<br />
clearance to the environmental improvement <strong>of</strong> squatter settlements; towards the setting-up <strong>of</strong><br />
policies to assist individuals to solve their own problems.”<br />
Aided self-help <strong>of</strong> the early 1960s too was more affordable than public housing though did not<br />
provide control. The later sites-and-services housing aimed to assist and repeat the success <strong>of</strong> the<br />
incremental house building and improvement process <strong>of</strong> the informal settlements to supply<br />
affordable housing units to the urban low income households. The provision <strong>of</strong> core units was the<br />
alternative that could be occupied relatively quickly by the household and extended when the<br />
occupier would afford the resources. The World Bank championing urban project assistance in the<br />
developing countries adopted Turner’s ideas; by the 1987 International Year <strong>of</strong> Shelter for the<br />
Homeless, these were well established, and remained broadly valid throughout. The following two<br />
subsections will summarize the approaches by the international agencies and governments to lowincome<br />
group’s housing, and analysis <strong>of</strong> spontaneous developments in the informal sector.<br />
3.1 International Community:<br />
The international agencies gave direction to the consulting community, governments, and the UN<br />
(Choguill 2007) as “a trendsetter for development thinking” (Huchzermeyer 1999). Abbott (2002)<br />
identified the radical shift from public housing, mandating self-help through sites-and-services and<br />
in situ slum upgrading when “the housing policies <strong>of</strong> almost all developing nations were devoted to<br />
complete houses for the poor" (Peattie & Doebele 1973). 3 Realising that this could not solve<br />
housing problems, international bodies and governments started assisting the poor to build their<br />
own houses in a cost-effective way after the 1960s.<br />
1972–82, 1983–93, and post-1993 were the phases when the international agencies providing<br />
financial assistance modified their housing development strategies (Pugh 2000), drifting from a<br />
focus on self help to a holistic development, relating housing to other development sectors<br />
(Kessides 1997). Full cost recovery was essential in the World Bank’s funding <strong>of</strong> sites-and-services<br />
schemes since 1972 as large-scale subsidies were infeasible (Choguill 2007; World Bank 1972,<br />
1973).<br />
Early planning evolved to manage housing development on vacant plots that had many political,<br />
pr<strong>of</strong>essional, funding and management advantages (Abbot 2002). Still unaffordable to at least 20%<br />
<strong>of</strong> people, it was found that sites-and-services schemes could not be sustained (Swan et al.1983);<br />
the site-based projects did not multiply benefits, address the subsidy issues, eradicate poverty, or<br />
increase land ownership to enhance the Poor’s access to housing, and hence were not replicable.<br />
Also, planning and building regulations based on western ideals inhibited an expansion <strong>of</strong><br />
affordable mass housing, with scope for extension and remodelling, employment generation, etc.<br />
3 Geddes had endorsed assisted self-help as early as 1918 and by the late 1950s its variations by Abrams (1964), Atkinson<br />
(1961) and Weissman (1960) were familiar. So, too, had major international agencies (Harris 2003) that incorporated<br />
economic arguments in favour <strong>of</strong> self-help in their wisdom. During the 1950s, they justified it for policies <strong>of</strong> assistance to<br />
owner-builders. The World Bank in its first country report acknowledged the economic importance <strong>of</strong> self-help in<br />
Colombia. By the early 1960s the economic benefits <strong>of</strong> self-help were part <strong>of</strong> conventional wisdom (Frieden 1965), taken<br />
for granted by many <strong>of</strong> the contributors to the collection written for a US Senate subcommittee in 1963.<br />
55
MAHBUBUR RAHMAN<br />
The 1976 Vancouver Habitat Conference recommended that the governments intervene to supply<br />
housing resources that involves large infrastructure and investment. The 1996 Istanbul Conference<br />
furthered through public-private partnership involving the stakeholders to identify and transform<br />
priorities into action plans: creating institutions for urban environmental improvement, and building<br />
capacities to participate and cooperate. The policy shift towards sustainability accepted informal<br />
settlements to bring a lasting development (World Bank 1991).<br />
The same standard <strong>of</strong> service delivery in settlement upgrading as in formally planned ones was<br />
ineffective. Thereafter, leaving low-income infrastructure provisions to the community to plan,<br />
build and manage themselves led to “progressive improvement model” (Choguill 1999). Similar to<br />
settlement upgrading, service provision could be gradually upgraded, and thus be affordable and<br />
sustainable. This sought to match the level to the ability <strong>of</strong> either the community or the local<br />
authority to manage in a sustainable way. Open and wide ranging in interventions, progressive<br />
infrastructure was successful in the Sri Lanka million houses program, and in other South Asian<br />
cities, owing to community decision-making.<br />
Sustainability cannot be brought without making the economy, environment and society parts <strong>of</strong> an<br />
overall development (Barbier 1988). Institutional reform had to be backed by comprehensive urban<br />
policies to sustain programs in the long run (World Bank 1983). Yet project-oriented theory,<br />
practice and policy <strong>of</strong> self help schemes that were to be translated into overall changes remained<br />
strong due to available funding, and instant benefits for all (Rahman 1999). In some cases only, e.g.<br />
in Karachi and Indonesia, large-scale upgrading became an essential part <strong>of</strong> the physical planning<br />
process <strong>of</strong> the urban areas affecting all informal settlements (Saleem 1983; Silas 1983).<br />
Redirected policies in the 1990s focussed on institutional reform. Gradually withdrawing from direct<br />
physical involvement, agencies generated and channelled funds in social housing with self-help<br />
components through structured finance and purpose-built institutions, the NGOs, 4 and CBOs. This<br />
allowed faster fund disbursement, better chance to reach target groups, and increase recovery by<br />
involving small groups. 5 The projects recovered the cost by setting target by affordability, and thus<br />
remained financially sustainable. Following the motto <strong>of</strong> affordability-accessibility-replicability<br />
(Choguill 1987), 6 this was more successful in countries with already developed housing finance<br />
systems (Rahman 1999). 7<br />
4 In addition, the NGOs can assist squatter settlements by helping to develop appropriate community organisations and<br />
mobilisation, provide technical and organisational skills for aiding self help house building and increasing their access to<br />
housing finance through creating financial co-operatives, provide loans directly or acting as intermediaries with formal<br />
commercial banks as guarantors for low income borrowers (Mitlin & Satterthwaite 1992; Igel & Srinivas 1996; Desai<br />
1996; UNCHS 1996a, 1996b; Rahman 1999).<br />
5 The Grameen Bank in Bangladesh developed credit and technical advice for women's enterprise, for housing, and for<br />
transforming social development among the poor; international agencies are supporting such initiatives (Rahman 1999).<br />
6 Accordingly, capital costs were to be set by the target group’s ability and willingness to pay, not by planning ideals and<br />
design standards. But cost recovery formula followed old convention where all components had to be paid for. It<br />
matched the Bank's priority for recovery so that it could repay the international credit and make economically and socially<br />
responsible uses <strong>of</strong> grants (Rahman 2004; Pugh 2000). At the end, the successful projects identified, improved and<br />
mainstreamed (Abbot 2002) could be replicated in similar situations elsewhere (Choguill 1987; Pugh 2000).<br />
7 In Bangladesh as international project grants were coming after the 1987 and 1988 floods, the government established a<br />
rural employment generation foundation to disburse and monitor them, instead <strong>of</strong> the central bank. An allocation <strong>of</strong> US$<br />
250 million by the World Bank in 1988 in India helped the Housing Development Finance Corporation to extend its<br />
credit coverage down the income ladder and stimulate local housing finance institutions. With Bank loans Chile<br />
introduced housing vouchers for sites-and-services schemes. In Brazil's Parana Market Towns Improvement Project in<br />
56
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
However, costs could be seldom recovered, project sites were remote from employment<br />
opportunities and unaffordable, institutional capability and expertise to implement and monitor was<br />
<strong>of</strong>ten weak, and therefore gentrification compounded the problems (ADB 1983), corruption was<br />
<strong>of</strong>ten rampant, inhibiting accessibility, and the projects did not lead to wider socio-economic<br />
development <strong>of</strong> the entire urban areas (Pugh 1990; Nientied & van der Linden 1985; Skinner et al.<br />
1987; Turner 1980). Moreover, outputs failed to make any qualitative or quantitative impact<br />
(Rahman 1999).<br />
Thus residents’ housing and non-housing choices do not need to be constrained to encourage<br />
affordable housing. It rather has the potential to improve the social sustainable environment in<br />
communities and cities (Forster-Kraus et al. 2009).<br />
Such holistic approach is set to develop finance more, reduce the backlogs, increase inadequate<br />
infrastructure, reform negative land management and land policy, introduce financial transparency<br />
to accelerate supplies in low-income housing, increase competitiveness <strong>of</strong> the construction industry,<br />
and establish or reform institutions (Pugh 2000). Social effectiveness was brought by decentralising<br />
responsibility <strong>of</strong> maintenance and cost recovery through the beneficiaries’ groups. Compared with<br />
the earlier shelter-oriented approach, approaches <strong>of</strong> the 1980s and 1990s prioritised broader and<br />
deeper institutional reforms and development, and creating a strong base for reorientation <strong>of</strong><br />
policies in the future. Also funding through the local government bodies proved appropriate in the<br />
backdrop <strong>of</strong> weak financial markets in many developing countries.<br />
Recent thoughts on development policy, instigated by Stiglitz (1998) and Wolfensohn (1999), drifted<br />
away from the above concept, which were followed by many developing countries that favoured<br />
opening up the market. The Bank (1999a) emphasised on broader urban issues in the late 1990s to<br />
enhance and sustain economic growth and modernisation. Thus improvement <strong>of</strong> living qualities,<br />
poverty reduction, job creation and production, environmental sustainability, and enhancement <strong>of</strong><br />
agglomeration economies were included as strategies to strengthen a balanced urban development<br />
(World Bank 1999). This gave priority in program finance and management to improve squatter<br />
settlements.<br />
3.2 Housing Process:<br />
As low income groups are unable to acquire good formal sector housing, unconventional or<br />
informal housing remains a necessary part <strong>of</strong> urban growth in the developing countries (UNCHS<br />
1996c; Gilbert 1990; Drakakis-Smith 1981). Appearing disorganised and inadequate initially, these<br />
with their own logic can be gradually upgraded as needs are felt and resources, including fund,<br />
materials, labour, etc. are available (Angel & Benjamin 1976). Many squatters go on to consolidate<br />
and improve their housing over time (Drakakis-Smith 1981; Turner 1976a), slowly shaping<br />
communities. 8 Low-income people can live in incomplete shelters to be improved later. Thus<br />
1983–88, a few towns created a municipal fund out <strong>of</strong> World Bank’s seed fund. The credit terms required sustainable<br />
finance through cost recovery achieved by skilled management <strong>of</strong> transactions. Yet, the local governments and<br />
households could have their own sub-projects, select the price according to their need, priority and affordability, through<br />
participation (World Bank 1994).<br />
8 This is not applicable to all squatter settlements or all in such settlements. In many Asian and African squatter settlements,<br />
the squatters do not move on from their initial stage <strong>of</strong> occupation (Shakur 1987). In Latin America where the squatters<br />
consolidated through improvement, there is a distinction between dynamic settlements and static and stagnant ones<br />
(Drakakis-Smith 1981). Owner occupation, strong group cohesion and the emergence <strong>of</strong> political leaders and CBOs<br />
57
MAHBUBUR RAHMAN<br />
compared to formal sector that ignores the needs <strong>of</strong> survival and flexibility <strong>of</strong> the low and<br />
intermittent income <strong>of</strong> the poor (Smets 1999), incremental upgrading distributes the affordable<br />
consumption and saving over time. This increases sustainability.<br />
Lower income dwellers can house themselves respectably at much lower than the formal sector<br />
cost as a large degree <strong>of</strong> the construction work would be carried out by the occupants using<br />
unconventional materials and techniques. The ‘process’ rather than the 'product' approach can halve<br />
the initial requirements for building compared to formal construction costs (Benjamin & McCallum<br />
1985), in exchange for social obligations to be met by the family over a period. The selfmanagement<br />
replaces up to a third <strong>of</strong> the labour cost (Payne 1983); 9 participatory environmental<br />
improvement is a saving too (Pugh 1994). In reality though most self help projects involve some<br />
degrees <strong>of</strong> paid labour and contract building; 92% <strong>of</strong> households in the sites-and-services projects<br />
in Lusaka used hired labour (Tipple 1994).<br />
A house is the largest commitment <strong>of</strong> a family (Maliene & Malys 2009). Despite a humble result,<br />
house building is an apocalyptic event for a low-income family, co-opting family members,<br />
marshalling all physical and monetary resources, collecting debts, and calling upon the community<br />
and the family for assuming new debt and obligations. Extended over time, this is the beginning <strong>of</strong><br />
a longer commitment to constantly make improvements and additions. Thus these turn out to be<br />
more acceptable and suitable to the socio-economic needs <strong>of</strong> lower-income people (Turner 1976a),<br />
and so are more affordable and sustainable.<br />
Informal sector self-help and self-built provide housing to 30–70% <strong>of</strong> urbanites in many developing<br />
countries (Keivani & Werna 2001). Self-management where skilled crew and hired labourers do<br />
most works, popular among the middle and upper-income groups, works with incremental building<br />
process too. The least common way is by the contractors for the wealthier people or organisations;<br />
better-<strong>of</strong>f is a society more <strong>of</strong> such houses emerge (Peattie 1987).<br />
Secure or expected occupancy rights motivate expression <strong>of</strong> built form in squatter settlements.<br />
While extending shelter, occupiers mark own aesthetic identity on it. This becomes sustainable due<br />
to a commitment <strong>of</strong> low-income people to place and home (Turner 1976a), perceiving the<br />
improvements as a part <strong>of</strong> wider resident activities are. Although the resourcing and organisation <strong>of</strong><br />
improving infrastructure and making personal investments to squatter houses have been discussed,<br />
enhancement <strong>of</strong> aesthetics and cultural amenities was less discussed. Against such functional<br />
dominance, Marcus (1995) focused on the personal meaning residents attach to their home that<br />
leads them to improve housing.<br />
Such deep human expression in self-help improvement is instigated by tenure form and length<br />
(Rahman, 1999; Pugh, 2000). 10 The mark <strong>of</strong> environmental change, local culture, and design and<br />
construction knowledge shows colour, adaptation, and space for rituals and festivals; and create<br />
specific and varied living environments (Rapoport 1988). A make-shift shack— the outcome <strong>of</strong><br />
which induce investment in housing and influence national political leaders for recognition and facilities were the<br />
distinguishing features <strong>of</strong> the former. The latter was characterised by high levels <strong>of</strong> renting, low group cohesion and<br />
community representation (UNCHS 1996c; Gilbert 1990). the level <strong>of</strong> owner occupation that induces both personal<br />
investment and group cohesion was the determining factor in this situation (UNCHS 1996c; Gilbert 1990).<br />
9 Materials cost is reduced by buying recycled and used items in informal sector. Family labour is usually free; skilled labour<br />
can be bartered for. Gerrul (1979) calculated that in lower-income housing, 35% labour is self-help; another 60% is semiskilled.<br />
10 For example, San Miguel, Mexico, or Klong Toey squatter settlement with 70,000 families in Bangkok's Port Authority land.<br />
58
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
rational thinking utilising limited available resources, reveals beliefs, aspirations, and the worldview,<br />
simultaneously impacting the political, the visual, and the cultural. As it is ‘architectural’<br />
same way as a ‘designed’ building, designers and planners could learn from these spontaneous<br />
open-ended, multi-sensory, semi-fixed settlements adding on elements, as architecture is about<br />
human drive, vision, interest and the identification <strong>of</strong> place (Unwin 1997).<br />
Most housing solutions focused on price reduction to close to the households’ ability to pay. These<br />
ignored the low-income families’ willingness to make extra effort to match financial requirements.<br />
With ownership prospect, household can readily devote more <strong>of</strong> their meagre resources, both<br />
monetary and commitment <strong>of</strong> non-monetary like spare time (Ward 1984a; Rahman 1999). Though<br />
low-income families have little savings, some <strong>of</strong> them could amass housing resources (Keare &<br />
Jimenez 1983), <strong>of</strong>ten by selling essential items (Rahman 2008). They improve affordability by<br />
using allocated space (CIVIS 2003), for example by renting out or using as workshops, <strong>of</strong>ten<br />
involving more family members (Setshedi 2006; Mai & Shamsuddin 2008). 11<br />
4. POVERTY AND SUSTAINABILITY<br />
Poverty is about the lack <strong>of</strong> capability to expand social opportunity in markets, in state policy, and<br />
in households; these development requisites focus on the freedom <strong>of</strong> individuals to choose values<br />
and lives worthy to them (Drèze & Sen 1995; Sen 1999; Sen & Wolfensohn 1999). Personal<br />
commitments and appropriate human bondage generated in low-income housing through freedom<br />
and control over the process can lead to poverty reduction. Fogel (1994) argued that improved<br />
nutrition, the advancement <strong>of</strong> medical knowledge, and the qualities <strong>of</strong> housing increase health and<br />
economic productivity over long-term development transitions. The principles are evident in the<br />
low-income housing requiring social co-operation to improve environmental conditions in a<br />
sustainable way.<br />
Environmental regulation combined with market based town planning, can play a part in<br />
ameliorating the slum problems if sufficient resources and powers are vested. The market<br />
sensitivity to building performance and impact is usually inhibited by price inflation linked to<br />
excessive demand for social, economic and other benefits. Integrating poverty alleviation and<br />
environmental improvement, Brundtland concluded that “the problems are entwined such that<br />
ecological sustainability cannot be achieved if the problem <strong>of</strong> poverty is not addressed” (Robinson<br />
2004). This linked the vast and complex issue <strong>of</strong> environmental deterioration with the equally vast<br />
and complex issue <strong>of</strong> human development, both affected by poverty.<br />
The mainstream now views environment differently from when related problems were regarded as<br />
minor, technical, soluble, and politically uncontentious in the 1980s. Economic growth and social<br />
progress could duly solve these, as increasing wealth created the resources including technology.<br />
Today, it is acknowledged that environmental problems require significant social and economic<br />
changes, not merely technical solutions. Sustainable development is the vehicle for this change,<br />
addressing the conflict between environmental protection and economic growth. It accepts that the<br />
former requires fundamental change in the direction <strong>of</strong> economic progress and institutional policy.<br />
But it argues that this is compatible with continued economic growth in a (regulated) global<br />
capitalist system.<br />
11 Projects would <strong>of</strong>ten be designed to integrate productive use enhanced by incentives like space and credit in Cairo,<br />
Mexico, Nairobi and Senegal (Ward 1984b), or use future income sources to assess affordability (Lee 1985).<br />
59
MAHBUBUR RAHMAN<br />
Favouring the human-centred nature <strong>of</strong> the Brundtland Report, the reformists suggested that the<br />
answer to both over- and under-consumption, i.e. the environmental concerns, lay in promoting<br />
more <strong>of</strong> sensitive human development that requires improvements in technology and efficiency.<br />
Devoid <strong>of</strong> spiritual values, or individual responsibility, it rather focused on collective institutional<br />
responses and social responsibility, and embodies an anthropocentric approach around human<br />
needs.<br />
The WCED recognised poverty as a major source <strong>of</strong> environmental degradation. Yet development<br />
agenda revolved around macro-economic stabilisation through relentless export-led growth and<br />
market liberalisation <strong>of</strong> the early-1980s. The 1997-98 financial crises exposed the limitations <strong>of</strong><br />
such approaches followed by many developing countries— the absence <strong>of</strong> a broad basis in the<br />
politics <strong>of</strong> socio-economic development. Countries like Thailand and Indonesia succumbed to<br />
international speculation and bad market governance (Pugh 2000).<br />
Stiglitz (1998) favoured medium-term strategic development policies to alleviate poverty through<br />
socio-economic transformation. He advocated holistic societal changes understood as development<br />
transitions, for example, improving the environmental and health dimensions, the changing<br />
volumes and characteristics <strong>of</strong> poverty, etc., not isolated development <strong>of</strong> individual sectors. In an<br />
overall context, such development policies could use transitions in a combination <strong>of</strong> varying<br />
emphasis on different sectors, based on the pragmatic context-based realities, and socio-economic<br />
opportunities.<br />
The WECD downplayed the extent to which poverty alleviation and environmental improvement<br />
result from wealth (Roseland 2000). Trainer (1990) was dismayed that it chose economic growth<br />
and attendant social and environmental impacts (e.g. tendency to exploit labour and the<br />
environment) over a consciously appropriate development strategy for the developing world (e.g.<br />
adequate housing and clean water rather than industry and export). NTFEE (1987) stated that<br />
“sustainable economic development does not require the preservation <strong>of</strong> the current stock <strong>of</strong> natural<br />
resources or any particular mix <strong>of</strong> human, physical and natural assets. Nor does it place limits on<br />
economic growth, provided that is both socially and environmentally sustainable.”<br />
The oxymoron that comes from combining sustainability with its greatest threat, development, is<br />
paralleled by calls to combat exclusion <strong>of</strong> the low-income group (Clark 2001), and restore a<br />
broader social and economic purpose <strong>of</strong> housing equity. However, the aspiration is not matched by<br />
the ideological commitment to provide the resources and powers necessary to change the status<br />
quo; politicians and bureaucrats may adopt language and sentiment without the will or means, or<br />
empowering the poor (Rahman 1999). Only a process <strong>of</strong> political mobilisation and mass education<br />
can change assumptions and behaviour. Under ideal conditions acquisition and incorporation <strong>of</strong><br />
intelligence replaces rhetoric, informs policy and leads to a more measured approach to current<br />
anxieties. But most <strong>of</strong> the time we deal in a less scientific or certain way as rhetoric and vision<br />
influence priorities and decisions.<br />
Sen (1987, 1999) criticised the way neo-institutional and ecological economics direct towards the<br />
conventional economic theories, based on individual capabilities and the concept <strong>of</strong> ‘social capital’,<br />
to address the social dimension <strong>of</strong> sustainable development. This meant the alternative<br />
combinations <strong>of</strong> functionings a person can achieve that in turn denotes the various things he may<br />
value— from elementary like shelter to complex like community participation (Sen 1999). Policies<br />
should not focus on collective outcomes such as the distribution <strong>of</strong> income, but rather on building<br />
60
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
individual capabilities, and ensuring that people have the freedom to convert economic wealth into<br />
desirable outcome.<br />
Ballet et al. (2003) defined socially sustainable development as one that “guarantees an<br />
improvement <strong>of</strong> the capabilities <strong>of</strong> social, economic or environmental well-being for all, through<br />
the aspiration <strong>of</strong> equity on the one hand, as intra-generational distribution <strong>of</strong> these capabilities, and<br />
their transmission across generations on the other hand”. He extended the notion <strong>of</strong> capabilities<br />
from individuals to cover also societies. The structure <strong>of</strong> capabilities expresses the adaptation <strong>of</strong> an<br />
individual or a society to a number <strong>of</strong> external constraints.<br />
The capability approach emphasises on the improvement <strong>of</strong> social conditions from one generation to<br />
another, and on the interactions between the three spheres <strong>of</strong> development—social, environmental<br />
and economic. In designing policies, not only the effects <strong>of</strong> economic and environmental actions on<br />
the social dimension, but also decisions within the social sphere itself have to be looked into. This<br />
expresses both an individualistic and a social point <strong>of</strong> view—on one hand, it looks at the capabilities<br />
<strong>of</strong> rationally and responsibly acting individuals, and on the other hand, at the social capabilities <strong>of</strong> a<br />
society and the roles <strong>of</strong> social actors. These are not necessarily in harmony with each other, since the<br />
improvement <strong>of</strong> education, health, employment, etc. may have harmful effects on some groups and<br />
thus even threaten social cohesion.<br />
Social policies such as poverty reduction programs may in fact adversely affect certain capabilities,<br />
with an increasing vulnerability <strong>of</strong> individuals and social inequalities as a result. Actors are thus<br />
themselves to decide which capabilities are to be considered. Finally, freedom is a key element in<br />
Sen's approach—all evidence seems to show that even the poor attach significant value to freedom,<br />
they may not seem ‘rational’ from traditional economic perspective <strong>of</strong> maximising utility. Thus<br />
arguments about self-help and identity in housing make more sense in terms <strong>of</strong> sustainability. The<br />
following subsection will discuss another important issue <strong>of</strong> sustainable housing, i.e. <strong>of</strong> increasing<br />
the capacity.<br />
4.1 Enablement:<br />
Reducing the state’s involvement in directly providing housing and expanding the role <strong>of</strong> the<br />
private market was accepted in the 1980s, pursued by most developing countries irrespective <strong>of</strong><br />
ideology or political structures (Israel 1990). Governments for long have played the hopeless role<br />
<strong>of</strong> ‘provider’ by constructing housing to directly solve the shortages in various countries. The<br />
enablement strategy, given birth by an urgent need for scaling up housing supply, was aimed at<br />
creating a congenial economic and social framework to enhance economic efficiency and social<br />
effectiveness so that capability to solve own housing problems could grow. This could be met by<br />
expanding the role <strong>of</strong> the private markets, rather than relying on project based approaches that<br />
could not alleviate housing shortages.<br />
Given the private sector dominates housing, enablement can expand housing production on a<br />
national scale by public support for the activities <strong>of</strong> the formal/informal markets and for the self<br />
help activities <strong>of</strong> the low income households. Attention was directed toward devising ways <strong>of</strong><br />
providing the financial, legal and institutional support (UNHSP 2005). Therefore, the international<br />
agencies started to encourage the creation <strong>of</strong> an enabling environment emanating into deregulation<br />
and institutional development <strong>of</strong> specially the land and housing markets to overcome the external<br />
constraints (LaNier et al. 1987; Kimm 1987; Loh 1987).<br />
61
MAHBUBUR RAHMAN<br />
This shift resulted in the adoption <strong>of</strong> a new World Bank housing sector policy paper, and formed<br />
the basis <strong>of</strong> the Global Strategy for Shelter to the Year 2000. This 1993 policy statement<br />
emphasised enablement, the sector's contribution to macroeconomic development, and pro-poor<br />
policies involving subsidies targeted only to the poor where required. It proposed the enablement <strong>of</strong><br />
private housing markets for scaling up housing production and developing the sector as a whole.<br />
Similarly, UNCHS in June 1996 adopted ‘adequate shelter for all and sustainable human<br />
settlements’ in the Habitat II, founding its shelter policies and recommendations on enabling<br />
strategies for private markets (UNCHS 1996a).<br />
Enablement would bring together technical know-how, capacity <strong>of</strong> development agencies, use<br />
available resources, and recognise and define responsibilities <strong>of</strong> all stakeholders, through an<br />
inclusive participatory approach among residents from all strata. The roles <strong>of</strong> each such partner in<br />
the multi-institutional and multi-organisational environment could be guided by the underlying<br />
socio-economic rationale. For example, private enterprises could contribute efficiency and<br />
entrepreneurship, CBOs could mediate between households and government agencies that provide<br />
urban management expertise, and the participants could provide various finances, self-help<br />
resources, and localised relevance in the upgrading efforts (Pugh 2000). These represent a complex<br />
process, with some risks <strong>of</strong> failure due to weak institutions, narrow coterie interests, corruption and<br />
market manipulation found in the developing countries.<br />
Such enablement framework was also relevant to new housing for other income groups too, where<br />
it could supply housing, and builders could access development finance in a competitive market.<br />
Implementation <strong>of</strong> proper land policies by coordinating with the infrastructure and utility agencies<br />
could ensure adequate supplies <strong>of</strong> well-placed ready land at affordable price. The legal system<br />
could protect property rights <strong>of</strong> all, developed finance institutions could introduce attractive<br />
instruments to generate and manage funds. The overall policy and enablement framework could<br />
have pro-poor and egalitarian elements for social-relevance and sustainability.<br />
Enablement brought sustainability in sites-and-services schemes in India, and in the small loan<br />
program in Sri Lanka, Hong Kong, Singapore, and Chile (Pugh 1997) that alleviated their housing<br />
crises. However, most developing countries could not achieve the effectiveness or the<br />
comprehensiveness <strong>of</strong> Chile or Singapore due to poor finance sector and institutions. Most <strong>of</strong> the<br />
countries not meeting the targets due to institutional incapacities could overcome the situation by<br />
enablement.<br />
Thus in the 1990s, it was evident that benefits could not be sustained without good governance<br />
(Rahman 1999). Therefore, within institutional-led reform, ‘enablement’ put governance into<br />
central positions in economic, education, health, environment, housing, urban and other sectors<br />
with a focus upon state–market–society relations. This encouraged community-based, participatory<br />
elements in upgrading the squatter settlements and owning community assets so that processes<br />
were more transparent and accountable and people were enabled to improve themselves (Rahman<br />
1999).<br />
The project-linked participation <strong>of</strong> the Sri Lanka Million Houses program (Lankatilleke 1990) had<br />
planning, design, implementation and maintenance stages. Hamdi and Goethert (1996) identified<br />
planning as crucial for the community and the city to jointly take key decisions and define the<br />
program. Crane (1950) viewed self-help as part <strong>of</strong> a process <strong>of</strong> community development, requiring<br />
and encouraging cooperation. The Recife Declaration on community control over decision-making<br />
_____________________________________________________<br />
* Corresponding Author. Tel.: +6017683404, Fax: +60326934844<br />
E-mail : mmrahman417@gmail.com<br />
62
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
stressed the importance <strong>of</strong> the integration <strong>of</strong> the informal city. However, the support to community<br />
participation in the 1980s was lost later (Abbot 2002).<br />
The enabling concept has been subject to much debate and criticism for its over-concentration on<br />
the private sector to work, and shunning other modes <strong>of</strong> housing (Keivani & Werna 2001).<br />
5. CONCLUDING DISCUSSIONS<br />
Despite efforts by all, continuously growing squatter settlements occupy a large section in many<br />
cities. 12 These have shown sustainability amidst squalor and disease. The locales for the life's<br />
drama and contributions in huge urban and socio-economic transformations, this has remained a<br />
dominant form <strong>of</strong> dwelling for many decades. These are necessary and important in terms <strong>of</strong> both<br />
product and process, and socio-economic and physical utility (Kellett & Napier 1995). The people<br />
living in these settlements are set in specific institutional conditions and processes that influence<br />
their housing and social status. 13 The intricacy, variety, accomplishment, and resource efficiency in<br />
squatter settlements and built forms also have social, cultural, economic, political, and architectural<br />
implications (Pugh 2000).<br />
Varying in theoretical, economic and technical characters, the role <strong>of</strong> spontaneous and informal<br />
housing in the developing countries urban development was recognised late (Lawrence 1997; Ling,<br />
1997). 14 Assisting self-help programs from 1950s, the funding agencies changed their methods from<br />
site-specific projects to programs mediated through formal institutions, and subsequently to<br />
developing policies, cooperation and participation. Some <strong>of</strong> the projects have also added economic<br />
and <strong>of</strong>ten aesthetic value to urban assets. 15<br />
Crane (1950) attacked the paternalistic thinking, opposing claims that only architects and allied<br />
pr<strong>of</strong>essionals can deal intelligently with housing. Turner (1976b) too condemned architect-designed<br />
low-income public housing for being expensive and authoritarian. Back (1962) and Safa (1964)<br />
found that such regimentation and lack <strong>of</strong> choice were disliked by occupants, whereas self-help<br />
12 Slum population in India has more than doubled in the past two decades; in 2001, 54.1% <strong>of</strong> Mumbaians lived there<br />
(NIHFW, 2006). Dharavi, Asia's second largest slum (Jacobson 2007) in central Mumbai houses 800,000 people (Davis<br />
2006). A quarter <strong>of</strong> Sao Paolo population lives in poor conditions. With more slums dwellers, Kolkata has a higher slum<br />
density. Based on water and sanitation access, 99% <strong>of</strong> Afghan and 94% in Central African Republic people live in slum<br />
condition; even a third <strong>of</strong> the Argentines experience the same. China had nearly 175 mil living in slum conditions, 158 mil<br />
in India, 42 mil. in Nigeria and 36 mil in Pakistan (UNFPA 2007). 327 million people live in slums in Commonwealth<br />
countries - almost one in six <strong>of</strong> its citizens (Comhabitat 2006). In 11 African, 2 Asian and 1 Pacific countries urbanizing<br />
rapidly, over two third urbanites live in slums.<br />
13 The gradually transformed and consolidated Santa Marta settlement, Colombia, is recognised and accepted into the formal<br />
sector. In Durban, spontaneous settlements are juxtaposed near formal settlements, being impermanent and temporarily<br />
linking kith and kin.<br />
14 Self-help construction existed in Roman towns, medieval Europe, and in early periods in the developed countries. Murals<br />
were found in Pompeii on this. The Swedish ‘magic houses’ were built by families in their free times; the local government<br />
provided land, materials and technical advice. In ancient Sri Lanka, self-help housing was the norm in urban settlements,<br />
e.g. Anuradhapura founded in 483 BC.<br />
15 Cross-subsidisation in land pricing and allocation enabled sites-and-services and squatter improvement programs to reach<br />
the poor in the 1970s and 1980s in Chennai. Housing investment and wealth <strong>of</strong> all income groups increased; the<br />
contracting between the World Bank, state government, and the project authorities blended state, market, and household<br />
self-help roles (Pugh 1990a, 1997). In the Kampung Improvement Program, the World Bank provided US$ 439 million in<br />
project loans, to improve living conditions, housing investment, incomes, and health. Some <strong>of</strong> its lessons led to wider<br />
community participation and deeper institutional reforms (World Bank 1995).<br />
63
MAHBUBUR RAHMAN<br />
provided a variety <strong>of</strong> social, psychological, and economic advantages that <strong>of</strong>fered a breakthrough in<br />
housing problem (Frankenh<strong>of</strong>f 1967).<br />
Though self help, household economics, affordability, and home sense could describe the roles <strong>of</strong><br />
individuals and households, the economists studied only market exchange value, the design and<br />
impact <strong>of</strong> subsidies, or social questions <strong>of</strong> poverty and inequality (Pugh 1997; Stretton 1976). They<br />
ignored non-economic resources as socio-economic assets, like time and energy used for home<br />
building, domestic chores, income generation, physical improvements, childrearing and human<br />
capital formation, and personal and community activities; most <strong>of</strong> these important aspects <strong>of</strong> lowincome<br />
housing are in deed measurable.<br />
Self help goes beyond the construction and management <strong>of</strong> housing and the environment. The<br />
domestic sector is more sustainable as it uses own resources to produce home-based goods and<br />
services, depending less on imported materials and technology. Those without proper housing lead<br />
diminished and socially excluded lives, unable to participate fully in the community; their further<br />
development depends on access to services and the security <strong>of</strong> a safe and healthy environment. Basic<br />
housing is an ‘entry fee’ providing foothold to subsequently access such other benefits (Peattie 1987).<br />
Market forces and <strong>of</strong>ficial guidance are not enough for developments to meet contemporary<br />
aspirations, rhetoric, commitment or technical possibilities; matters <strong>of</strong> environmental justice are<br />
also involved as proper housing brings social and economic goods (Clark 2001). The domestic,<br />
commercial, and the public sectors are interdependent in bringing overall socio-economic<br />
development <strong>of</strong> the low-income groups. This raises the importance <strong>of</strong> domestic economics in<br />
sustainable development by way <strong>of</strong> affordable housing and environmental improvement, and<br />
supplements other areas by contributing to the human and labour development.<br />
The state is forced to tolerate some illegal and irregular housing (UNCHS 1996c; Gilbert 1990;<br />
Drakakis-Smith 1981); it denies better housing to maintain the status quo. While a few accept<br />
aesthetic <strong>of</strong> squatter settlements, the authorities loath such view (Peattie 1987), which have<br />
influenced housing policies and actions like the demolition <strong>of</strong> shelters and destruction <strong>of</strong><br />
communities. Though all communities have rights to improve their settlement, the legal,<br />
pr<strong>of</strong>essionals and participatory processes may not concur on squatter upgrading. 16<br />
Some politicians, planners, and intellectuals had been sympathetic to the idea that settlements <strong>of</strong> the<br />
poor were communities and deserved respect. For example, Jacobs (1961) articulated the functional<br />
aspects <strong>of</strong> the communities that planners and politicians liked to label slums; it was “slums <strong>of</strong><br />
hope” to Stokes (1963) and others. With the acceptance <strong>of</strong> in situ improvement and regeneration,<br />
self help can stand alongside modern technology in cultural–aesthetic form (Rahman & Mai 2010).<br />
Housing and environment <strong>of</strong> low-income groups can be improved by recognising <strong>of</strong> squatters<br />
housing rights, stable growth <strong>of</strong> income, and the development <strong>of</strong> social capital and empowerment.<br />
Thereon, social, ethical, and aesthetic expressions cover the full range <strong>of</strong> living, and encompass<br />
environmental, social, economic and political facets that encourage people to value lives. Allowing<br />
one to participate and express attachment in a varied socio-political context results in more<br />
commitments to affordable and sustainable improvements— incremental upgrading <strong>of</strong> low-quality<br />
makeshift shacks into wholesome structures.<br />
16 For example, social groups in Jordan strongly contended priorities and access to political and economic power (Raed<br />
1998). The government's attempts at urban renewal in Delhi by the late 1950s ran into resistance and had engendered a<br />
growing recognition that low-income communities had intrinsic value (Clinard 1966).<br />
64
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
It is essential to implement housing through participatory democracy (UKGP 1998). Local<br />
governments are the only elected representative making accountable planning and development<br />
decisions for the community; this is critical in moving toward sustainable communities (Roseland<br />
2000). Social homogeneity, community leadership, prior social co-operation experience, visible<br />
outcome, prospective ownership and the affordability can help to achieve consensus regarding<br />
development objectives and means (Rahman 1999). Environmental improvements can be converted<br />
into action plans and partnership by distributing responsibilities, attribution <strong>of</strong> costs and self help,<br />
and participatory and transparent management. In essence, both the process and the project need<br />
good governance, organisation, management, and policy (Pugh 2000).<br />
If legitimised and assisted, the large informal sector in most developing countries providing for the<br />
low-income groups can contribute in socio-economic developments (de Soto 1989; Fernandez &<br />
Varley 1998), in conserving economy, construction, environment, and health, and hence beckon for<br />
sustainable improvement. The regeneration schemes are sustainable as these improve living<br />
conditions, providing social opportunities for millions to add more socio-economic and<br />
environmental values than high-pr<strong>of</strong>iled projects. This potential varies with the characters <strong>of</strong> the<br />
squatter settlements. 17 Improvement can take place spontaneously where a form <strong>of</strong> tenure security is<br />
envisioned. Political skills and pressures <strong>of</strong>ten influence the selection <strong>of</strong> improvements and the<br />
distribution <strong>of</strong> costs and benefits among households. State-assisted regenerations <strong>of</strong>ten involve<br />
redesigning and re-aligning lay-outs (Potter & Lloyd-Evans 1998), which disrupts socio-economic<br />
network and identity, instead <strong>of</strong> preserving things <strong>of</strong> value to the target group that is possible only as<br />
part <strong>of</strong> overall housing development and urban macro-spatial planning and stockholders participation.<br />
Unprecedented urban growth in the face <strong>of</strong> increasing poverty and social inequality means that<br />
there will be about 2 billion people living in the slums in 2030 (UNCHS 2007); the MDG urged the<br />
international community to improve the lives <strong>of</strong> a part <strong>of</strong> them. Urbanisation <strong>of</strong> poverty (Whelan<br />
2004) is increasing number <strong>of</strong> underprivileged in urban areas. It is not possible to scale up housing<br />
production to meet the requirements <strong>of</strong> a billion people living in severe housing conditions through<br />
project based policies predominant in developing countries (Tipple 1994; UNCHS 1996a).<br />
International agencies want to replace them, which in themselves were not very successful, with<br />
sustainable housing policies that can work on a national level.<br />
Nobel economic laureates advocated state's welfare roles covering institutional reform, social and<br />
private property rights, and governance quality. They focused on the way the economic ethics and<br />
quality <strong>of</strong> institutions defined as norms, property rights, compliance procedures, etc. influenced<br />
performance in long-term growth. The World Bank prioritised institutional reform, at the heart <strong>of</strong><br />
modern policy interest in governance, in current urban development and policy agendas in the<br />
developing countries (Pugh 2000). Their policies were driven by the drive to balance the tensions<br />
between economic growth, the environment, and social impacts. The three-pronged approach<br />
conceptualised urban development that promoted economic growth, maintained social inclusion and<br />
minimised environmental impact. Most <strong>of</strong> the focus in the literature has been placed on economy and<br />
environment, though sustainable housing is more than just that. The economic sector addressing the<br />
financial aspects <strong>of</strong> social justice, accompanying the environmental sustainability, is an important<br />
17 In cases, the populations have expectations <strong>of</strong> imminent redevelopment, whereas in others de facto occupancy rights seem<br />
secure. Sometimes a settlement generates its own leadership and organizational structures which can be used for<br />
negotiating with politicians and bureaucracies for installing infrastructure. Other settlements either have apathy or<br />
powerlessness.<br />
65
MAHBUBUR RAHMAN<br />
element <strong>of</strong> housing, while the environmental limits are an important constraining influence on<br />
economic growth.<br />
Sustainability in housing, viewed as an end in itself, has no meaning. Robinson (2004) argued to<br />
integrate the social dimensions <strong>of</strong> sustainability with the biophysical dimensions; solutions<br />
addressing environmental, social or economic concerns only are insufficient. This required a transdisciplinary<br />
thinking that actively creates synergy, not just summation. Sustainability must also be<br />
integrated across sectors or interests; governments alone have neither the will nor the capability to<br />
accomplish it. Hence the private sector must be involved, supplemented by the monitoring, capacity<br />
building, and alternative service delivery roles by the NGOs.<br />
Sustainable development implies a shift in the capacity <strong>of</strong> individuals, companies and nations to<br />
use rightful resources under favourable legal and economic arrangements. Although conventional<br />
analyses recognise the need for changing the practices, few realise that moving toward a sustainable<br />
society requires more than minor adjustments. Sustainable community development initiatives are<br />
not only interventions, but also learning processes for mobilising positive changes. Changed social<br />
behaviour can prevent a host <strong>of</strong> environmental and related social disasters, to create healthy,<br />
sustainable more pleasant and satisfying communities than today. It emphasises on the efficient use<br />
<strong>of</strong> urban space, minimising the consumption <strong>of</strong> essential natural capital, multiplying social capital,<br />
and mobilising citizens and their governments toward these ends.<br />
Sustainability, “an attack on conventional thinking and practice” (Gibson 2001), and a framework<br />
for thinking about urban futures, provides an alternative with optimism. The global audience has<br />
pinned its hopes on sustainable development to solve the urgent environmental and societal<br />
problems (Roseland 2000). However, the right to affordable secure housing must be severed from<br />
property speculation and economic competition. Though, policies for sustainable housing in<br />
isolation may not overcome the urban problems, without them no solution can be found.<br />
6. REFERENCES<br />
[1.] Abbott, John (2002) An analysis <strong>of</strong> informal settlement upgrading and critique <strong>of</strong> existing<br />
methodological approaches. Habitat International Volume 26, Issue 3, September 2002, pp.<br />
303-315<br />
[2.] Abrams, C. (1964) Housing in the Modern World. Faber & Faber, London.<br />
[3.] ADB (1983) Regional Seminar on Financing Low-Income Housing- summary report. Asian<br />
Development Bank, Manila.<br />
[4.] Agyeman, J., & Evans, B. (Eds.) (1996) Statement <strong>of</strong> Purpose. Local Environment 1(1).<br />
[5.] Angel, S. & Benjamin, S.N. (1976) Seventeen Reasons why the Squatter Problem cannot be<br />
Solved. EKISTICS vol. 41, no. 242, pp. 20-26.<br />
[6.] Atkinson, G.A. (1961) Jobbing builders or Self-help for African housing. Journal <strong>of</strong> African<br />
Administration 13, pp. 46–49.<br />
[7.] Awaal, S. (1982) A Study <strong>of</strong> the Influence <strong>of</strong> the Overseas Workers Remittance on the Land<br />
Market <strong>of</strong> Dhaka. Unpublished Masters Thesis, Human Settlement Division, Asian Institute <strong>of</strong><br />
Technology, Bangkok.<br />
[8.] Back, K.W. (1962) Slums, projects, and people. Social psychological problems <strong>of</strong> relocation in<br />
Puerto Rico, Duke University Press, Durham, NC.<br />
66
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
[9.] Ballet, J., Dubois, J.L. & Mahieu, F.R. (2003) Le développement socialement durable: un<br />
moyen d'intégrer capacités et durabilité. Paper Presented at the Third Conference on the<br />
Capability Approach, University <strong>of</strong> Pavia, 6–9 September, 2003.<br />
[10.] Barbier, E (1988) The Economics <strong>of</strong> Environment and Development. Elgar, Cheltenham.<br />
[11.] Baross, P. & van der Linden, J. (Eds.) (1990) The Transformation <strong>of</strong> Land Supply Systems in<br />
Third World Cities. Avebury, Aldershot.<br />
[12.] Benjamin, S.N. & McCallum, D. (1985) Low-Income Housing in the Third World- broadening<br />
the economic perspective. Institute <strong>of</strong> Technology, Bandung.<br />
[13.] Choguill, C.L. (2007) The search for policies to support sustainable housing. Habitat Int. Vol.<br />
31, Issue 1, pp. 143-149<br />
[14.] Choguill, C.L. (1999) Sustainable Human Settlements: some second thoughts. In: A.F. Foo &<br />
B. Yuen (Eds.), Sustainable cities in the 21st century. The National University <strong>of</strong> Singapore,<br />
Singapore, pp. 131–142.<br />
[15.] Choguill, C. (1987) New Communities for Urban Squatters- lessons from the Plan that failed in<br />
Dhaka, Bangladesh. Plenum Press, NY.<br />
[16.] CIVIS (2003) The Enabling Environment for Housing Finance in Kenya: Shelter Finance for<br />
the Poor Series. Cities Alliance issue IV, April, 2003.<br />
[17.] Clark, M. (2001) Domestic futures and sustainable residential development. Futures Volume<br />
33, Issue 10, pp. 817-836.<br />
[18.] Clinard, M. (1966) Slums and community development,experiments in self-help. Free Press,<br />
NY.<br />
[19.] Comhabitat (2006) Briefing paper produced for the Commonwealth Civil Society Consultation.<br />
Marlborough House, London.<br />
[20.] Crane, J. L. (1950). How technical assistance in housing can help to achieve the purposes <strong>of</strong><br />
international cooperation. International Office, HHFA, Washington, DC, 28 December, 1950.<br />
[21.] Daly, H.E. (1992) Steady state economics (2nd ed). Earthscan, London.<br />
[22.] Davis, M. (2006) Planet <strong>of</strong> Slums. La Découverte, Paris.<br />
[23.] de Soto, H (1989) The Other Path: the invisible revolution in the Third World. Harper & Row,<br />
NY.<br />
[24.] Desai, V. (1996) Access to Power and Participation. Third World Planning Review 18 2, pp.<br />
217–242.<br />
[25.] Drakakis-Smith, D. (1981) Urbanization, housing and the development process. Croom Helm,<br />
London.<br />
[26.] Drèze, J. & A. Sen (1995) India's Economic Development and Social Opportunity. Clarendon,<br />
Oxford.<br />
[27.] Dubois, J.L. & Mahieu, F.R. (2002) La dimension sociale du développement durable: reduction<br />
de la pauvreté ou durabilité sociale? In: J.Y. Martin and L. Guillaume (Eds.), Développement<br />
Durable: Doctrines, Pratiques, Evaluations. IRD, Paris, pp. 73–94.<br />
[28.] El Sarafy, S. (1989) The proper calculation <strong>of</strong> income from depletable natural resources. In:<br />
Y.J. Ahmad, S. El Serafy and E. Lutz (Eds.), Environmental accounting for sustainable<br />
development. The World Bank, Washington, DC.<br />
[29.] Empacher, C. (2002) Die Sozialen Dimensionen der Nachhaltigkeit—Vorscläge zur<br />
Konkretisierung und Operationalisierung. Vortrag au der ordentlichen Mitgliederversammlung<br />
67
MAHBUBUR RAHMAN<br />
des Doktoranden-Netzwerk Nachhaltiges Wirtschaften, 26 April. Köln. Institut für sozialökologische<br />
Forschung. http://www.isoe.de/ftp/kerpen.pdf.<br />
[30.] European Commission (2001) A sustainable Europe for a better world- a European Union<br />
strategy for sustainable development. European Commission, Brussels.<br />
[31.] Fernandez, E. & Varley, A. (1998) Law, the city and citizenship in developing countries: an<br />
introduction. In: E. Fernandez & A. Varley (Eds.), Illegal Cities: Law and Urban Change in<br />
Developing Countries. Zed Books, London, pp. 3–17.<br />
[32.] Fogel, R. (1994) Economic growth, population theory, and psychology: the bearing <strong>of</strong> longterm<br />
processes on the making <strong>of</strong> public policy. American Economic Review June, pp. 369–395.<br />
[33.] Foy, G. & Daly, H.E. (1992) Allocation, distribution and scale as determinants <strong>of</strong><br />
environmental degradation: Case studies <strong>of</strong> Haiti, El Salvador and Costa Rica. In: A.<br />
Markandya & J. Richardson (Eds.), Environmental Economics. Earthscan, London.<br />
[34.] Frankenh<strong>of</strong>f, C. (1967) Elements <strong>of</strong> an economic model for slums in a developing country.<br />
Economic Development and Cultural Change 16, pp. 27–36.<br />
[35.] Frieden, B.J. (1965) Search for housing policy in Mexico City. Town Planning Review 36<br />
(1965), pp. 75–94.<br />
[36.] Forster-Kraus, S., Reed, R. & Wilkinson, S. (2009) Affordable Housing in the Context <strong>of</strong><br />
Social Sustainability. ISA International Housing Conference, 1- 4 September 2009, Glasgow.<br />
[37.] Gerrul, B. (1979) Economic Effects <strong>of</strong> Housing in Developing Countries; Vol. III & IV- types<br />
<strong>of</strong> low-income residential areas in Dhaka. Technical University <strong>of</strong> Darmstadt, Darmstadt.<br />
[38.] Gibson, Robert (2001) Specification <strong>of</strong> sustainability-based environmental assessment decision<br />
criteria and implications for determining significance in environmental assessment. Available<br />
from: .<br />
[39.] Gilbert, A. (1990) The Costs and Benefits <strong>of</strong> Illegality and Irregularity in the Supply <strong>of</strong> Land.<br />
In: P. Baross, P. & J. van der Linden (Eds.), The Transformation <strong>of</strong> Land Supply Systems in<br />
Third World Cities. Avebury, Aldershot, pp. 17–36.<br />
[40.] Guha R. & Martinez-Alier, J. (1997) Varieties <strong>of</strong> Environmentalism: Essays North and South.<br />
Earthscan, London.<br />
[41.] Greider, W. (1997) One World, Ready or Not: The Manic Logic <strong>of</strong> Global Capitalism. Simon,<br />
New York.<br />
[42.] Grimes, O.F. (1976) Housing for Low-Income Urban Families. John Hopkins University Press,<br />
Washington, DC.<br />
[43.] Hamdi, N. & Goethert, R. (1996) Action planning for cities, a guide to community practice.<br />
Wiley, Chichester, NY.<br />
[44.] Hardoy, J.E., Miltin, D. & Satterthwaite, D. (1992) Environmental Problems in Third World<br />
Cities. Earthscan, London.<br />
[45.] Harris, Richard (2003) Learning from the past: international housing policy since 1945. A<br />
double irony: the originality and influence <strong>of</strong> John F.C. Turner. Habitat International Vol. 27,<br />
Issue 2, June 2003, pp. 245-269<br />
[46.] Holden, Meg (2006) Urban indicators and the integrative ideals <strong>of</strong> cities, Cities Volume 23,<br />
Issue 3, June 2006, pp. 170-183<br />
[47.] Huchzermeyer, M. (1999) The exploration <strong>of</strong> appropriate informal settlement intervention in<br />
South Africa: Contributions from a comparison with Brazil. unpublished Ph.D. thesis,<br />
University <strong>of</strong> Cape Town, South Africa.<br />
68
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
[48.] Igel, B. & Srinivas, H. (1996) The Co-Option <strong>of</strong> Low-Income Borrowers by Informal Credit<br />
Suppliers. Third World Planning Review 18 3, pp. 287–305.<br />
[49.] Israel, A. (1990) The Changing Role <strong>of</strong> the State: Institutional Dimensions. World Bank Policy,<br />
Research and External Affairs Working Paper # 495, Washington, DC.<br />
[50.] Jacobs, J. (1961) The Death and Life <strong>of</strong> Great American Cities. Vintage, NY.<br />
[51.] Jacobson, M. (May 2007). Dharavi: Mumbai's Shadow City. National Geographic, National<br />
Geographic Society. http://ngm.nationalgeographic.com/2007/05/dharavi-mumbaislum/jacobson-text;<br />
retrieved on 2009-04-28.<br />
[52.] Keare, D.H. & Jimenez, E. (1983) Progressive Development and Affordability in the Design <strong>of</strong><br />
Urban Shelter Projects. Working Paper # 560, The World Bank, Washington DC.<br />
[53.] Keivani, Ramin & Werna, Edmundo (2001) Modes <strong>of</strong> Housing Provision in Developing<br />
Countries. Progress in Planning Vol. 55, Issue 2, February 2001, pp. 65-118<br />
[54.] Kellett, P. & Napier, M. (1995) Squatter architecture? A critical examination <strong>of</strong> vernacular<br />
theory and spontaneous settlement with reference to South America and South Africa.<br />
Traditional Dwellings and Settlements Review 6 2, pp. 7–24<br />
[55.] Kessides, C. (1997) World Bank experience with the provision <strong>of</strong> infrastructure services for the<br />
urban poor: preliminary identification and review <strong>of</strong> best practices. The World Bank, TWU-<br />
OR8, Washington, DC.<br />
[56.] Kimm, P.M. (1987) Housing Progress in Developing Countries, in Proceedings <strong>of</strong> the Second<br />
International Shelter Conference and Vienna Recommendations on Shelter and Urban<br />
Development. National Association <strong>of</strong> Realtors, Washington, DC, pp. 72–75.<br />
[57.] LaNier, R., Oman, C.A. & Reeve, S. (1987) Encouraging Private Initiative. USAID,<br />
Washington, DC.<br />
[58.] Lankatilleke, L. (1990) Community Action Planning: A case study from Sri Lanka. Trialog<br />
23/24, pp. 24–27.<br />
[59.] Lawrence, R. (1997) Space and Text. In: R. Lawrence & A. Wallace-Hadrill (Eds.), Domestic<br />
Space in the Roman World: Pompeii and Beyond. Journal <strong>of</strong> Roman Archaeology<br />
Supplementary Series No. 22, pp. 7–14.<br />
[60.] Lehtonen, Markku (2004) The Environmental–Social Interface <strong>of</strong> Sustainable Development:<br />
capabilities, social capital, institutions. Ecological Economics Volume 49, Issue 2, 1 June 2004,<br />
pp. 199-214.<br />
[61.] Ling, R. (1997) The Insula <strong>of</strong> the Meander at Pompeii, Vol. 1. The Structure Clarendon,<br />
Oxford.<br />
[62.] Loh, P. (1987) Policy and Programs for Urban and Shelter Lending: The Next Decade. In:<br />
Proceedings <strong>of</strong> the Second International Shelter Conference and Vienna Recommendations on<br />
Shelter and Urban Development. National Association <strong>of</strong> Realtors, Washington, DC, pp. 67–<br />
71.<br />
[63.] Maliene, Vida & Malys, Naglis (2009) High Quality Housing- a key issue in delivering<br />
sustainable communities. Building & Environment Vol. 44, Issue 2, pp. 426-430.<br />
[64.] Mangin, W. (1967) Latin American Squatter Settlements: a problem and a solution. Latin<br />
American Research Review 2, pp. 65–98.<br />
[65.] Marcus, S. (1995) House as a Mirror <strong>of</strong> Self: explaining the deeper meanings <strong>of</strong> home, Conari<br />
Press, Berkeley.<br />
69
MAHBUBUR RAHMAN<br />
[66.] Mitlin, D. and Satterthwaite, D. (1992) Scaling-Up in Urban Areas. In: M. Edwards and D.<br />
Hulme (Eds.), Making a Difference: NGOs and Development in a Changing World. Earthscan,<br />
London, pp. 169–179.<br />
[67.] Nientied, P. & van der Linden, J. (1985) Approaches to low-income housing in the Third<br />
World: some comments. International Journal <strong>of</strong> Urban and Regional Research 9, pp. 311–329.<br />
[68.] NIHFW (2006) Slum population in million plus cities. Slums in India - an Overview. Urban<br />
Health Resource Unit, National Institute <strong>of</strong> Health and Family Welfare, Ministry <strong>of</strong> Health and<br />
Family Welfare, Government <strong>of</strong> India, New Delhi.<br />
[69.] NTFEE (1987) Report <strong>of</strong> the National Task Force on Environment and Economy. Canadian<br />
Council <strong>of</strong> Resource and Environment Ministers, Ottawa.<br />
[70.] Norgaard, R.B. (1994) Development Betrayed: The End <strong>of</strong> Progress and a Coevolutionary<br />
Revisioning <strong>of</strong> the Future. Routledge, London.<br />
[71.] Norton, Bryan (1999) Pragmatism, Adaptive Management, and Sustainability. Environmental<br />
Values 8, pp. 455-466.<br />
[72.] Passet, R. (1996) L'Economique et le vivant, 2 édition. Payot, Paris.<br />
[73.] Payne, G. (1983) Low-Income Housing in the Developing World- the role <strong>of</strong> sites-and-services<br />
and squatter upgrading. John Wiley & Sons, NY.<br />
[74.] Peattie, L.R. (1987) Shelter development and the poor. In: L. Rodwin (Ed.), Shelter, Settlement<br />
and Development. Allen & Unwin, Boston, pp. 263–280.<br />
[75.] Peattie, L. & Doebele, W.A. (1973). Review <strong>of</strong> J.F.C. Turner & Robert Fichter (Eds.), Freedom<br />
to Build. Journal American Institute <strong>of</strong> Planners, 39, pp. 66–67.<br />
[76.] Pezzoli, K. (1997) Sustainable development: a transdisciplinary overview <strong>of</strong> the literature.<br />
Journal <strong>of</strong> Environmental Planning & Management 40 (5), pp. 549–574.<br />
[77.] Potter, R. & Lloyd-Evans, S. (1998) The City in the Developing World. Longman, Harlow.<br />
[78.] Pugh, C. (1990) Housing and Urbanization: A Study <strong>of</strong> India. Sage, New Delhi.<br />
[79.] Pugh, C. (1994) The development <strong>of</strong> housing finance and the global strategy for shelter. Cities<br />
11, pp. 384–392.<br />
[80.] Pugh, C. (1995) The role <strong>of</strong> the world bank in housing. In: B. Aldrich and R. Sandhu (Eds.),<br />
Housing the urban poor: Policy and practice in developing countries, Zed Books, London.<br />
[81.] Pugh, C. (1997) Poverty and progress? Reflections on housing and urban policies in developing<br />
countries, 1976–96. Urban Studies 34 10, pp. 1547–1596<br />
[82.] Pugh, C. (2000) Squatter Settlements: their sustainability, architectural contributions, and socioeconomic<br />
roles. Cities Vol. 17, Issue 5, October 2000, pp. 325-337<br />
[83.] Raed, H. (1998) Public participation and community organisation in the low-income housing<br />
production: the Jordanian experience. Unpublished Ph.D. thesis, University College London.<br />
[84.] Rahman, M.M. (1990) Seventeen Facts on Dhaka’s Bastees. EKISTICS, Vol. 57, No. 342-343,<br />
May-August, pp. 172-180.<br />
[85.] Rahman, M.M. (1999) Role <strong>of</strong> the NGOs in Urban Housing for the Poor. Robert McNamara<br />
Fellowship Report; World Bank, Washington DC.<br />
[86.] Rahman, M.M. (2004) Housing Affordability- ability or willingness? Symposium <strong>of</strong> Housing<br />
II- affordable dwelling, March 27-30, High Commission for the Development <strong>of</strong> Arriyadh,<br />
KSA.<br />
70
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
[87.] Rahman, M.M. & Mokhtar Mai (2010) World Bank’s changing approach to Sustainability <strong>of</strong><br />
Housing Affordability in Developing Economis. In: MHA Samad et.al (Eds.), Issues on Global<br />
Energy Crisis and its Impact on Design, USM, Penang, pp. 118-135.<br />
[88.] Rapoport, A. (1988) Spontaneous Settlements as Vernacular Design. in Patton C. (ed)<br />
Spontaneous Shelter: International Perspectives and Prospects, Temple University Press,<br />
Philadelphia, pp. 51–57.<br />
[89.] Rees, W. (1996) Revisiting carrying capacity: Area-based indicators <strong>of</strong> sustainability,<br />
Population and Environment 17 (3).<br />
[90.] Robinson, John (2004) Squaring the circle? Some thoughts on the idea <strong>of</strong> sustainable<br />
development. Ecological Economics Volume 48, Issue 4, 20 April 2004, pp. 369-384<br />
[91.] Robinson, J. and Tinker, J. (1997) Reconciling ecological, economic, and social imperatives: a<br />
new conceptual framework. In: T. Schrecker (Ed.), Surviving Globalism: Social and<br />
Environmental Dimensions, Macmillan, London, pp. 71–94.<br />
[92.] Roseland, Mark (2000) Sustainable Community Development: integrating environmental,<br />
economic, and social objectives. Progress in Planning, Vol. 54 Issue 2, pp. 73-132.<br />
[93.] Safa, H.I. (1964) From shanty town to public housing. A comparison <strong>of</strong> family structure in two<br />
urban neighbourhoods in Puerto Rico. Caribbean Studies 4 (1964), pp. 3–12.<br />
[94.] Saleem, S. (1983) Land Management in the katchi abadis <strong>of</strong> Karachi. In: S. Angel, R.W.<br />
Archer, S. Tanphiphat and E.A. Weglin (Eds.), Land for Housing the Poor, Select Books,<br />
Singapore, pp. 144–155.<br />
[95.] Sen, A.K. (1999) Development as Freedom. Anchor Books, NY.<br />
[96.] Sen, A.K (1987) Commodities and Capabilities, Oxford India Paperbacks, Oxford Univ. Press,<br />
Oxford.<br />
[97.] Sen, A.K. & Wolfensohn, J. (1999) Development: a coin with two sides, World Bank,<br />
Washington DC.<br />
[98.] Setshedi, Gift Phalatse (2006) Architecture, Literature and the Advancement <strong>of</strong> Township<br />
Enterprises. In: M. Rahman (Ed.), Society, Architects and Emerging Issues, CAA-IAB,<br />
London-Dhaka, pp. 85-92.<br />
[99.] Shakur, M.T. (1987) Approaches towards Housing Low Income Communities in the Third<br />
World: a Literature Review. University <strong>of</strong> Liverpool, Working Paper, p. 30.<br />
[100.] Silas, J. (1983) Spatial structure housing delivery, land tenure and the urban poor in Surabaya,<br />
Indonesia. In: S. Angel, S., R.W. Archer, S. Tanphiphat and E.A. Weglin (Eds.), Land for<br />
Housing the Poor, Select Books, Singapore, pp. 211–233.<br />
[101.] Skinner, R., Taylor J. and Weglin E.A. (Eds.) (1987) Shelter Upgrading for the Urban Poor:<br />
evaluation <strong>of</strong> Third World experience, Island Publishing House, Manila.<br />
[102.] Smets, P. (1999) Housing finance trapped in a dilemma <strong>of</strong> perceptions: affordability criteria for<br />
the urban poor in India question. Housing Studies 14, pp. 821–838.<br />
[103.] Stiglitz, J. (1998) Towards a new paradigm for development: strategies, policies and processes.<br />
9th Raul Prebisch Lecture, UN Conference on Trade and Development, Geneva.<br />
[104.] Stokes, C. (1963) A theory <strong>of</strong> slums. Land Economics 38 (1963), pp. 187–197.<br />
[105.] Stren, R., and Polese, M. (2000) Understanding the new socio-cultural dynamics <strong>of</strong> cities:<br />
comparative urban policy in a global context. In: M. Polese and R. Stren (Eds.), The Social<br />
Sustainability <strong>of</strong> Cities, University or Toronto Press, Toronto, pp. 3–38.<br />
71
MAHBUBUR RAHMAN<br />
[106.] Stretton, H. (1976) Capitalism, Socialism and the Environment, Cambridge University Press,<br />
London.<br />
[107.] Swan, P.J. et al., (1983) Management <strong>of</strong> sites and services housing schemes: The Asian<br />
experience, Wiley, Chichester (1983).<br />
[108.] Tipple, G.A. (1994) A Matter <strong>of</strong> Interface: the Need for a Shift in Targeting Housing<br />
Intervention. Habitat International 18 4, pp. 1–15.<br />
[109.] Trainer, T. (1990) A Rejection <strong>of</strong> the Brundtland Report. ifda dossier 77, 71–84, May/June<br />
1990.<br />
[110.] Turner, A. (1980) The Cities <strong>of</strong> the Poor. Croom Helm, London.<br />
[111.] Turner, J.F.C. (1976a) Housing by People- towards autonomy in building environments.<br />
Marion Boyars, London.<br />
[112.] Turner, J.F.C. (1976b) Approaches to government-sponsored housing. Ekistics 41 (1976), pp.<br />
4–7.<br />
[113.] Turner, J.F.C. (1972) The Re-Education <strong>of</strong> a Pr<strong>of</strong>essional, and Housing as a verb. In: J.F.C.<br />
Turner and R. Fichter (Eds.), Freedom to Build, Macmillan, NY.<br />
[114.] Turner, J.F.C. (1968) Uncontrolled urban settlements. Problems and policies. International<br />
Social Development Review 1, pp. 107–130.<br />
[115.] Turner, J.F.C. (1967) Barriers and channels for housing development in modernising countries.<br />
Journal <strong>of</strong> the American Institute <strong>of</strong> Planners 33 3, pp. 167–181.<br />
[116.] UK Green Party (1998) Manifesto, policy HO101,<br />
http://www.brighton.co.uk/greenparty/housing.htm, accessed on 10.01.09.<br />
[117.] UNCHS-Habitat (2007) UN-Habitat Sustainable Urbanization: local action for urban poverty<br />
reduction, emphasis on finance and planning. 21st Session <strong>of</strong> the Governing Council, Nairobi.<br />
[118.] UNCHS (2006) State <strong>of</strong> the World's Cities Report 2006/7, Third World Urban Forum,<br />
Vancouver.<br />
[119.] UNCHS (2003) The Challenge <strong>of</strong> Slums: new Global Report on Human Settlements, UN-<br />
Habitat, Nairobi.<br />
[120.] UNCHS-Habitat (1996a) The Istanbul Declaration and Habitat II Agenda, Istanbul.<br />
[121.] UNCHS-Habitat (1996b) The Future <strong>of</strong> Human Settlements: Good Policy Can Make a<br />
Difference, Istanbul, A/CONF.165/7.<br />
[122.] UNCHS-Habitat (1996c) An Urbanizing World: Global Report on Human Settlements 1996,<br />
Oxford University Press, Oxford.<br />
[123.] UN Dept. <strong>of</strong> Economic and Social Affairs (1957) Financing <strong>of</strong> housing and community<br />
improvement programmes, UN (ST/SOA/32), NY.<br />
[124.] UNFPA (2007) State <strong>of</strong> World Population 2007, United Nations Population Fund, NY.<br />
[125.] Unwin, S. (1997) Analysing Architecture, Routledge, London.<br />
[126.] Ward, P.M. (1984a) Self-Help Housing- a critique, Mansell, London.<br />
[127.] Ward, P.M. (1984b) Mexico- beyond Sites and Services. In: G. Payne (Ed.), Low-Income<br />
Housing in the Developing World, John Wiley and Sons, NY, pp 149-159.<br />
[128.] Watts, K. (1997) Outwards from home- A planner's odyssey, The Book Guild, Sussex.<br />
[129.] Weissman, E. (1960) Mutual aid in low-cost housing. Annals <strong>of</strong> the American Academy <strong>of</strong><br />
Political and Social Science 329 (1960), pp. 107–114.<br />
72
AFFORDABILITY, ENABLEMENT AND SUSTAINABILITY OF LOW-INCOME HOUSING<br />
[130.] Whelan, S. (2004) One third <strong>of</strong> the world’s urban population lives in a slum. International<br />
Committee <strong>of</strong> the Fourth International, www.wsws.org<br />
[131.] Wolfensohn, J. (1999) A proposal for a comprehensive development framework. Speech to the<br />
Board, Management and Staff <strong>of</strong> the World Bank Group, World Bank, Washington, DC.<br />
[132.] Woolcock, M. (2001) The place <strong>of</strong> social capital in understanding social and economic<br />
outcomes. The contribution <strong>of</strong> Human and Social Capital to Sustained Economic Growth and<br />
Well-Being. International Symposium Report. Human Resources Development Canada and<br />
OECD, pp. 65–88.<br />
[133.] World Bank (1999) A Strategic View <strong>of</strong> Urban and Local Government Issues: Implications for<br />
the Bank. World Bank, Washington, DC.<br />
[134.] World Bank (1995) Indonesia impact evaluation report. Enhancing the quality <strong>of</strong> life in urban<br />
Indonesia: the legacy <strong>of</strong> the Kampung improvement program. Operations Evaluation<br />
Department, Report # 14747-IND, World Bank, Washington, DC.<br />
[135.] World Bank (1994) Twenty years <strong>of</strong> lending for urban development, 1972–92, World Bank<br />
Operations Evaluation Department, Report # 13117, Washington, DC.<br />
[136.] World Bank (1993) Housing: Enabling Markets to Work, World Bank, Washington, DC.<br />
[137.] World Bank (1991) Urban policy and economic development: An agenda for the 1990s, World<br />
Bank, Washington DC.<br />
[138.] World Bank (1983) Learning by Doing, World Bank, Washington, DC.<br />
[139.] World Bank (1973) Housing Enabling markets to work, World Bank, Washington, DC.<br />
[140.] World Bank (1972) Urbanization, World Bank, Washington, DC.<br />
[141.] World Commission on Environment & Development (1987) Our common future, Oxford<br />
University Press, Oxford.<br />
73
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 75-85 (2010)<br />
© 2009 ALHOSN University<br />
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061<br />
ALUMINUM USING DESIRABILITY FUNCTIONS<br />
Mahmoud I. Awad 1* , Elsayed Orady 2 , Chenmaya Dandekar 2<br />
1<br />
Industrial Engineering Department, ALHOSN University, P.O. Box: 38722, Abu Dhabi, UAE<br />
2 Industrial and Manufacturing Systems Engineering Department, University <strong>of</strong> Michigan-Dearborn, Dearborn, MI. USA<br />
ABSTRACT: This paper investigates the effect <strong>of</strong> drilling process parameters, such as cutting speed, feed, and peck<br />
angle on the drilled hole quality characterized by surface roughness and burrs’ formation. Empirical models<br />
describing process performance which can be used for prediction are presented. The models for average<br />
performance and variability were used to optimize the roughness and burr using desirability functions.<br />
Keywords: Drilling process, Robustness, Desirability functions<br />
1. INTRODUCTION:<br />
Due to its good mechanical properties such as machinability and low density, Aluminum<br />
is commonly used in a wide range <strong>of</strong> industries and constitutes about 40% <strong>of</strong> all metal-cutting<br />
operations (Hamade and Ismail (2005)). Drilling is a cutting process in which a hole is originated<br />
or enlarged by means <strong>of</strong> a multipoint, fluted, end cutting tool typically aided by cutting fluids.<br />
As the drill is rotated and advanced into the workpiece, material is removed in the form <strong>of</strong> chips<br />
that move along the fluted shank <strong>of</strong> the drill. Chips are produced within the workpiece and move<br />
in direction opposite to axial movement <strong>of</strong> the drill. Although long spiral chips usually result<br />
from drilling, adjustment <strong>of</strong> the feed rate can result in chips with a range <strong>of</strong> shapes and sizes.<br />
Consequently, chip disposal in drilling and the effectiveness <strong>of</strong> cutting fluids are important.<br />
The most two important performance measures used to determine drilling quality are surface<br />
roughness and burr height. Two surface roughness measurements are very common: R<br />
a<br />
and R<br />
q<br />
where R<br />
a<br />
is the arithmetic mean <strong>of</strong> the departures <strong>of</strong> the roughness pr<strong>of</strong>ile from the mean line,<br />
while R<br />
q<br />
measures the maximum departure (See Figure 1). R<br />
a<br />
and R<br />
q<br />
are highly correlated, so<br />
minimizing one <strong>of</strong> them will lead to minimizing the other. R<br />
a<br />
is expressed mathematically as<br />
where l m<br />
denotes cut-<strong>of</strong>f length.<br />
R<br />
l m<br />
1<br />
R<br />
a<br />
f ( y)<br />
dx,<br />
(1)<br />
l<br />
q<br />
q<br />
m<br />
0<br />
R is the RMS parameter corresponding to<br />
<br />
1<br />
l<br />
m<br />
l m<br />
<br />
0<br />
2<br />
f<br />
( y)<br />
dx,<br />
R<br />
a<br />
by<br />
(2)<br />
____________________________________<br />
* Corresponding Author. Tel.: +971 4070585<br />
E-mail : m.awad@<strong>alhosn</strong>u.ae<br />
75
MAHMOUD I. AWAD, ELSAYED ORADY, CHENMAYA DANDEKAR<br />
Figure 1: Surface roughness<br />
Burr is plastically deformed projected material, generated on the part edge during drilling.<br />
Both roughness and burr cause reliability problems and performance degradation especially in precise<br />
assemblies. Burr formation requires additional manufacturing operations with additional cost and<br />
time for disassembly and de-burring. The geometry and drilling process parameters <strong>of</strong> classical<br />
drilling have been well studied over the years and gained more importance lately. For example Pande<br />
and Relekar (1986) have found that feed rate is the most significant factor in the drilling process.<br />
Lauderbaugh (2009) used experimentally verified simulation to predict burr height, force and<br />
temperature for 2024-T351 and 7075-T6 aluminum. He concluded that feed rate, chisel edge to drill<br />
diameter ratio, drill diameter, yield strength and point angle are the most significant parameters<br />
affecting burr height. Pena et al. (2005) proposed a monitoring method, based on internal signals from<br />
spindle torque, to detect non-desired burr formation during drilling operations.<br />
Other researchers investigated tool wear as a function <strong>of</strong> cutting speed and coating material,<br />
see Nouari et al. (2005) and List et al. (2005).<br />
Most <strong>of</strong> the <strong>engineering</strong> applications, including drilling process, would comprise <strong>of</strong> multiple<br />
responses such as surface roughness and burr height. Moreover, variability <strong>of</strong> process or product has<br />
to be minimized and mean has to be optimized. Optimizing multiple responses simultaneously is a<br />
difficult task that has been tackled by many researchers. One <strong>of</strong> the early methods proposed by<br />
Taguchi (1987) used for multi-response optimization advocates the use <strong>of</strong> signal-to-noise ratios<br />
(S/N). This methodology specifically utilizes both experimentation and optimization methods to<br />
determine the system optimum operating conditions to minimize performance variability and<br />
deviation from target value <strong>of</strong> interest. Many detractors such as Box (1988) and Nair (1992)<br />
challenged S/N ratio methodology and question Taguchi choice <strong>of</strong> experimental designs and methods<br />
<strong>of</strong> statistical analysis. Khuri and Conlon (1981) used the generalized distance approach to find the<br />
optimal settings that minimize the distance function over the experimental region. Pignatielo (1993)<br />
used a quadratic loss function to solve the multi-response problem. In this article, a desirability<br />
function method (Deringer & Suich, 1980; Harington, 1965), which transforms the multi-response<br />
problem into a single response problem by maximizing the combined desirability, is used. More<br />
details about desirability method are provided in section 2.2.<br />
In the next section, we review background information on experimental design and optimization<br />
model using desirability function. In section 3, we utilize the above methods in optimizing the<br />
drilling process relative to surface roughness and burr height and its variability’s. Finally, we<br />
conclude the paper with a discussion <strong>of</strong> our findings.<br />
2. BACKGROUND: EXPERIMENTATION AND DESIRABILITY OPTIMIZATION<br />
2.1. Experimental designs<br />
Experimental design is a scientific approach to planning an experiment using statistical<br />
techniques (Montgomery, 1993). The process <strong>of</strong> planning an experiment is performed so that the<br />
76
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061 ALUMINUM USING DESIRABILITY FUNCTIONS<br />
experimental data collected can be analyzed in order to obtain valid and objective conclusions. In<br />
general, experimental data are subject to experimental errors and statistical approach is the only<br />
objective approach to analyze such data. Three basic principles <strong>of</strong> experimental design that must<br />
be emphasized when designing an experiment are the principles <strong>of</strong> replication, randomization,<br />
and blocking.<br />
Replication is the repetition <strong>of</strong> an experiment. It allows the experimenter to estimate the<br />
experimental error, which is used as a basis for determining whether observed differences in<br />
the data are really statistical differences. In addition, experimenter is usually interested in<br />
estimating the effect <strong>of</strong> different levels <strong>of</strong> experimental factors. More importantly, replication<br />
enables the experimenter to estimate and model variability in terms <strong>of</strong> standard deviation or<br />
variance.<br />
Randomization is an important concept underlying the use <strong>of</strong> statistical methods in<br />
experimental design. It should be <strong>applied</strong> to two aspects <strong>of</strong> the experiment, which are 1) to<br />
randomly allocate the experimental units (e.g. workpiece, material, etc.) to different<br />
experimental settings, and 2) to randomly determine the order in which the individual runs<br />
are to be performed. Statistical methods usually require that the experimental errors be<br />
independently distributed random variables. Randomization helps assuring that this<br />
assumption is met. Furthermore, properly applying randomization would balance out the<br />
effect <strong>of</strong> extraneous factors that may be present.<br />
Blocking is a technique used to increase the precision <strong>of</strong> an experiment’s result. A block is a<br />
portion <strong>of</strong> the experimental units that is more homogenous than the entire set <strong>of</strong> experimental<br />
units. Blocking involves making comparisons among the conditions <strong>of</strong> interest in the<br />
experiment within each block.<br />
2.2 Desirability Models<br />
Derringer and Suich (1980) developed a procedure which can be very useful when several<br />
responses are involved. The method utilizes a desirability function to be maximized in which<br />
practitioners can assign their own priorities and desires on the response values and build it into<br />
the optimization procedure. The method makes use <strong>of</strong> an objective function, D, called the<br />
desirability function and transforms an estimated response into a scale free value (d i ), called<br />
desirability as shown in 3.<br />
Where n is the number <strong>of</strong> responses, and w i is importance <strong>of</strong> the response. The desirable scale<br />
ranges from zero (least desirable) to one (most desirable) and have different forms depending on<br />
the goal <strong>of</strong> responses as shown in Equations 4-6:<br />
(4)<br />
77
MAHMOUD I. AWAD, ELSAYED ORADY, CHENMAYA DANDEKAR<br />
(5)<br />
(6)<br />
Where wt i is a weight factor range from 0.1 to 10 and controls the shape <strong>of</strong> the desirability<br />
function. For example, if wt i =1, desirable scale d i will vary from 0 to 1 in a linear fashion. In this<br />
paper, and without any general loss, the weight factor will be assumed one for all responses.<br />
Taking into account all n responses, clearly one wishes to choose the conditions x for the process<br />
parameters to maximize D. A value <strong>of</strong> D close to 1.0 implies that all responses are in a desirable<br />
range simultaneously.<br />
3. DRILLING EXPERIMENT<br />
Machining processes, in general, have many parameters that significantly affect the<br />
performance <strong>of</strong> the processes. The influences could be on the machine performance, the tool used<br />
to perform the processes, and/or the products produced. In any machining process, like any<br />
production machine, the ultimate goal is to produce high quality products with minimal cost.<br />
The primary objective <strong>of</strong> the drilling experiment conducted in this paper is to establish a<br />
relationship between the quality <strong>of</strong> the machined holes and machining variables. This is because,<br />
to the best <strong>of</strong> the authors’ knowledge, there is no relationship (i.e. empirical or physics models)<br />
that has been established for predicting the quality <strong>of</strong> drilled holes by using the machining<br />
parameters.<br />
In general, machining processes have numerous parameters. Some parameters related to the<br />
machine such as machine dynamics and accuracy, other parameters related to the tool such as the<br />
ability <strong>of</strong> the tool to perform a cut with certain cutting conditions, and finally parameters related<br />
to the workpieces such as the mechanical and physical properties <strong>of</strong> the workpiece materials, and<br />
the geometry <strong>of</strong> the workpiece. For the drilling process in particular, there are many methods<br />
used to produce a hole. The first step is to machine a center for the hole using combination center<br />
tool. The depth <strong>of</strong> the center is a function <strong>of</strong> the diameter <strong>of</strong> the hole. Depending on the diameter<br />
and length <strong>of</strong> the hole, the procedure to produce the hole varies. The diameter <strong>of</strong> the hole could<br />
be machined using a drill with the same diameter, or the hole could be machined in two or more<br />
78
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061 ALUMINUM USING DESIRABILITY FUNCTIONS<br />
drilling sequences by using drills with different sizes to reach the desired diameter <strong>of</strong> the hole.<br />
On the other hand, depending on the depth <strong>of</strong> the hole, it could be drilled in one sequence or one<br />
tool feed, or using peck procedure.<br />
3.1 Experimentation Design<br />
In order to limit the number <strong>of</strong> experimental factors to a manageable number, three cutting<br />
parameters were selected as controllable experimental factors. These include cutting speed, feed,<br />
and peck. Regarding the range <strong>of</strong> the input variables, the cutting speed and feed for the selected<br />
tool and workpiece pair materials were determined using the Machinery’s Handbook.<br />
Several experimental designs were considered such as full factorial, fractional, Taguchi<br />
orthogonal arrays, and regression models. Since our prime goal is to construct an empirical<br />
model, regression designed experiments with number <strong>of</strong> levels equal to number <strong>of</strong> runs were<br />
chosen. The other designs are rigid in terms <strong>of</strong> number <strong>of</strong> levels. More levels tested empower the<br />
model in capturing any non-linearity if exist.<br />
Six <strong>of</strong> 6"<br />
4" 1"<br />
workpieces were cut from a 6-foot long 6061 Aluminum bar. Each workpiece<br />
was drilled 15 times. Since the objective <strong>of</strong> the experiment is to construct an empirical prediction<br />
model, each drilling parameters was not treated as fixed effect factors, but was treated as<br />
continuous variables. The drilling parameters (speed, feed and peck length) were randomly<br />
generated from its input range using the Random function in Excel “rand()” to drill 15 holes in<br />
each workpiece. Two blocks <strong>of</strong> parameter values for the drilling process were randomly<br />
generated as shown in Table 1.<br />
Table 1: Two blocks <strong>of</strong> parameter values for the drilling process<br />
Standard Order Block Cutting speed (S)<br />
(fpm)<br />
Feed (f)<br />
(ipr)<br />
Peck depth (P)<br />
(in.)<br />
1 1 364 0.007 0.75<br />
2 1 329 0.004 0.086<br />
3 1 252 0.003 0.066<br />
4 1 308 0.005 0.1<br />
5 1 208 0.008 0.1<br />
6 1 287 0.006 0.4<br />
7 1 275 0.008 0.79<br />
8 1 377 0.007 0.55<br />
9 1 366 0.001 0.64<br />
10 1 253 0.004 0.07<br />
11 1 364 0.006 0.33<br />
12 1 361 0.010 0.85<br />
13 1 315 0.006 0.5<br />
14 1 341 0.009 0.9<br />
15 1 225 0.001 0.05<br />
16 2 278 0.005 0.138<br />
17 2 214 0.005 0.65<br />
18 2 273 0.008 0.53<br />
19 2 216 0.003 0.72<br />
20 2 336 0.006 0.16<br />
21 2 204 0.005 0<br />
22 2 391 0.002 0.35<br />
79
MAHMOUD I. AWAD, ELSAYED ORADY, CHENMAYA DANDEKAR<br />
23 2 358 0.007 0.53<br />
24 2 370 0.003 0.6<br />
25 2 208 0.0075 0.23<br />
26 2 331 0.009 0.1<br />
27 2 305 0.0052 0.38<br />
28 2 274 0.0068 0.1<br />
29 2 239 0.0066 0.67<br />
30 2 294 0.003 0.5<br />
The Six workpieces were randomly divided into two sets <strong>of</strong> three workpieces, where each set<br />
received a random selection <strong>of</strong> a parameter set, and the three workpieces in each set served as<br />
replications for each parameter set. Two CNC programs were developed and loaded in the<br />
machine (Fadal TRM Vertical milling machine) to perform the drilling process using six-half<br />
inch Coated-HSS twist drills.<br />
3.2 Experimentation Results<br />
The response variables are measured using a Mitutoyo Surftest 402 and a Surftest Analyzer.<br />
The surface roughness measurements Ra<br />
were obtained using proper scale and resolution. The<br />
burr heights were measured using a dial indicator with 0.001 inch resolution. For each hole, two<br />
measurements were carried out for each <strong>of</strong> the surface roughness parameters and four<br />
measurements were taken for the burr heights. Results are shown in Appendix I. Both means <strong>of</strong><br />
Ra and Burr (Ra mean , Burr mean ), along with standard deviations (Ra STDE , Burr STDE ) are calculated<br />
for each run. The responses were consequently expressed in form <strong>of</strong> regression equations as<br />
follows:<br />
(7)<br />
The analysis <strong>of</strong> variance (ANOVA) is summarized in tables 2-5:<br />
Table 2. Response Surface Regression: Ra mean<br />
Term Coefficient SE Coef T P<br />
Constant 329.1 51.14 6.435 0.000<br />
f 2470.7 3947.5 0.626 0.538<br />
S -0.6 0.18 -3.291 0.003<br />
P -300.4 101.40 -2.963 0.007<br />
P 2 100.6 65.46 1.537 0.138<br />
f P -13522.0 7292.43 -1.854 0.077<br />
S P 1.0 0.33 3.041 0.006<br />
80
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061 ALUMINUM USING DESIRABILITY FUNCTIONS<br />
Table 3. Response Surface Regression: Ra STDE<br />
Term Coefficient SE Coef T P<br />
Constant 92.5 31.53 2.934 0.007<br />
f 5581.3 2388.15 2.337 0.028<br />
S -0.4 0.11 3.251 -0.003<br />
P -107.0 52.52 -2.038 0.053<br />
f P -13949.3 4349.02 -3.207 0.004<br />
S P 0.6 0.20 3.215 0.004<br />
Table 4. Response Surface Regression: Burr mean<br />
Term Coefficient SE Coef T P<br />
Constant 0.050 0.013 3.978 0.001<br />
f -8.325 2.633 -3.162 0.004<br />
P 0.008 0.010 0.880 0.388<br />
f 2 319.409 163.113 1.958 0.062<br />
f S 0.010 0.008 1.262 0.220<br />
f P -1.557 1.623 -0.959 0.347<br />
Table 5. Response Surface Regression: Burr STDE<br />
Term Coefficient SE Coef T P<br />
Constant 0.00364 0.00409 0.889 0.382<br />
f -0.97956 0.31322 -3.127 0.004<br />
S 0.00003 0.00001 2.027 0.053<br />
P -0.01071 0.00972 -1.102 0.281<br />
P 2 0.01267 0.00964 1.315 0.200<br />
Although the main effect <strong>of</strong> feed (f) is not significantly impacting Ra, the interaction <strong>of</strong> feed<br />
and peck (fP) and the interaction <strong>of</strong> speed and peck (fS) are significant. All variables and their<br />
interactions impacted the standard deviation <strong>of</strong> Ra.<br />
On the other hand, feed was the only factor impacted burr mean significantly. Both feed and speed<br />
impacted the standard deviation <strong>of</strong> burr.<br />
3.3 Desirability Optimization<br />
In this multi-objective optimization problem, four different objectives were selected to be<br />
optimized: Ra mean, Ra standard deviation, Burr height mean, and Burr height standard deviation.<br />
The goal <strong>of</strong> all <strong>of</strong> the four objectives is the lower the better (LTB), i. e. minimization. In multiresponse<br />
optimization desired importance is given to all the responses. It’s the author’s point <strong>of</strong> view<br />
that Burr is more important than surface finish. Table 6 shows the range <strong>of</strong> the responses, goals,<br />
weights, and importance assigned to each response.<br />
Table 6. Response goals<br />
Term Lower Target Upper Weight Importance<br />
Ra mean (micro inch) 120 120 200 1 1<br />
Ra STDE (micro inch) 0 10 20 1 1<br />
Burr mean (inch) 0 0 0.04 1 8<br />
Burr STDE (inch) 0 0.001 0.005 1 1<br />
81
MAHMOUD I. AWAD, ELSAYED ORADY, CHENMAYA DANDEKAR<br />
In addition to response goals, there are some constraints that limit the search space in the<br />
considered optimization problem to the allowed ranges for the process parameters given by the<br />
validity range <strong>of</strong> the experimental models.<br />
Minitab S<strong>of</strong>tware was used to search for the optimal solution using desirability functions<br />
considering the process parameters constraints, the optimal solution is summarized in Figure 2.<br />
Optimal<br />
D<br />
High<br />
Cur<br />
0.81476 Low<br />
Composite<br />
Desirability<br />
0.81476<br />
Ra_mean<br />
Minimum<br />
y = 127.0951<br />
d = 0.91131<br />
Burr_mea<br />
Minimum<br />
y = 0.0094<br />
d = 0.76497<br />
Ra_sigma<br />
Targ: 10.0<br />
y = 9.8466<br />
d = 0.98466<br />
f S P<br />
0.0107 391.0 1.0<br />
[0.0096] [324.8889] [0.5688]<br />
0.0011 204.0 0.0513<br />
Burr_sig<br />
Targ: 0.0010<br />
y = 0.0010<br />
d = 0.99822<br />
Figure 2: Optimization results<br />
The results indicate that feed should be set at a value closer to the maximum value (0.0096),<br />
high speed (324.9), and intermediate peck angle value (0.57). Setting the drilling process at these<br />
values is expected to provide an acceptable results relative to the desired targets set in Table 6. For<br />
example, the optimal settings will result in a surface roughness <strong>of</strong> 127micro inch which is acceptable<br />
compared to the target which is 120 micro inch. Similarly, the optimal settings resulted in a burr<br />
mean <strong>of</strong> 0.0094 inch which is acceptable compared to the zero burr height target.<br />
4. DISCUSSION AND CONCLUSION:<br />
To verify the results obtained in the previous section, several confirmation runs were<br />
conducted with the optimal settings. Figures 3 and 4 depict the results <strong>of</strong> the confirmation runs <strong>of</strong> Ra<br />
and burr height in the form <strong>of</strong> a boxplot respectively with confidence intervals <strong>of</strong> the means. A t-test<br />
<strong>of</strong> t Ra mean was conducted where (H 0 : Ra_ave = 127, vs H a : Ra_ave 127). The results indicated<br />
that there is no sufficient evidence to reject the null hypothesis with a p-value <strong>of</strong> 0.124. Similarly, a t-<br />
test for burr height was conducted (H 0 : Burr_ave = 0.0094, vs H a : Burr_ave 0.0094) and indicated a<br />
lack <strong>of</strong> evidence to disprove H 0 with a p-value <strong>of</strong> 0.053. Although the roughness and burr means<br />
were close to the predicted values, the standard deviations <strong>of</strong> both responses were <strong>of</strong>f. The standard<br />
deviation <strong>of</strong> the confirmation runs indicated a standard deviation <strong>of</strong> 22 micro inch (predicted =10<br />
82
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061 ALUMINUM USING DESIRABILITY FUNCTIONS<br />
micro inch) for Ra and 0.0033 inch (predicted =0.001inch) for burr height. The inaccuracy <strong>of</strong><br />
variability prediction for Ra and burr height is due to the nature <strong>of</strong> the responses. <br />
Figure 3 Boxplot <strong>of</strong> Ra confirmation runs with 95% confidence interval<br />
Ho<br />
_<br />
X<br />
0.0050<br />
0.0075<br />
0.0100<br />
Burr height<br />
0.0125<br />
0.0150<br />
0.0175<br />
Figure 4 Boxplot <strong>of</strong> burr height confirmation runs with 95% confidence interval<br />
Other variables may influence the quality <strong>of</strong> drilled hole such as tool wear, cutting forces (torque<br />
and thrust force) and machine and fixture rigidity. These factors were not included in the<br />
conducted experiments and can be added in future research. In this study two measures <strong>of</strong> the<br />
hole quality were included, namely; burr height and surface roughness. However, there are other<br />
quality measures that can be included in the model to describe the hole quality in more<br />
comprehensive manner. These measures may include circularity (roundness), cylindericity, and<br />
center position.<br />
The optimization scheme proposed in this research simultaneously optimizes both the mean and<br />
standard deviation for multiresponse system design problems. The proposed method incorporates<br />
noise factors within the experimentation phase in order to determine the settings <strong>of</strong> the control<br />
factors to create response models that are insensitive to internal or external noise. The proposed<br />
approach also uses a nonlinear programming technique to model a multiresponse problem as a<br />
single objective function using a desirability index. Highest desirability could be obtained at<br />
high level <strong>of</strong> feed, high speed, and an intermediate peck. Using desirability functions, both targets<br />
were achieved simultaneously and different importance levels were assigned to responses.<br />
5. REFERENCES<br />
[1.] Box, G. (1988). “Signal-to-noise ratios performance criteria, and transformations.”,<br />
Technometrics, Vol. 30(1), pp. 1–17<br />
[2.] Coleman, D. E., and Montgomery, D. C. “A Systematic Approach to Planning for a Designed<br />
Industrial Experiment (with Discussion).,” Technometrics. Vol. 35, pp. 1-27 (1993).<br />
[3.] Derringer, G., & Suich, R. (1980). “Simultaneous optimization <strong>of</strong> several response variables.”,<br />
Journal <strong>of</strong> Quality Technology, Vo. 12(4), pp. 214–219<br />
83
MAHMOUD I. AWAD, ELSAYED ORADY, CHENMAYA DANDEKAR<br />
[4.] Furness, R. J., Wu, C.L., and Usloy, A.G., (1992). “Statistical Analysis <strong>of</strong> the Effects <strong>of</strong> Feed<br />
Speed and Wear on Hole Quality in Drilling.” Sensors and Signal Processing for Manufacturing,<br />
PED-Vol. 55, pp. 97-112,<br />
[5.] Gallowaym, D. F. (1995). “Some experiments on influence <strong>of</strong> Various Factors on Drill<br />
Performance.” Transaction <strong>of</strong> the ASME, pp. 191-231<br />
[6.] Hamade, R.F., Ismail, F. (2005). “A case for aggressive drilling <strong>of</strong> aluminum.”, Journal <strong>of</strong><br />
Materials Processing Technology, Vol. 166, pp. 86-97<br />
[7.] Khuri, A. I., & Conlon, M. (1981). “Simultaneous optimization <strong>of</strong> multiple responses represented<br />
by polynomial regression functions.”, Technometrics, Vol. 23(4), pp. 363–375<br />
[8.] Kuehl, R. O. (1999). Design <strong>of</strong> experiments: Statistical principles <strong>of</strong> research design and analysis,<br />
Duxbury Press, New York<br />
[9.] Lauderbaugh L. Ken (2001). “An Exit Burr Model for Drilling <strong>of</strong> Metals.” Transaction <strong>of</strong> the<br />
ASME, Journal <strong>of</strong> Manufacturing Science and Engineering, Vol. 23, pp. 562-66<br />
[10.] Lauderbaugh L. Ken (2009). “Analysis <strong>of</strong> the effect <strong>of</strong> process parameters on exit burrs in drilling<br />
using a combined simulation and experimental approach.”, Journal <strong>of</strong> Materials Processing<br />
Technology, Vol. 209, pp. 1909-1919.<br />
[11.] List, G., Nouri, M., Gehin, D., Gomez, S., Manaud, J.P., Petitcorps, Y. L., Girot, F. (2005).<br />
“Wear behavior <strong>of</strong> cemented carbide tools in dry machining <strong>of</strong> aluminum alloys.” International<br />
Wear, vol. 259, pp. 1177-1189.<br />
[12.] Montgomery, D. C., Design and Analysis <strong>of</strong> Experiments (2000), 5 th Ed., Wiley, New York,<br />
[13.] Myers, R. H., and Montgomery, D. C., Response surface methodology (2002). Process and<br />
product optimization using designed experiments, Wiley, New York,<br />
[14.] Machinery’s Handbook, 25 th Edition, 1996 Industrial Press Inc., New York (1996).<br />
[15.] Madhavan Nair, R., Durairajan, B. and Bahr, B., (2002). “High Speed Drilling <strong>of</strong> Al-2024-T3<br />
Alloy.” SAE Technical Paper 2002-01-1516<br />
[16.] Nair, V. N., Abraham, B., MacKay, J., Nelder, J. A., Box, G., Phadke, M. S., et al. (1992).<br />
“Taguchi’s parameter design: A panel discussion.”, Technometrics, Vol. 34(2), pp. 127–161..<br />
[17.] Nouri, M., List, G., Gehin, D., (2005). “Effect <strong>of</strong> machining parameters and coating on wear<br />
mechanisms in dry drilling <strong>of</strong> aluminum alloys.” International Journal <strong>of</strong> Machine Tools and<br />
Manufacturing, vol. 45, pp. 1436-1442.<br />
[18.] Pan, H. L., Chen, Y., Orady, E., and Yan, F. (1995). "Tool Wear Monitoring by Acoustic Analysis<br />
Through a Solid Path.," International Journal <strong>of</strong> Production Research. Vol. 33, No. 9, pp. 2455-<br />
2470<br />
[19.] Pan, H.L., Orady, E, and Chen, Y. (1994). "Tool Useful Life Evaluation and Monitoring by<br />
Thermal EMF," Proceedings <strong>of</strong> the International Symposium on Manufacturing Science and<br />
Technology for the 21st Century, Beijing, China,<br />
[20.] Pande, S.S., Relekar, H.P., (1986). Investigations on reducing burr formation in drilling.<br />
International Journal <strong>of</strong> Machine Tool Design Research, Vol. 26 (3), pp. 339–348.<br />
[21.] Pande, B., Aramendi, G., Rivero, A., de Lacalle, L.N., (2005).” Monitoring <strong>of</strong> drilling for burr<br />
detection using spindle torque”. International Journal <strong>of</strong> Machine Tools & Manufacture, Vol. 45,<br />
pp.1614-1621.<br />
[22.] Pignatiello, J. J. (1993). “Strategies for robust multiresponse quality <strong>engineering</strong>”, IIE<br />
Transactions, Vol. 25(3), pp. 5–15<br />
[23.] Taguchi, G. (1987). System <strong>of</strong> experimental design: Engineering methods to optimize quality and<br />
minimize costs. Dearborn, MI: American Supplier Institute.<br />
[24.] Thangaraj, A. and Langenstien, M., (1990). “Investigation into the Relationship between Hole<br />
Quality and Drilling Forces.” Proceedings <strong>of</strong> Manufacturing International ’90: Advances in<br />
Materials and Automation, pp. 149-157<br />
84
DRILLING PROCESS ROBUST OPTIMIZATION FOR 6061 ALUMINUM USING DESIRABILITY FUNCTIONS<br />
APPENDIX I:<br />
Ra (micro inch)<br />
Burr Height (inches)<br />
Standard<br />
Order<br />
Hole<br />
No.<br />
Drilling<br />
order 1 2 3 Ra_mean Ra_STDE 1 2 3 Burr_M Burr_STDE<br />
1 1 7 140.0 142.5 113.0 131.8 16.4 0.01575 0.01880 0.03050 0.02168 0.00779<br />
2 2 9 146.0 134.5 145.0 141.8 6.4 0.01075 0.01338 0.01238 0.01217 0.00132<br />
3 3 5 173.0 153.5 126.5 151.0 23.4 0.00750 0.00975 0.01250 0.00992 0.00250<br />
4 4 15 220.0 205.0 175.0 200.0 22.9 0.04613 0.03113 0.04200 0.03975 0.00775<br />
5 5 14 133.0 129.5 131.5 131.3 1.8 0.00763 0.01240 0.01325 0.01109 0.00303<br />
6 6 2 138.0 122.0 156.0 138.7 17.0 0.00875 0.01700 0.00625 0.01067 0.00563<br />
7 7 11 163.0 151.0 159.0 157.7 6.1 0.01925 0.01425 0.04850 0.02733 0.01850<br />
8 8 6 123.0 131.0 131.0 128.3 4.6 0.00950 0.01038 0.01250 0.01079 0.00154<br />
9 9 1 115.0 110.0 185.0 136.7 41.9 0.01625 0.03400 0.04975 0.03333 0.01676<br />
10 10 8 171.0 165.0 157.0 164.3 7.0 0.01425 0.00893 0.01163 0.01160 0.00266<br />
11 11 10 145.0 129.5 123.5 132.7 11.1 0.00800 0.00875 0.01595 0.01090 0.00439<br />
12 12 3 115.0 128.0 121.5 121.5 6.5 0.01125 0.01300 0.01100 0.01175 0.00109<br />
13 13 4 135.0 142.0 119.0 132.0 11.8 0.00950 0.01613 0.01425 0.01329 0.00341<br />
14 14 12 135.5 127.0 153.0 138.5 13.3 0.00775 0.00775 0.01195 0.00915 0.00242<br />
15 15 13 152.5 119.5 113.5 128.5 21.0 0.01425 0.02425 0.01750 0.01867 0.00510<br />
16 1 9 167.0 117.0 146.0 143.3 25.1 0.01500 0.00375 0.01063 0.00979 0.00567<br />
17 2 6 143.0 152.0 167.0 154.0 12.1 0.01513 0.01350 0.02063 0.01642 0.00373<br />
18 3 5 180.5 210.0 210.0 200.2 17.0 0.02150 0.03263 0.01775 0.02396 0.00774<br />
19 4 10 201.0 200.0 220.0 207.0 11.3 0.01950 0.01288 0.01175 0.01471 0.00419<br />
20 5 11 148.0 350.0 270.0 256.0 101.7 0.00988 0.00963 0.00725 0.00892 0.00145<br />
21 6 8 199.0 148.0 144.5 163.8 30.5 0.00988 0.01050 0.01175 0.01071 0.00095<br />
22 7 3 143.0 142.0 148.0 144.3 3.2 0.00550 0.00625 0.01050 0.00742 0.00270<br />
23 8 2 138.0 107.0 133.0 126.0 16.6 0.01675 0.00850 0.01213 0.01246 0.00414<br />
24 9 7 159.0 185.5 245.0 196.5 44.0 0.02638 0.01700 0.02618 0.02318 0.00536<br />
25 10 12 159.0 153.0 176.0 162.7 11.9 0.02080 0.01900 0.02100 0.02027 0.00110<br />
26 11 1 157.5 193.0 177.5 176.0 17.8 0.00925 0.01250 0.01400 0.01192 0.00243<br />
27 12 14 149.0 152.0 153.0 151.3 2.1 0.00713 0.00775 0.00900 0.00796 0.00095<br />
28 13 4 130.5 139.0 133.0 134.2 4.4 0.00963 0.01200 0.01300 0.01154 0.00173<br />
29 14 13 132.5 155.0 151.0 146.2 12.0 0.03038 0.00675 0.00750 0.01488 0.01343<br />
30 15 15 165.5 202.0 186.5 184.7 18.3 0.02970 0.03850 0.01563 0.02794 0.01154<br />
85
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 87-105 (2010)<br />
© 2009 ALHOSN University<br />
NEIGHBORHOOD DESIGN AND WALKABILITY:<br />
A SYNTHESIS FROM PLANNING, DESIGN,<br />
TRANSPORTATION AND<br />
ENVIRONMENTAL HEALTH FIELDS<br />
Mohamad Kashef *<br />
Department <strong>of</strong> Architecture, ALHOSN University, P.O Box 38772, Abu Dhabi, UAE<br />
ABSTRACT: This study develops a nuanced understanding <strong>of</strong> issues related to neighborhood design and<br />
walkability, which received increased attention over the last two decades. It examines a broad range <strong>of</strong><br />
environmental health, transportation, and design/planning studies. The study argues that the bulk <strong>of</strong> published<br />
empirical research relied on statistical tools that measured the design ingredients rather than the morphological and<br />
spatial structures that enhance walkability. The qualitative design aspects <strong>of</strong> built environments were hardly<br />
translated to quantifiable variables in much <strong>of</strong> the reviewed empirical research. This paper highlights critical<br />
configurational and space syntax dynamics that complement the current understanding <strong>of</strong> neighborhood design and<br />
walkability. It opens the door for further cross disciplinary research aimed at understanding and potentially<br />
optimizing the relationship between urban form and pedestrian movement.<br />
1. INTRODUCTION<br />
This study engages the planning and urban design literature as well as transportation and<br />
environmental health fields to develop an interdisciplinary understanding <strong>of</strong> issues related to<br />
neighborhood design and walkability. It examines intellectual diversion and/or conversion<br />
between environmental health studies, designers/planners narratives, and transportation analyses.<br />
Over the last four or five decades, various design and planning publications called for reviving<br />
traditional built forms and reintroducing mixed uses to enhance walkability and urban vitality<br />
(Jacobs [53]; Alexander [4]; Rowe et al [81]; Krier [58]; Trancik [90]; Barnette [10]). The New<br />
Urbanists also expressed a renewed interest in traditional spatial configurations and building<br />
typologies. They contended that fine-grain connected urban blocks, mix <strong>of</strong> uses, and attention to<br />
architectural details such as sidewalks, porches, and visual stimuli can encourage walking,<br />
bilking, and civic engagement. The presumption is that technological and functional imperatives<br />
<strong>of</strong> the dispersed metropolis did not change the basic human desire for well-designed<br />
neigborhoods that support residents’ daily needs within a five-minute walking distance (Duany et<br />
al [24]; Katz [56]; Dutton [26]). A host <strong>of</strong> recent empirical (statistically-supported)<br />
investigations from the public and environmental health fields, on the other hand, pointed to the<br />
lack <strong>of</strong> causality between neigborhood design and walkability. Most <strong>of</strong> these studies noted that<br />
certain neigborhood types with design features such as connected sidewalks, architectural variety,<br />
etc. are positively correlated with walking and biking. They generally explained such associations<br />
in terms <strong>of</strong> the market self-selection process, i.e., people who like walking and biking commonly<br />
choose to live in places that support their way <strong>of</strong> life (Cervero et al [19]; Atkinson et al [5]; Reed<br />
et al [79]; Oaks et al [72]; Lovasi et al [65]; Nagel et al [70]).<br />
This paper argues that the bulk <strong>of</strong> published empirical research relied on statistical tools that<br />
measured the design ingredients rather than the morphological and spatial structures that enhance<br />
____________________________________<br />
* Corresponding Author. Tel.: +971 4070511<br />
E-mail : m.kashef@<strong>alhosn</strong>u.ae<br />
87
MOHAMMED KASHEF<br />
walkability. The paper highlights shortcomings <strong>of</strong> statistically-supported research in capturing<br />
the essence <strong>of</strong> qualitative design values and transforming them into measurable study variables.<br />
Critical configurational and syntactical dynamics (Hiller [51], [49], [48]; Bafna [7]; Peponis et al<br />
[74]; Baran et al [9]) were not considered in measuring the cause and effect relationship between<br />
design and walkability. The paper is organized into three major parts: the first part reviews the<br />
evidence from environmental health fields; the second part encompasses the results from<br />
planning, design and transportation studies; the paper concludes with a discussion that<br />
synthesizes the evidence and provides an interdisciplinary perspective that potentially benefits<br />
public policy debates regarding land use and subdivision regulations that support walking and<br />
biking.<br />
2. ENVIRONMENTAL AND PUBLIC HEALTH STUDIES<br />
These comprise a broad range <strong>of</strong> empirical investigations within health related disciplines such as<br />
public health, nutrition, exercise physiology, epidemiology, and preventive medicine, among<br />
several others. They addressed the nexus between walking and neigborhood design from a variety<br />
<strong>of</strong> perspectives that consider objective and perceived built environments, personal attitudes and<br />
behavior, safety (real and perceived), environmental constraints, purpose (leisure or<br />
work/shopping related nonmotorized travel, as well as health and lifestyle characteristics). The<br />
presence <strong>of</strong> sidewalks is <strong>of</strong>ten touted by the New Urbanism as an important environmental<br />
resource that supports walking in neigborhoods (Calthorpe et al [16]; Fulton [35]; Duany et al<br />
[24], [25]; Katz [56]). The evidence from health studies regarding the impact <strong>of</strong> sidewalks on<br />
physical activity and walking in general is mixed. The following discussion articulates the<br />
findings <strong>of</strong> health studies under three themes or headlines: walking behavior and design,<br />
sidewalks, and street connectivity. These themes are not mutually exclusive; there is a great deal<br />
<strong>of</strong> overlap and such articulation is merely <strong>of</strong>fered as an organizational framework for the large<br />
amount <strong>of</strong> data addressed below.<br />
2.1 Walking Behavior and Design<br />
In an attempt to predict walking behavior, Ryan Rhodes et al [80] integrated personality, the<br />
perceived environment, and time planning into a framework called “TPB” (theory <strong>of</strong> planned<br />
behavior). The study included various measures for personality traits, actual walking,<br />
neigborhood geography (sidewalks, architecture, aesthetics, etc.), and attitudes towards outdoor<br />
physical activity for leisure purposes. Research instruments were administered to 358 residents <strong>of</strong><br />
British Columbia, Canada. The study model provided evidence that leisure-time walking is<br />
largely intention based behavior with an additional independent contribution from physical<br />
proximity and accessibility <strong>of</strong> neigborhood retail and recreation facilities. As shown in the<br />
adapted model diagram (Fig. 1), neigborhood design and aesthetics impact walking but only<br />
through mediating intention based variables that include attitudes and other social cognitive<br />
constructs. In that sense, the environment provides infrastructure support but is not the cause for<br />
walking per se. Notable among the results <strong>of</strong> this study is the impact <strong>of</strong> proximity <strong>of</strong> parks to<br />
homes. Individuals who are habitually active and lived at a reasonable walking distance from<br />
recreation venues were more inclined to walk (Rhodes et al [80]). Proximity <strong>of</strong> homes to exercise<br />
facilities was also positively correlated with the frequency <strong>of</strong> exercise among San Diego residents<br />
(Sallis et al [82]). These results are <strong>of</strong> particular significance to community planning practice. It<br />
88
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
shows that adding destinations <strong>of</strong> interest and recreation venues can promote walking among<br />
active residents and help close the intention-walking/activity gap.<br />
Mixed-Use<br />
Affective<br />
Attitude<br />
Design<br />
Elements<br />
Intention<br />
Walking<br />
Instrumental<br />
Attitude<br />
Aesthetics<br />
Figure 1. Perceived environment and theory <strong>of</strong> planned behavior model to predict walking<br />
(Adapted from: Rhodes et al [80])<br />
2.2 Sidewalks<br />
Sidewalk availability did not figure as a prominent indicator <strong>of</strong> physical activity among residents<br />
<strong>of</strong> different neigborhoods in Washington State. Objectively measured built environment<br />
characteristics including sidewalks within a mile from research participants’ homes accounted for<br />
only one percent <strong>of</strong> residents’ total physical activity (Lovasi et al [65]). Research conducted in<br />
Portland, Oregon (fifty-six neigborhoods) yielded similar results. Findings did not suggest that<br />
built environment plays a significant role in the walking and physical activity <strong>of</strong> older adults.<br />
Results were consistent with previous studies in which sidewalk presence and neighborhood<br />
aesthetics were associated with increased physical activity among active residents who habitually<br />
walk (Nagel et al [70]; Atkinson et al [5]). In a stark contrast with previous results, the findings<br />
<strong>of</strong> some other studies pointed to positive associations between the perceived presence <strong>of</strong><br />
neighborhood sidewalks and physical activity behaviors (Ainsworth et al [2]; Addy et al [1]; Lee<br />
et al [61]). These findings were significant because research participants were classified as<br />
irregular walkers, namely, white older adults and African American women. This, in a sense,<br />
substantiates the New Urbanists’ claims regarding the impact <strong>of</strong> neigborhood design on walking<br />
behavior.<br />
Various studies reported that both objective and subjective characteristics <strong>of</strong> the built<br />
environment were correlated with walking, cycling, and outdoor physical activity in general.<br />
Objectively measured correlates included the physical quality <strong>of</strong> walking routes and the presence,<br />
type, and geographic proximity <strong>of</strong> destinations such as retail shops, parks, and other recreation or<br />
cultural facilities. Subjective variables included the perception <strong>of</strong> safety (traffic and crime<br />
related), the convenience <strong>of</strong> neigborhood services, and the perceived visual quality <strong>of</strong><br />
neigborhood landscape and architecture (Moudon et al [69]; McCormack et al [67]; Boehmer et<br />
89
MOHAMMED KASHEF<br />
al [11]; Lesliea et al [62]; Geus et al [36]). Using a multi-stage sampling design and existing GIS<br />
databases, Li et al (2005) surveyed all neigborhoods in Portland, Oregon (521, 121 people) and<br />
reported positive associations between urban form and walking activity. Places with mixed-use<br />
and higher levels <strong>of</strong> green and open spaces were significantly associated with increased walking<br />
activity. The results <strong>of</strong> the latter surveys were noteworthy because findings showed that variation<br />
in walking activity across neigborhoods was accounted for by environmental characteristics.<br />
2.3 Street Connectivity<br />
Street connectivity also emerged as an important attribute <strong>of</strong> walkable neigborhoods. This is a<br />
measure <strong>of</strong> the accessibility achieved by having smaller blocks (less than 1000 feet long) and<br />
ease <strong>of</strong> movement between developed blocks without being obstructed by dead-ends. A<br />
connectivity index is commonly derived by dividing the number <strong>of</strong> roadway links by the number<br />
<strong>of</strong> roadway intersections and nodes (cul-de-sacs included). A higher number indicates a wellconnected<br />
street system (Ewing, 1996). Data from Metropolitan Atlanta’s region, Oakland, New<br />
Zealand, and Perth, Australia showed a significant correlation between increased levels <strong>of</strong><br />
outdoor physical activity among adults and well-connected streets, continuous pedestrian<br />
sidewalk network, and mix <strong>of</strong> land uses (Frank et al [33]; Badland et al [6]; Chin et al [20]).<br />
Proximity <strong>of</strong> workplace and commercial destinations was highlighted by some studies as<br />
significant contributors to total transport-related walking. It was estimated that individuals who<br />
regularly walk to work would approximately accumulate 166 weekly minutes <strong>of</strong> walking, which<br />
meets health guidelines for adults (Cerin et al [18]). Residents <strong>of</strong> mixed-use high walkability<br />
neigborhoods reported 70 more minutes <strong>of</strong> physical activity per week than did residents <strong>of</strong> singlefamily<br />
low-walkability neigborhoods. They also had lower obesity prevalence after adjusting for<br />
individual demographics (Saelens et al [83], [84]).<br />
Neighborhood design and aesthetics figured prominently in other studies as conducive to active<br />
aging, enhanced walkability, and higher perception <strong>of</strong> safety and sense <strong>of</strong> community (Michael et<br />
al [68]; Leyden [63]; Wood et al [92]). This again substantiates some <strong>of</strong> the New Urbanists’<br />
claims regarding the impact <strong>of</strong> Physical elements such as building aesthetics, streetscapes, vistas,<br />
porches, etc. on to the livability and vitality <strong>of</strong> neigborhoods. Such elements are frequently cited<br />
in the New Urbanism literature as the building blocks <strong>of</strong> communities. They have the potential <strong>of</strong><br />
enhancing social interaction and hence igniting the culture <strong>of</strong> community. The New Urbanists<br />
generally contend that reinstating the traditional morphology <strong>of</strong> street and civic architecture can<br />
contribute to restoring the lost sense <strong>of</strong> place and reviving many long-gone communal practices.<br />
At present, suburban patterns are completely dependent on the private automobile and lack the<br />
basic ingredients conducive to collective social and cultural dynamics (Duany [24; Katz [56];<br />
Calthorpe [16], [17]; Kelbaugh [57]). Rob Boer et al [58] adopted the Smart Scorecard developed<br />
by the New Urbanism (Fleissig et al, 2002) as a research instrument to assess the impact <strong>of</strong><br />
neigborhood design on walking trips in ten U.S. metropolitan areas. The Smart Scorecard<br />
evaluates various built environment measures such as density, mix <strong>of</strong> uses, street network<br />
connectivity, block sizes, diversity, landscape, etc. Based on data analysis from the 2000 U.S.<br />
census, the study indicated that only three <strong>of</strong> the ten critical components <strong>of</strong> the Smart Scorecard<br />
were found to be associated with walking, namely: accessibility and mobility choices<br />
(connectivity), compactness (density), and mix <strong>of</strong> uses (Boer et al [12]).<br />
90
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
2.4 Summarizing the Results <strong>of</strong> Health Studies<br />
It seems that health-related studies <strong>of</strong>fer more evidence on the positive side <strong>of</strong> association<br />
between built environment and walkability. However, the findings <strong>of</strong> these studies were largely<br />
based on cross-sectional data. Most <strong>of</strong> the positive associations were derived from respondents<br />
perceived or subjective characterizations <strong>of</strong> the built environment. Almost none <strong>of</strong> the reviewed<br />
health studies claimed to have identified causal linkages between environmental attributes and<br />
physical activity; study results could not attest to whether the criteria <strong>of</strong> time precedence and nonspuriousness<br />
were established. Self-selection always figured in study limitations and caveats as a<br />
possible explanation for the positive associations <strong>of</strong> neighborhood design and walkability; people<br />
who walk regularly choose neighborhood environments that promote walkability and not vice<br />
versa.<br />
Two very recent studies with quasi-longitudinal design provided contradictory evidence. Handy<br />
et al [45] pointed to a causal linkage between neigborhood environment and physical activity.<br />
Wells et al [91], on the other hand, asserted that neigborhood features alone such as sidewalks,<br />
front porches, small set-back distances, and aesthetics may not be enough to affect walking.<br />
Residents <strong>of</strong> New Urbanism neigborhoods did not walk significantly more than those living in<br />
conventional suburbs (Wells et al [91]). Research findings from the Health field did not provide a<br />
definitive answer as to the relationship <strong>of</strong> walkability and neighborhood design one way or the<br />
other. The only possible recommendation out <strong>of</strong> health studies is that neighborhood design does<br />
affect the opportunities for walkability; it has the potential to enhance nonmotorized forms <strong>of</strong><br />
travel, especially for a select group <strong>of</strong> physically active individuals. The following part <strong>of</strong> the<br />
paper examines the evidence from the design, planning, and transportation fields. While studying<br />
similar constructs, the approach in most <strong>of</strong> these studies varies from health studies.<br />
Transportation and some planning studies generally adopt quantitative methodologies with a<br />
variety <strong>of</strong> measuring devices that include structural equations, systems analysis, computer<br />
modeling and simulation, etc. Other planning and design studies are qualitative in nature, at times<br />
intuitive, and provide rich historiography and in-depth narratives about cities, communities, and<br />
the forces that shape them.<br />
3. TRANSPORTATION AND PLANNING/DESIGN STUDIES<br />
There is a great deal <strong>of</strong> overlap between planning, design, and transportation studies, especially in<br />
regard to the impact <strong>of</strong> design on physical activity. Transportation studies focus more on travel<br />
related issues such as modal split, traffic management and safety, as well as street network<br />
optimization. Planning and design fields, on the other hand, capitalize on the data produced by<br />
transportation studies to establish connections between neigborhood architecture, street network<br />
integration, walkability, and sociability. Transportation and planning issues are inextricably<br />
linked, which may not allow addressing them separately. The following discussion articulates the<br />
findings under three themes: land use and transportation, walkability and place attachment, and<br />
space syntax and connectivity. The issues discussed under these categories are overlapping, and<br />
such articulation is mainly <strong>of</strong>fered as an organizational framework for the large amount <strong>of</strong> data<br />
addressed below.<br />
91
MOHAMMED KASHEF<br />
3.1 Land Use and Transportation<br />
Coordinating transportation and land use emerged as a critical component <strong>of</strong> viable economic<br />
development policies, infrastructure planning and management, and more significantly for this<br />
paper, as a potential modifier <strong>of</strong> travel behavior. Zoning and land use are used exchangeably to<br />
denote an approach to planning urban and suburban developments. Traditional zoning, <strong>of</strong>ten<br />
referred to as “Euclidean Zoning” after the approach was approved in the Supreme Court<br />
decision in 1926, began as a political response to the obnoxious effects <strong>of</strong> industry on human<br />
settlements. Zoning regulation progressed further and enacted exclusive zones for <strong>of</strong>fice use,<br />
commercial, and different housing categories. The rationale behind land use regulation was to<br />
promote the public’s health, safety, and welfare. It was also meant to stabilize property values,<br />
encourage stable and congenial neigborhoods, and regulate traffic flow. Over the past several<br />
decades, zoning was subject to regular criticism and commonly described as inflexible and<br />
exclusionary. It codified the relocation <strong>of</strong> commercial and <strong>of</strong>fice functions to suburbs, causing a<br />
spatial mismatch between lower classes in downtowns and suburban employment locations. The<br />
single-use areas prescribed by zoning resulted in sprawling suburban patterns with substantial<br />
distances and voids between related urban functions. The private automobile was the most logical<br />
choice, and other modes <strong>of</strong> transport such as public transit, walking, or biking became unviable.<br />
There is a heated planning debate regarding the role that zoning played in shaping the American<br />
urban/suburban form. Some planning studies portrayed zoning as a cultural artefact that reflected<br />
long-standing societal choices, which continue to be asserted in buying decisions and outright<br />
preference for auto-dependent, single-use suburban locations. These referred to the efficiency <strong>of</strong><br />
market-driven policies and the economic rationale <strong>of</strong> bid-rent curves which show how the highest<br />
and best use <strong>of</strong> land is determined. They maintained that zoning bylaws and land use regulations<br />
do not have as much effect on decentralizing American cities as consumer preferences and tastes;<br />
developers are market-conscious and do not usually risk building an unacceptable product<br />
(Kashef [55]). Other planning and design studies referred to zoning as the culprit behind the<br />
dispersion and fragmentation <strong>of</strong> the North American built form. In that sense, suburban sprawl<br />
was promoted by a combination <strong>of</strong> zoning regulations, mortgage policies, and massive highway<br />
and road building programs that were aided by the federal and state governments. These policies<br />
helped instate the private automobile as the one and only viable mode <strong>of</strong> transport within most<br />
American cities. Most people had limited choices and were sold on a highly publicized and<br />
government-subsidized ideal (the American dream) that shaped the development market over the<br />
past several decades (Kashef [55]; Ewing [28]; Duany et al [25]). The New Urbansits have<br />
always emphasized the need to replace Euclidean zoning with flexible built form controls that<br />
provide more options and create socially and physically integrated communities and<br />
neigborhoods. The latter should promote mix <strong>of</strong> uses and built forms and coordinate land use and<br />
transportation, thereby allowing a rich modal split between motorized (car/public transit) and<br />
non-motorized (walking/biking) transportation. The New Urbanist agenda is predicated on the<br />
hypothesis that changing neighborhood land use configurations and increasing their density<br />
would influence travel behavior and promote public transit, walking and biking (Duany et al [24],<br />
[25]; Calthorpe et al [16]; Katz [56]; Dutton [26]).<br />
The findings <strong>of</strong> some recent transportation and planning studies cast doubts on the potential <strong>of</strong><br />
neigborhood design on travel behavior. Bagley et al [8] examined the relationship <strong>of</strong> residential<br />
92
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
neigborhood type to travel behavior, incorporating attitudinal, lifestyle, and demographic<br />
variables. Data were collected from five neigborhoods in the San Francisco Bay area with singleand<br />
mixed-use configurations. The results showed that people with different travel propensities<br />
select themselves into neigborhoods that support their lifestyle and attitudinal predispositions.<br />
Mixed land use configuration in and <strong>of</strong> itself did not affect travel behavior (Bagley et al [8]).<br />
Other Bay Area studies have shown that mixed land uses and close proximity to retail services<br />
may induce nonmotorized transport. However, the data analysis revealed that land use on its own<br />
had a modest and sometimes statistically insignificant effect on walking and biking. Various<br />
exogenous factors, such as topography, darkness, and rainfall had far stronger influences than<br />
land use. Demographic characteristics <strong>of</strong> trip makers were also far stronger predictors <strong>of</strong> walking<br />
and biking rather than land use factors (Cervero et al [19]; Schwanen et al [86]). Evidence from<br />
the RESIDE project in Australia also contradicted the New Urbanism hypothesis. The RESIDE<br />
project aimed at evaluating the impact <strong>of</strong> land use mix among other urban design factors on<br />
walking for recreation and transport. It was a five-year longitudinal study that used a baseline<br />
survey completed by home owners before and after moving from single-family estates in monozoned<br />
suburbs to mixed-use neighborhoods. The results underscored the process <strong>of</strong> self-selection<br />
and provided no evidence that residents became more physically active in mixed-use<br />
neigborhoods than other types <strong>of</strong> estates (Pikora et al [75]).<br />
Reid Ewing, on the other hand, argued that development patterns and especially land use mix<br />
affect travel patterns. By operationalizing various aspects <strong>of</strong> the built environment, Ewing<br />
compared the extremes <strong>of</strong> sprawl and compact mixed use developments. He emphasized that<br />
relatively higher densities and balanced mix <strong>of</strong> residential, commercial, and employment<br />
activities would render active travel modes such as walking, biking, and transit more viable<br />
(Ewing [27]). Such assertions were supported by various other studies that focused on<br />
disaggregate data measuring walking activity on the household and individual levels. A study <strong>of</strong><br />
Austin neigborhoods in Texas showed that mixed-use developments with relative proximity <strong>of</strong><br />
retail, employment, and residential uses generated almost twice the walking activity <strong>of</strong><br />
conventional suburbs (Handy [37]). The findings <strong>of</strong> this and other studies suggested that<br />
utilitarian trips to work, shopping, etc. were the source <strong>of</strong> overall differences in walking activity<br />
between mixed-use and conventional suburbs (Handy [40]; Saelens et al [83]; Rajmani et al [77];<br />
Handy et al [38], [39], [43]).<br />
A general sense <strong>of</strong> unease, however, characterizes the presentation <strong>of</strong> findings <strong>of</strong> various<br />
empirical investigations. Kevin Krizek [59] conducted a longitudinal study about land use and<br />
travel patterns <strong>of</strong> 6,144 households in Seattle, Washington. The study analyzed the changing<br />
patterns <strong>of</strong> travel behavior for families who moved from conventional suburbs to mixed-use New<br />
Urbanist neighborhoods. The findings generally indicated that changes in urban form and land<br />
use triggered changes in modal split. Overall vehicle miles traveled (VMT) were decreased after<br />
moving to mixed-use environments. A closer look at the data revealed that such VMT reduction<br />
was not universal and it was mainly generated by households who were previously active and<br />
used transit even before they moved to New Urbanist communities (Krizek [59]).<br />
93
MOHAMMED KASHEF<br />
3.2 Walkability and Place Attachment<br />
Walkability here refers to non-work related physical activity that does not necessarily rely on<br />
mixed land use configurations that allow people to walk from home to work or major shopping<br />
centers and vice versa. It is more about walking for leisure purposes, recreation (parks and<br />
playgrounds), marginal shopping errands (buying a Popsicle), and/or exercise. Place attachment<br />
is a complex and pertinent construct that figured in most design and planning literature as both a<br />
by-product <strong>of</strong> and contributor to walking and overall human presence in the public sphere.<br />
Walking in that sense does not contribute much to reducing VMT; it does not substitute for<br />
automobile driving but may have positive impact on people’s daily physical activity levels and<br />
overall health. Place attachment is <strong>of</strong>ten described as a “positive affective relationship between<br />
people and place” that occurs because <strong>of</strong> people’s satisfaction and identification with a specific<br />
place. Identification with a place results in emotional connections with its physical attributes that<br />
people <strong>of</strong>ten associate with social events. Place and associated emotions and events become<br />
inseparable or interchangeable in people’s minds. The meaning <strong>of</strong> place here encompasses built<br />
environments as well as natural and cultural landscapes.<br />
Designers have generally maintained that certain spatial configurations, built forms, and design<br />
elements can play a significant role in promoting walkability, safety, and place attachment. The<br />
most significant design theme arising from the design and planning literature and potentially<br />
affecting neighborhood walkability and sociability is the interface between the private and public<br />
realms (Kelbaugh [57]; Jacobs [53]; Alexander [4]; Duany et a [25]). It is essentially about the<br />
articulation <strong>of</strong> the outer skin <strong>of</strong> buildings or homes and disposition <strong>of</strong> buildings in space. Various<br />
studies indicate that increasing the interface between private and public realms tends to heighten<br />
the level <strong>of</strong> activity around the edges <strong>of</strong> buildings. From a residential design perspective,<br />
elements that contribute to increased private/public interface include windows <strong>of</strong> active living<br />
spaces, transitional spaces such as entryways or porches, as well as the availability and<br />
distribution <strong>of</strong> common spaces. As long as home privacy is maintained, the presence <strong>of</strong> such<br />
elements creates opportunities for walking and social interaction among inhabitants (Duany et al<br />
[25]).<br />
The interaction referenced here can be characterized as passive or active. The passive social<br />
interaction falls under a category that Bill Hillier calls ‘social or spatial co-presence.’ Hillier<br />
contends that co-presence and therefore co-awareness amongst the individuals living in and<br />
passing by an area can be considered a raw material for social interaction. Whilst co-present<br />
individuals are not a community and may not know or acknowledge each other, they in fact form<br />
a social resource for communal behavior. Co-presence is a very important psychological resource<br />
in and <strong>of</strong> itself and can be brought about by design (Hillier [51]). Jane Jacobs’s ‘eyes on the<br />
street’ and Newman’s ‘defensible space’ have earlier noted the significance <strong>of</strong> home windows<br />
and visible entrance doors in providing surveillance and self-policing <strong>of</strong> public spaces (Jacobs<br />
[53]). Active interaction is commonly perceived as a by-product <strong>of</strong> increasing the pattern <strong>of</strong><br />
‘probabilistic interfaces,’ a heightened sense <strong>of</strong> co-awareness or co-presence, leading to overt<br />
social acknowledgment (Hillier [51]: 129-214). Transitional spaces such as home porches<br />
provide a setting for conversation, a kind <strong>of</strong> ‘Shakespearean’ theatrical platform for social ‘play.’<br />
Porches project the human presence within the house to passersby, creating a sense <strong>of</strong> ‘plot’; a<br />
voluntary and potentially entertaining social exchange (Figs. 2, 3). The availability <strong>of</strong> the home<br />
94
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
porch is significant whether a conversational exchange takes place or not. The porch acts as a<br />
precondition or a ‘stage’ for a social play that otherwise would not have been possible.<br />
Figure 2. A single family home with a porch fronting a<br />
neighborhood sidewalk, New Bern inner suburbs, North<br />
Carolina (built in the 1940s).<br />
Figure 3. A single family home in a modern suburb,<br />
New Bern, North Carolina (built in 2006).<br />
Another important dimension <strong>of</strong> the notion <strong>of</strong> walkabilty and sociability <strong>of</strong> built environments,<br />
and a factor in the private/public interface, is the idea <strong>of</strong> closeness or compactness. The New<br />
Urbanism called for shrinking private space and positioning homes closer to the curbside.<br />
Reducing the distance between home porches and sidewalks contributes to the ‘Shakespearean’<br />
plot and may enhance passive (co-awareness) and potentially active social exchanges between<br />
residents and passersby. New Urbanism codes typically prescribe a 15-20 feet optimum distance<br />
between sidewalks and porches to facilitate social encounters (Duany [25]). Reducing the<br />
physical footprint <strong>of</strong> buildings within a given development would also free more space for public<br />
activities. Public places in the form <strong>of</strong> parks and/or civic facilities such as picnic shelters, park<br />
gazebos, playgrounds, sports fields, etc. provide venues for chance encounters and serve as<br />
symbols <strong>of</strong> places (Langdon [60]). The urban design literature in general, and the New Urbanism<br />
praxes in particular, recognize sense <strong>of</strong> place as a function <strong>of</strong> sense <strong>of</strong> space which requires<br />
heightened attention to spatial configurations, design details, and disposition <strong>of</strong> public functions.<br />
The planners, designers, and New Urbanists’ case regarding the impact <strong>of</strong> spatial form and design<br />
details on human behavior is theoretically supported by environmental psychology and/or<br />
environmental sociology studies. Individuals associated with the latter strand <strong>of</strong> research aim to<br />
develop a body <strong>of</strong> knowledge that would enable designers, engineers, and other pr<strong>of</strong>essionals to<br />
envision optimum physical environments that enhance human functions and social relations.<br />
Some environmental sociology studies showed statistical evidence that design and spatial<br />
allocation <strong>of</strong> homes within residential blocks can have an impact on residents’ social relations<br />
and interactions. They asserted that facilitating people closeness by proper arrangement <strong>of</strong> private<br />
and shared public spaces enhances social contact and group/place attachment. Emily Talen [89]<br />
referred to housing studies in Chicago and Pittsburgh that highlight the role <strong>of</strong> the neighborhood<br />
physical environment in shaping residents’ social networks and interactions. Other researchers<br />
95
MOHAMMED KASHEF<br />
have found a direct correlation between home design and family interaction. Strategically located<br />
kitchens and living spaces tend to foster interaction and provide more opportunities for<br />
socialization among family members.<br />
John Peponis et al [74] have also pointed to direct and statistically supported evidence that <strong>of</strong>fice<br />
layout can contribute to the density <strong>of</strong> different network interaction and productivity. Integrated<br />
<strong>of</strong>fice layouts establish patterns <strong>of</strong> informal co-awareness and interaction, thus facilitating<br />
communication, sharing <strong>of</strong> ideas, and the joint exploration <strong>of</strong> possibilities in the non-routine<br />
phases <strong>of</strong> knowledge work. Other recent planning studies provided some credence to the New<br />
Urbanism claims regarding the impact <strong>of</strong> neighborhood design on non-work walking activity.<br />
Among the most important variables predicting a difference in walking was a change in the<br />
attractiveness <strong>of</strong> built environment. All else equal, people walked more after moving to visually<br />
pleasing neighborhood environments with variety <strong>of</strong> housing styles, well-maintained sidewalks<br />
and landscape, better street lighting, and more opportunities for social interaction. Such changes<br />
did not necessarily translate to less driving; it mainly impacted leisure-based walking (Handy<br />
[42], [43]). These results were corroborated by various other studies inside the U.S. and other<br />
cities in Australia. Design aesthetics, perceived safety levels, sidewalk treatments, trees, and the<br />
overall quality and variety <strong>of</strong> buildings were conducive to increased walkability and sociability<br />
(Rajmani et al [77]; Doyle et al [23]).<br />
3.3 Space Syntax and Connectivity<br />
Bill Hillier and Julienne Hanson probably coined the term ‘space syntax’ which delineates a set<br />
<strong>of</strong> spatial concepts and principles that underlie human interaction with the built environment<br />
(Hiller et al [52]). The aim <strong>of</strong> space syntax is to optimize the spatial allocations and connections<br />
between human functions so as to produce the most effective configurations and circulation<br />
networks. The assumption is that movement through space is a key organizing element in the<br />
social functioning <strong>of</strong> human settlements. Hillier contends that socioeconomic forces shape the<br />
city primarily through the relations between movement and the structural layout <strong>of</strong> local districts.<br />
A mixed-use strategy that brings homes closer to retail cannot generate movement unless shops<br />
are selectively located on integrating lines that carry the most movement in relation to<br />
surrounding spaces (Hillier [50], [51]). Hillier’s work sparked the development <strong>of</strong> a large volume<br />
<strong>of</strong> research around the world that used space syntax logic to examine spatial configurations <strong>of</strong><br />
homes, <strong>of</strong>fices, neigborhoods, cities, metropolitan regions or movement networks <strong>of</strong> entire<br />
countries (Baran et al [9]; Hillier [49], [48]; Bafna [7]; Jiang et al [54]; Nophaket et al [71];<br />
Raford et al [76]; Peponis et al [74]). There are some contentious issues surrounding the<br />
application <strong>of</strong> space syntax tools to the real world. Space syntax relies primarily on twodimensional<br />
mapping and geometrical techniques that discard vertical and horizontal dimensions<br />
as critical components <strong>of</strong> spatial structures (Ratti [78]; Hillier [46]). A detailed investigation <strong>of</strong><br />
the space syntax and related contentious issues is outside the scope <strong>of</strong> the current paper. The<br />
objective here is to highlight the value <strong>of</strong> the space syntax approach in understating the dynamics<br />
<strong>of</strong> street network design and walkability. A neighborhood space syntax analysis starts with an<br />
axial representation <strong>of</strong> its street layout. The axial map shows the neighborhood streets as a series<br />
<strong>of</strong> intersecting lines creating a network <strong>of</strong> links and nodes. Each street or line on the map has a<br />
different level <strong>of</strong> connectivity with the network based on the number <strong>of</strong> other streets that intersect<br />
it. Streets that generate more intersections hold higher value as elements <strong>of</strong> integration in the<br />
96
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
network. Important functions that tend to catalyze movement such as parks, shops, or other<br />
public amenities should be located on streets with the highest integration value in the network.<br />
Cumulatively, networks with more open (three- or four-way) street intersections create higher<br />
connectivity values (a grid system is the optimum).<br />
Connectivity indices measure the directness and availability <strong>of</strong> alternative routes between<br />
different locations within the road network. They are commonly used to assess the walkability<br />
potential <strong>of</strong> street networks. A connectivity index is commonly derived by dividing the number<br />
<strong>of</strong> roadway links by the number <strong>of</strong> roadway intersections and nodes (cul-de-sacs included). A<br />
higher number indicates a well-connected street system (Ewing [28]). Figures 4 & 5 juxtapose<br />
two hypothetical subdivision layouts with different connectivity potential. A link in the<br />
connectivity index is simply a stretch <strong>of</strong> road between two nodes. A node ranges from a dead-end<br />
cul-de-sac to a three- or-four-way intersection. The more links available in comparison to the<br />
number <strong>of</strong> nodes, the higher the connectivity and the potential <strong>of</strong> the street network to generate<br />
more pedestrian movement (Schmidt et al [85]). Various studies indicated that smaller blocks<br />
(less than 1000 feet long) and higher connectivity indices tend to generate more pedestrian<br />
movement (Ewing [28]; Southworth [88]; Krizek [59]; Cerin et al [18]; Frank et al [33]; Badland<br />
et al [6]; Chin et al [20]).<br />
Figure 4.<br />
Typical Subdivision Tract with Poor Connectivity<br />
Connectivity Index: Roadway Links/Nodes: 8/9 0.89<br />
Subdivision Statistics:<br />
No. <strong>of</strong> lots 84; average lot size 10,000 square feet<br />
No. <strong>of</strong> external access points 1; no. <strong>of</strong> dead-end<br />
streets 5<br />
Poor internal connectivity: Pedestrians have to<br />
backtrack from dead-end streets to navigate<br />
different areas <strong>of</strong> the subdivision. Roads do not<br />
encourage walking or biking.<br />
Poor external connectivity: There is a single<br />
access point on the collector road and no<br />
possibilities for alternative connectors with neighboring subdivisions.<br />
Figure 5.<br />
Typical Subdivision Tract Outfitted with Good<br />
Connectivity<br />
Connectivity Index: Roadway Links/Nodes: 15/10 =<br />
1.5<br />
Subdivision Statistics:<br />
No. <strong>of</strong> lots 84; average lot size 10,000 square feet<br />
<br />
11<br />
97
MOHAMMED KASHEF<br />
<br />
<br />
<br />
No. <strong>of</strong> external access points 4; no. <strong>of</strong> dead-end streets 1 (less than 200 feet road stretch)<br />
Good internal connectivity: Pedestrians can navigate different areas <strong>of</strong> the subdivision along a<br />
continuous loop and small block sizes that encourage walking and biking.<br />
Good external connectivity: 1 main access along the collector road and 3 other alternative<br />
accesses that enhance connectivity with neighboring subdivisions and encourage walking and<br />
biking.<br />
Increasing street connectivity would mean better intra- and-inter-neighborhood accessibility not<br />
only for pedestrians and cyclers, but also for cars. In that sense, neighborhoods would be<br />
connected by internal streets that allow low-speed travel across a large number <strong>of</strong> neigborhoods<br />
in suburban regions. This could potentially reduce congestion along arterial roads and provide<br />
more route options when some links are blocked by traffic. Despite concerns about safety and<br />
increased car travel along suburban roads, transportation studies indicated that improved<br />
connectivity in suburban regions would provide a much needed flexibility <strong>of</strong> movement,<br />
especially for emergency vehicles. Building some redundancy in road networks is critical to their<br />
long-term viability and ability to generate a rich modal split between motorized and<br />
nonmotorized forms <strong>of</strong> transportation. Pedestrians and cyclers would be encouraged to travel<br />
along low-speed internal suburban roads rather than venturing into high-speed regional<br />
transportation corridors (Crane [21]; Ewing et al [29]; Handy et al [41]; Langdon [60]; Alba et al<br />
[3]). The findings <strong>of</strong> the Atlanta and Central Puget Sound traffic studies showed that doubling<br />
suburban road intersections would decrease regional vehicle miles traveled by approximately five<br />
to ten percent. Such decrease was attributed to the additional route options, shorter distances<br />
between origins and destinations as well as a better modal split between driving, walking, and<br />
biking (SMARTRAQ [87]; LUTAQH [66])<br />
3.4 Summarizing the Results <strong>of</strong> Planning, Design, and Transportation Studies<br />
Some transportation and planning studies corroborated the findings <strong>of</strong> health studies regarding<br />
the role <strong>of</strong> self-selection process in explaining the difference in walkability between conventional<br />
suburbs and mixed-use neigborhoods; people seem to select themselves into neigborhoods that<br />
support their lifestyle. These studies generally doubted the efficacy <strong>of</strong> the New Urbanism design<br />
techniques in changing people’s attitudes and encouraging a more active lifestyle. Planning<br />
studies, however, implicated zoning and government policies in entrenching the suburban ideal<br />
and creating car-dependent, low-density suburbs. Design and planning narratives criticized the<br />
exclusionary practices <strong>of</strong> Euclidean zoning. In that sense, zoning compartmentalized land uses,<br />
reduced densities, and eliminated functional overlaps that used to define vibrant human<br />
settlements. The ensuing metropolitan form was riddled with social and spatial mismatches<br />
between housing, commercial, and employment activities. Zoning and its attendant financial<br />
policies and infrastructure projects resulted in unsustainable urban growth with restricted<br />
individual options and inflexible development patterns. Two significant themes figured<br />
prominently in planning and design analyses. The first theme was the nexus between walkability<br />
and place attachment. Statistically supported investigations within environmental psychology and<br />
sociology asserted that good design can be used to bolster the experiential quality <strong>of</strong> built<br />
environments. Design provides possibilities for social interaction that enriches people’s lives and<br />
enhances their sense <strong>of</strong> place. Identification with a place results in emotional connections with its<br />
physical attributes. Homes, windows, sidewalks, porches, landscape, parks, etc. become critical<br />
98
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
correlates <strong>of</strong> people’s presence in the public sphere. The second theme was space syntax and<br />
connectivity. Spatial relations, not architecture per se, define the ability <strong>of</strong> built environments to<br />
generate effective patterns <strong>of</strong> movement. Neighborhood walkability gets depleted once<br />
movement networks are compartmentalized through dead-ends and reduced intersections in the<br />
name <strong>of</strong> exclusivity and efficiency. Integrated spatial networks, higher street intersection<br />
densities, and supportive architectural configurations are all crucial for creating walkable<br />
neighborhoods. The crux <strong>of</strong> the planning and design analyses is that cities should create welldefined<br />
interfaces between buildings and streets and between localized movement in<br />
neighborhood streets and the more globalized pattern <strong>of</strong> movement in metropolitan areas.<br />
4. SYNTHESIS AND FUTURE RESEARCH DIRECTIONS<br />
Though pointing to positive associations between built environment and walkability, the<br />
empirical evidence discussed in this paper is not sufficient to frame a theory <strong>of</strong> built environment<br />
and human behavior. It did not provide a definitive answer one way or the other to the<br />
relationship between walkability and built environment. The available evidence can hardly be<br />
used as a firm basis for policy guidance. However, juxtaposing the results <strong>of</strong> studies from<br />
environmental health and planning, design, and transportation fields revealed some<br />
inconsistencies in the way attributes <strong>of</strong> built environment were operationalized and measured in<br />
health and planning empirical investigations. The qualitative aspects <strong>of</strong> built environments<br />
highlighted in the design literature were hardly translated to quantifiable variables in much <strong>of</strong><br />
these empirical studies. For example, the issue <strong>of</strong> sidewalks was mostly operationalized in a<br />
binary ‘0/1’ mode, available or unavailable. The design and space syntax literature indicated that<br />
there are critical configurational and spatial relationships that must exist before physical design<br />
elements can achieve their desired behavioral objectives. Sidewalks need to be located near home<br />
fronts to enhance the private/public interface. Home architecture in return should encompass<br />
transitional spaces that project human presence to the public space, e.g. porches and visible<br />
entryways. Windows <strong>of</strong> active spaces are also crucial to enhance these configurational and<br />
sociospatial dynamics. Other important design values include visual interest, coherence,<br />
enclosure, legibility, and variety. None <strong>of</strong> the reviewed statistically-driven studies translated such<br />
qualities into measurable variables. It is worth noting that Ewing et al [30] developed operational<br />
definitions and measurement protocols for such qualitative urban design values as imageability,<br />
enclosure, human scale, transparency, and complexity. These measurement protocols were not<br />
utilized or tested in the most recent empirical studies.<br />
The connectivity index used in various empirical investigations is another example <strong>of</strong> such<br />
reductionist statistical measures. The notion that street networks have more four-way<br />
intersections as well as more links than nodes does not say anything about the integrity <strong>of</strong> the<br />
road and block system, the distribution <strong>of</strong> key public functions within the street network, or the<br />
overall synergy achieved between form and function. The connectivity index is a very effective<br />
measure for studying the flexibility <strong>of</strong> road networks for motorized traffic but not an<br />
encompassing measure <strong>of</strong> intention-based and/or environment-provoked walking behavior. It is<br />
definitely a helpful measure, but may not supplant for a qualitative multi-dimensional<br />
understanding <strong>of</strong> the integration between local and regional road networks and disposition <strong>of</strong><br />
buildings and uses in space. It may not also be used as a unique measure to establish a cause and<br />
effect relationship between built environment and walking behavior.<br />
99
MOHAMMED KASHEF<br />
The New Urbanism literature points to the need for creating a synergy between various design<br />
qualities in order to produce walkable and vital built forms. A piecemeal or reductionist approach<br />
that utilizes some elements and removes others may not achieve the desired objectives. The<br />
position taken here is empirically based and calls for internalizing qualitative urban design values<br />
and translating them into measurable protocols. Reductionist measures employed in many<br />
empirical studies should be replaced by rich protocols that benefit from the New Urbanism<br />
approach to understanding human settlements. Implicit in many <strong>of</strong> the recent empirical<br />
investigations is that the New Urbanism developments embody most or all <strong>of</strong> the good urban<br />
design principles that may enhance walkability. The New Urbanists have certainly succeeded to<br />
create that image. However, New Urbanist developments manifest a variety <strong>of</strong> approaches and<br />
selective inclusion <strong>of</strong> design principles in primarily pr<strong>of</strong>it-seeking development projects. This is<br />
not something that runs contrary to the movement mission. The New Urbanism aims to achieve<br />
its design objectives through private market dynamics. Thus, including New Urbanist<br />
developments for statistical and comparative analyses should be accompanied by in-depth,<br />
critical evaluation and assessment <strong>of</strong> their design and spatial morphology.<br />
Despite its current limitations, the space syntax approach holds a great potential for<br />
understanding morphological structures and built configurations. It goes beyond the simplistic<br />
binary characterizations <strong>of</strong> mixed-use/single-use, available/unavailable sidewalks, or<br />
connected/disconnected built environments. Space syntax analysis was used in some empirical<br />
investigations as a two-dimensional mapping technique to assess connectivity <strong>of</strong> road systems.<br />
But Hillier’s analyses highlighted a comprehensive approach to understanding human settlements<br />
that goes beyond two-dimensional mapping. It tackled movement through space, spatial<br />
allocation <strong>of</strong> important public functions, and configurational relationships between buildings and<br />
streets. Bringing homes closer to retail cannot generate movement unless shops are selectively<br />
located on integrating lines that carry the most movement in relation to surrounding spaces. This<br />
is critical for applying space syntax and understanding the nexus between built environment and<br />
walkability. Mixed-use was incorporated in some <strong>of</strong> the reviewed empirical investigations as one<br />
<strong>of</strong> the determinants <strong>of</strong> walkability. Critical evaluations <strong>of</strong> spatial relationships between homes<br />
and shops were generally lacking. This paper takes the position that integrating two- and-threedimensional<br />
metrics with space syntax analyses holds a great potential for future research aimed<br />
at understanding the complex relationship between built forms and human living patterns. The<br />
quest for understanding form and behavior would also require a concerted effort to develop<br />
comprehensive measurement protocols for a variety <strong>of</strong> qualitative urban design values.<br />
5. REFERENCES<br />
[1] Addy CL, Wilson DK, Kirtland KA, Ainsworth BE, Sharpe P, Kimsey D. Association <strong>of</strong><br />
perceived social and physical environmental supports with physical activity and walking behavior.<br />
Am J Public Health. 2004; 94:440-443.<br />
[2] Ainsworth B. E., Wilcox S., Thompson W. W., Richter D. L., Henderson K. A. (2003) Personal,<br />
social and physical environmental correlates <strong>of</strong> physical activity in African American women in<br />
South Carolina. Am Journal <strong>of</strong> Preventive Medicine, 25 (3, Supplement 1): 23-29.<br />
[3] Alba, C. A., and Beimborn, E. (2005) Analysis <strong>of</strong> the effects <strong>of</strong> local street connectivity on<br />
arterial traffic, Transportation research Board Annual Meeting, www.trb.org.<br />
100
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
[4] Alexander, C. (1977) A pattern Language: Towns, Buildings, Construction. Oxford University<br />
Press, New York<br />
[5] Atkinson, J. L., Sallis, J. F., Saelens, B. E., Cain, K. L., and Black, J. B. (2004) The association <strong>of</strong><br />
neigborhood design and recreational environments with physical activity, American Journal <strong>of</strong> Health<br />
Promotion, 19 (4): 304-309.<br />
[6] Badland, H. M., Sch<strong>of</strong>ield, G. M., and Garrett, N. (2008) Travel behavior and objectively<br />
measured urban design variables: Associations for adults traveling to work, Health & Place, 14: 85–<br />
95.<br />
[7] Bafna, S. (2003) Space syntax: a brief introduction to its logic and analytical techniques,<br />
Environment and Behavior, 35 (1): 17-29.<br />
[8] Bagley, M. N., and Mokhtarian P. L. (2002) The impact <strong>of</strong> residential neighborhood type on travel<br />
behavior: A structural equations modeling approach, The Annals <strong>of</strong> regional Science, 36:279–297.<br />
[9] Baran, P. K., Rodriguez, D. A., and Khattak, A. J. (2008) Space syntax and walking in a New<br />
Urbanist and suburban neighborhoods, Journal <strong>of</strong> Urban Design, 13: (1) 5-28.<br />
[10] Barnette, J. (1995) The Fractured Metropolis: Improving the New City, Restoring the Old City,<br />
Reshaping the Region. Harper Collins, New York.<br />
[11] Boehmer, T. K., Hoehner, C. M., wyrwich, K. W., Ramirez, L. K., and Brownson, R. C. (2006)<br />
Correspondence between perceived and observed measures <strong>of</strong> neigborhood environmental supports<br />
for physical activity. Journal <strong>of</strong> Physical Activity and Health, 3: 22-36.<br />
[12] Boer, R., Zheng, Y., Overton, A., Ridgeway, G. K., and Cohen, D. A. (2007) Neighborhood<br />
design and walking trips in ten U.S. metropolitan areas, American Journal <strong>of</strong> Preventive Medicine, 32<br />
(4): 289-304.<br />
[13] Bogart, W. (1998) The Economics <strong>of</strong> Cities and Suburbs, Prentice Hall: New Jersey.<br />
[14] Bourne L. S. (1996), "Reinventing the Suburb: Old Myths and New Realities", Process in<br />
Planning, 46:3.<br />
[15] Burchell W. D., Listokin, D., Galley C. C. (2000) Smart growth: more than a ghost <strong>of</strong> urban<br />
policy past, less than a bold new horizon, Housing Policy Debate, 11 (4): 821-879.<br />
[16] Calthorpe P, Richmond H (1992) Land use and transportation planning: Sustainable growth. In:<br />
Green M (ed) Changing America: Blueprints for the new administration. Newmarket Press,<br />
New York, pp 699–710<br />
[17] Calthorpe, P. (1993) The Next American Metropolis: Ecology, Community and the American<br />
Dream, Princeton Architectural Press, New York.<br />
[18] Cerin, E., Leslieb, E., Toit, L. D., Owen, N., and, Frank, L. D. (2007) Destinations that matter:<br />
Associations with walking for transport, HEALTH & PLACE, 13: 713-724.<br />
[19] Cervero, R. and Duncan M. (2003) Walking, bicycling, and urban landscapes: evidence from the<br />
San Francisco Bay Area, American Journal <strong>of</strong> Public Health, 93 (9): 1478-1483.<br />
[20] Chin, G. K.,Van Niel, K. P., Corti, B. G., and Knuiman, M. (2007) Accessibility and<br />
connectivity in physical activity studies: The impact <strong>of</strong> missing pedestrian data, Preventive Medicine,<br />
46: 41-45<br />
[21] Crane, R., (2000) The influence <strong>of</strong> urban form on travel: an interpretive review, Journal <strong>of</strong><br />
Planning Literature, 15 (1): 3-23.<br />
[22] DiPietro L. (2001) Physical activity in aging: changes in patterns and their relationship to health<br />
and function. Journal <strong>of</strong> Gerontology and Biological and Medical Science, 56(spec no. 2):13–22.<br />
[23] Doyle, C., Schwartz, A. K., Schlossberg, M., and Stockard, J. (2006) Active communities:<br />
environments and health, Journal <strong>of</strong> the American Planning Association, 72 (1): 19-31.<br />
[24] Duany, A. & Z. Plater (1991) Towns and Town-Making Principles, Rizzoli, New York.<br />
[25] Duany, A. and Plater-Zyberk, E. (2000) Suburban Nation: the Rise <strong>of</strong> Sprawl and the Decline <strong>of</strong><br />
the American Dream, North Point Press, New York.<br />
101
MOHAMMED KASHEF<br />
[26] Dutton, J. A. (2000) New American Urbanism: Re-forming the Suburban Metropolis, Skira<br />
Architecture Library, Milano.<br />
[27] Ewing, R. (2004) Can the physical environment determine physical activity levels? Exercise and<br />
Sport Sciences Reviews, 33 (2): 69-75.<br />
[28] Ewing, R. (1996) Best development practices: doing the right thing and making money at the<br />
same time, Planners Press (www.planning.org).<br />
[29] Ewing, R. and Cervero, R. (2002) Travel and the built environment: synthesis, Transportation<br />
research record 1780 (TRB), www.trb.org.<br />
[30] Ewing, R., Handy, S., Brownson, R. C., Clemente, O., and Winston, E. (2006) Identifying and<br />
measuring urban design qualities related to walkability, Journal <strong>of</strong> Physical Activity and Health, 3<br />
(1): s223-s240.<br />
[31] Gratz R. B. & Norman M. (1998) Cities Back From the Edge: New Life for Down town, John<br />
Wiley & Sons, New York.<br />
[32] Fleissig, W., and Jacobsen, V. (2002) Smart Scorecard for development projects, In<br />
Collaboration with the Congress for New Urbanism and the U.S. Environmental Protection Agency<br />
(http://www.cnu.org/cnu_reports/scorecard_exp.pdf)<br />
[33] Frank, L. D., Schmid, T. L., Sallis, J. F. Chapman, J. and Saelens, B. E. (2005) Linking<br />
objectively measured physical activity with objectively measured urban form: findings from<br />
SMARTRAQ, American Journal <strong>of</strong> Preventive Medicine, 28 (2S2): 117-125.<br />
[34] Friedman, B., Gordon, S. P. (1994) Effect <strong>of</strong> Neotraditional Neighborhood Design on Travel<br />
Characteristics, Transportation Research Record 1466, TRB, National Research<br />
Council, Washington, D.C., 63–70.<br />
[35] Fulton W (1996) The new urbanism. Lincoln Institute <strong>of</strong> Land Policy, Cambridge, MA.<br />
[36]Geus, B. D., Bourdeaudhuij, I. D., Jannes, C., and Meeusen, R. (2008) Psychological and<br />
environmental factors associated with cycling for transport among a working population, HEALTH<br />
EDUCATION RESEARCH, 23 (4) 697-708.<br />
[37] Handy, S. L. (1996) Urban form and pedestrian choices: a study <strong>of</strong> Austin neighborhoods,<br />
Transportation Research Record 1552, National Research Council, Washington,<br />
D.C., TRB: 135-144,<br />
[38] Handy, S. L., and Clifton, K. J. (2001) Local shopping as a strategy for reducing automobile<br />
travel, Transportation, 28: 317-364.<br />
[39] Handy, S. L., Boarnet, M. G., Ewing, R., and Killingworth, R. E. (2002) How the built<br />
environment affects physical activity: views from urban planning, American <strong>journal</strong> <strong>of</strong> Preventive<br />
Medicine, 23(2S): 64-73.<br />
[40] Handy, S. L. (2004) Critical assessment <strong>of</strong> the literature on the relationships among<br />
transportation, land use, and physical activity. Department <strong>of</strong> environmental Science and Policy,<br />
University <strong>of</strong> California, Davis. Prepared for the Committee on Physical Activity, Health,<br />
Transportation, and Land Use.<br />
[41] Handy, S., Paterson, R. G., and Butler, K. (2004) Planning for street connectivity: getting from<br />
here to there, Planning Advisory Service report 515, American Planning Association,<br />
www.planning.org.<br />
[42] Handy, S. L. (2005) Which comes first: the neighborhood or the walking? ACCESS, 26: 16-21.<br />
[43] Handy, S. L. (2006) Questioning assumptions, Planning, January, 2006: 36-37.<br />
[44] Handy, S. L., Cao, X., and Mokhtarian P. (2006) Self-selection in the relationship between the<br />
built environment and walking, Journal <strong>of</strong> the American Planning association, 72 (1): 55-74.<br />
102
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
[45] Handy, S. L Cao, X., and Mokhtarian, P. L. (2008) The causal influence <strong>of</strong> neighborhood design<br />
on physical activity within the neighborhood: evidence from Northern California, American Journal<br />
<strong>of</strong> Health Promotion, 22 (5): 350-358<br />
[46] Hillier, B. and Penn, A. (2004) Rejoinder to Carlo Ratti, Environment and Planning B; Planning<br />
and Design, 31: 501-511.<br />
[47] Hiller, B., and Netto, V. (2002) Society seen through he prism <strong>of</strong> space: outline <strong>of</strong> a theory <strong>of</strong><br />
society and space, URBAN DESIGN International, 7: 181-203.<br />
[48] Hillier, B. (2002) A theory <strong>of</strong> the city as object: or, how spatial laws mediate the social<br />
construction <strong>of</strong> urban space, URBAN DESIGN International, 7: 153-179.<br />
[49] Hillier, B. (1999) The hidden geometry <strong>of</strong> deformed grids, Environment and Planning B, 26:<br />
169-191.<br />
[50] Hiller, B. (1997) Cities as movement economies, in: P. Droedge (Ed.) Intelligent Environments,<br />
Elsevier, Amsterdam.<br />
[51] Hillier B. (1996) Space is the Machine: A Configurational Theory <strong>of</strong> Architecture, Cambridge<br />
University Press, Cambridge.<br />
[52] Hillier B. and J. Hanson (1984) The Social logic <strong>of</strong> Space, Cambridge University Press, London<br />
[53] Jacobs, J. (1961) The Death and Life <strong>of</strong> Great American Cities, Random House, Inc. New York<br />
[54] Jiang, B. and Claramunt, C. (2002) Integration <strong>of</strong> space syntax into GIS: new perspectives for<br />
urban morphology, Transactions in GIS, 6 (3): 295-309.<br />
[55] Kashef, M. (2008) Architects and planners approaches to urban form and design in the Toronto<br />
region: a comparative analysis, Ge<strong>of</strong>orum 39 (1): 414-37<br />
[56] Katz P. (1994) The New Urbanism: Toward an Architecture <strong>of</strong> Community, MacGraw-Hill, New<br />
York<br />
[57] Kelbaugh D. (1997) Common Place: Toward Neighbourhood and Regional Design, University<br />
<strong>of</strong> Washington Press, Washington.<br />
[58] Krier, R. (1979) Urban Space, Rizzoli, New York<br />
[59] Krizek, K. J. (2003) Operationaliz.ing neighborhood accessibility for land use-travel behavior<br />
research and modeling. Journal <strong>of</strong> Planning Education and Research, 22(3): 270-28.<br />
[60] Langdon, P. (2004) Street connectivity: emerging as a new municipal goal, New Urban News,<br />
New Urban Publications, www.newurbannews.com, 21-24.<br />
[61] Lee, C. and Moudon, V. (2006) Correlates <strong>of</strong> walking for transportation or recreation Purposes,<br />
Journal <strong>of</strong> Physical Activity and Health, 3 (Suppi 1): S77-S98.<br />
[62] Lesliea, E., saelensb, B., Frankc, L., Owen, N., Baumand, A., C<strong>of</strong>feee, N., and Hugo, G. (2005)<br />
Residents’ perceptions <strong>of</strong> walkability attributes in objectively different neighborhoods: a pilot study,<br />
Health & Place, 11: 227-236.<br />
[63] Leyden, K. M. (2003) Social capital and the built environment: the importance <strong>of</strong> walkable<br />
neigborhoods, American Journal <strong>of</strong> Public Health, 93 (9) 1546-1551.<br />
[64] Li, F., Fisher, J. K., Brownson, R. C., Bosworth, M (2005) Multilevel modeling <strong>of</strong> built<br />
environment characteristics related to neighbourhood walking activity in older adults, Journal <strong>of</strong><br />
Epidemiology and Community Health, 59: 558-564.<br />
[65] Lovasi, G. S., Moudon, A. V., Pearson, A. L., Hurviz, P. M., Larson, E. B., Siscovick D. S.,<br />
Berke, E. M., Lumley, T. and Psaty, B. M. (2008) Using built environment characteristics to predict<br />
walking for exercise, International Journal <strong>of</strong> Health Geographics, 7 (10).<br />
[66] LUTAQH Project (2006), Central Puget Sound, www.metrokec.gov.<br />
[67] McCormack, G. R., Cerin, E., Leslie, E., Du Toit, L., and Owen, N. (2008) Objective versus<br />
perceived walking distances to destinations: correspondence and predictive validity, Environment and<br />
Behavior, 40 (3): 401-425.<br />
103
MOHAMMED KASHEF<br />
[68] Michael, Y. L., Green, M. K., and Farquhar, S. A. (2006) Neigborhood design and active aging,<br />
HEALTH & PLACE, 12: 734-740.<br />
[69] Moudon, A. V., Dr. es Sc.; Lee, C., Cheadle, A. D., Garvin C., Johnson, D. B., Schmid, T. L.,<br />
Weathers, R. D. (2007) Attributes <strong>of</strong> environments supporting walking, American Journal <strong>of</strong> Health<br />
Promotion, 21 (5): 448-459.<br />
[70] Nagel, C. L., Carlson, N. E., Bosworth, M., and Michael, Y. L. (2008) The relation between<br />
neigborhood built environment and walking activity among older adults, American Journal <strong>of</strong><br />
epidemiology Advance Access.<br />
[71] Nophaket, N., and Fujii, A. (2004) Syntactic and network pattern structures <strong>of</strong> city: comparison<br />
<strong>of</strong> grid and meandering street patterns in Kyojima and Honjo, Journal <strong>of</strong> Asian Architecture and<br />
Building Engineering, 349-356.<br />
[72] Oaks, M. J., Forsyth, A., Shmitz, K. H. (2007) The effects <strong>of</strong> neigborhood density and street<br />
connectivity on walking behaviour: the twin cities walking study, Epidemiologic perspectives &<br />
Innovation, 4 (16).<br />
[73] O'Sullivan, A. (1996) Urban Economics, Irwin, Chicago.<br />
[74] Peponis, J., Bafna, S., Bajaj, R., Bromberg, J., Congdon, C., Rashid, M., Warmels, S., Zhang, Y.<br />
and Zimring, C. (2007) Designing space to support knowledge work, Environment and Behavior, 38:<br />
815-840.<br />
[75] Pikora, T., Corti, B. G., Knuiman, M., Niel, K. V., Bulsara, M., Bull, F., Shilton, T. (2007) The<br />
impact <strong>of</strong> urban design on walking and cycling: the RESIDE project, Planning and Transportation<br />
research Center (PATREC).<br />
[76] Raford, N., and Ragland, D. R. (2003) Space syntax: an innovative pedestrian volume modeling<br />
tool for pedestrian safety, Institute <strong>of</strong> Transportation Studies, UC Berkeley traffic Safety Center, UC<br />
Berkeley.<br />
[77] Rajmani, J., Bhat, C. R., Handy, S., Knaap, G., Song, Y. (2003) Assessing the impact <strong>of</strong> urban<br />
from measures in nonwork trip mode choice after controlling for demographic and level-<strong>of</strong>-service<br />
effects, Transportation Research Board (TRB), Annual Meeting CD-ROM.<br />
[78] Ratti, C. (2004) Space syntax: some inconsistencies, Environment and Planning B: Planning and<br />
Design, 31: 487-499.<br />
[79] Reed, J. A., Wilson, D. K., Ainsworth, B. E., Bowels, H., and Mixon, G. (2006) Perceptions <strong>of</strong><br />
neigborhood sidewalks on walking and physical activity patterns in a south eastern community in the<br />
US, Journal <strong>of</strong> Physical Activity and Health, 3: 242-253.<br />
[80] Rhodes R. E., Courneya K. S., Blanchard C. M., and Plotnik<strong>of</strong>f R. C. (2007) Prediction <strong>of</strong><br />
leisure-time walking: and integration <strong>of</strong> social cognitive, perceived environmental, and personality<br />
factors, International Journal <strong>of</strong> Behavioral Nutrition and Physical Activity, 4 (51).<br />
[81] Rowe C. and Koetter, F. (1978) Collage City, MIT, Cambridge<br />
[82] Sallis J. F., Melbourne H, and H<strong>of</strong>stetter C. R., et al (1990) Distance between homes and<br />
exercise facilities related to frequency <strong>of</strong> exercise among San Diego residents. Public Health<br />
Rep, 105:179-185.<br />
[83] Saelens, B. E., Sallis, J. F., and Frank, L. D. (2003) Environmental correlates <strong>of</strong> walking and<br />
cycling: findings from the transportation, urban design, and planning literatures, Environment and<br />
Physical Activity, 25: 80-91.<br />
[84] Saelens, B. E., Sallis, J. F., Black, J. B., and Chen D. (2003) Neighborhood-based differences in<br />
physical activity: and environment scale evaluation, American Journal <strong>of</strong> Public Health, 93 (9):<br />
1552-11558.<br />
[85] Schmidt, S. J., and Wells, J. S. (2005) Transit village monitoring research: connectivity<br />
measures, Alan M. Voorhees Transportation Center (VTC), Rutgers.<br />
104
NEIGHBOURHOOD DESIGN AND WALKABILITY<br />
[86] Schwanen, T., and Mokhtarian, P. L. (2007) Attitudes towards travel and land use and choice <strong>of</strong><br />
residential neigborhood type: evidence from the San Francisco Bay Area, Housing Policy Debate, 2<br />
(1): 171-207.<br />
[87] SMARTRAQ Project (2204), Atlanta, www.smartraq.net.<br />
[88] Southworth, M (2005) Designing the walkable city, Journal <strong>of</strong> Urban Planning and<br />
development, 131 (4): 246-257.<br />
[89] Talen, A (1999) Sense <strong>of</strong> community and neighborhood form: an assessment <strong>of</strong> the social<br />
doctrine <strong>of</strong> New Urbanism, Urban Studies, 36 (8): 1379-99.<br />
[90] Trancik, R. (1986) Finding Lost Space: Theories <strong>of</strong> Urban Design, Van Nostrand Reinhold, New<br />
York.<br />
[91] Wells, N. M., and Yang, Y. (2008) Neighborhood design and walking: a quasi-experimental<br />
longitudinal study, American Journal <strong>of</strong> Preventive medicine, 34 4): 313-319<br />
[92] Wood, L., Shannon, T., Bulsara, M., Pikora, T, McCormiack, G., and Corti, B. G. (2007) The<br />
anatomy <strong>of</strong> the safe and social suburb: an exploratory study <strong>of</strong> the built environment, social capital<br />
and residents’ perceptions <strong>of</strong> safety, HEALTH & PLACE, 14: 15-31<br />
105
106
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 107-118 (2010)<br />
© 2009 ALHOSN University<br />
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
P. Rengaraju, C-H. Lung 1 , A. Srinivasan 2 , R.H.M. Hafez 1*<br />
1 Department <strong>of</strong> Systems and Computer Engineering, Carleton University, Ottawa, Ontario, Canada<br />
2 EION Inc. Ottawa, Ontario, Canada<br />
ABSTRACT: In the last few years the usage <strong>of</strong> multimedia applications and VoIP using mobiles over the<br />
internet has drastically increased. Providing Quality <strong>of</strong> Service (QoS) for the end user applications has become<br />
an important consideration for the upcoming fourth generation WiMAX and LTE networks. Research is<br />
continuing in Physical (PHY) layer, Medium Access Control (MAC) layer and cross layers in order to improve<br />
the QoS for both WiMAX and 3GPP-LTE networks. This paper discusses fundamental issues in QoS<br />
improvements in WiMAX networks in each layer.<br />
Key words: IEEE 802.16; WiMAX; QoS; QoE<br />
1. INTRODUCTION<br />
WIMAX stands for Worldwide Interoperability for Microwave Access. It is the technology<br />
aimed to provide broadband wireless data access over long distances [5]. It is based on<br />
Institute <strong>of</strong> Electrical and Electronics Engineers (IEEE) 802.16 standards [1], [2] and the<br />
standard defines the PHY and MAC layer functionalities. The technology provides basic<br />
Internet Protocol (IP) connectivity to the user and the communication is connection oriented.<br />
The initial IEEE 802.16d standard is specific to fixed and nomadic users. Based on increasing<br />
mobile users requirements and to increase the coverage region, the standard released the IEEE<br />
802.16e and IEEE 802.16j [2] versions. Now the IEEE 802.16m [3], [4] task group are<br />
working to support the International Mobile Telecommunications (IMT) - Advanced<br />
requirements. One <strong>of</strong> the most controversial requirements <strong>of</strong> IEEE 802.16m, during the<br />
development <strong>of</strong> its system requirement document [4] was full backwards compatibility and<br />
interoperability with the legacy systems.<br />
The MAC and PHY functions can be classified into three categories, namely data plane,<br />
control plane and management plane. The data plane comprises functions in the data<br />
processing path such as header compression as well as MAC and PHY data packet processing<br />
functions. A set <strong>of</strong> L2 control functions is needed to support various radio resource<br />
configuration, coordination, signaling and management. This set <strong>of</strong> functions is collectively<br />
referred to as control plane functions. The individual data and control plane functionalities <strong>of</strong><br />
PHY and MAC layer is shown in Fig.1. A management plane is also defined for external<br />
management and system configuration. Therefore, all management entities fall into the<br />
management plane category [5]. Some <strong>of</strong> the MAC layer functional blocks are discussed in<br />
section III, MAC layer QoS. improvements. Relay functions and advanced air interface<br />
support functions are specific to 802.16j and 802.16m standard.<br />
________________________<br />
* Corresponding Author.<br />
E-mail : hafez@sce.carlton.ca<br />
107
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
The different types <strong>of</strong> PHY carriers are Single Carrier (SC) for 11 to 66GHz, Single Carrier<br />
access (SCa) for less than 11GHz, Orthogonal Frequency Division Multiplexing (OFDM) and<br />
Orthogonal Frequency Division Multiple Access (OFDMA). Most <strong>of</strong> the IEEE 802.16e based<br />
nodes, IEEE802.16j and IEEE 802.16m nodes uses only OFDMA. The QoS improvements on<br />
OFDMA carriers are sub channelization, Multiple Input Multiple Output (MIMO), Adaptive<br />
Antenna System (AAS) and Space Time Coding etc.<br />
The term QoS refers to the probability <strong>of</strong> the telecommunication network meeting a given<br />
traffic contract. The parameters that can be used to measure it quantitatively are throughput,<br />
transmission delay or packet delay, delay jitter, percentage <strong>of</strong> packets lost etc. On top <strong>of</strong> this,<br />
the International Telecommunication Union (ITU) recently released the Quality <strong>of</strong> Experience<br />
(QoE) framework ITU-T (2004) in which an explicit distinction has been made between QoS<br />
and QoE. The QoS expresses the ‘degree <strong>of</strong> objective service performance’ and QoE the<br />
‘overall acceptability <strong>of</strong> an application or service, as perceived subjectively by the end user’.<br />
From the last few years, many researches are on going to increase the capacity and improve<br />
the QoS <strong>of</strong> the network [5]. This paper analyses the QoS improvements in each layer and the<br />
cross layer.<br />
The rest <strong>of</strong> the paper is organized as follows. Section II and III surveys the QoS<br />
improvements in PHY and MAC layer. Section IV surveys end-to-end (E2E) QoS and cross<br />
layer improvements. The last section V is the conclusion <strong>of</strong> the Paper.<br />
2. PHY LAYER QOS IMPROVEMENTS<br />
There are two ways to improve the PHY layer QoS; 1. Increase the spectral efficiency <strong>of</strong><br />
available spectrum by advanced OFDMA and MIMO techniques. 2. Re-use <strong>of</strong> available<br />
frequency spectrum to increase the capacity. OFDMA demonstrates superior performance in<br />
non-line-<strong>of</strong>-sight (N-LOS) multi-path channels with its relatively simple transceiver structures<br />
and allows efficient use <strong>of</strong> the available spectrum resources by time and frequency sub<br />
channelization. On the other hand MIMO technology multiplies throughput without requiring<br />
additional frequency bandwidth, enhances link reliability through spatial diversity and<br />
enlarges the coverage area by increasing the transmission range [5]. The Partial Frequency<br />
Reuse (PFR) [11], [12] and Full Frequency Reuse (FFR) [10] can be <strong>applied</strong> to WiMAX<br />
network to increase the capacity <strong>of</strong> the network.<br />
108
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
MIMO Techniques<br />
Examples <strong>of</strong> MIMO techniques include single-user MIMO (SU-MIMO), multiuser MIMO<br />
(MU-MIMO), and cooperative relay [7]. Multiuser interference becomes a bottleneck for<br />
increasing multiuser MIMO throughput at high SNRs or for a relatively larger number <strong>of</strong><br />
multiuser MIMO users. This places a more stringent requirement on channel state information<br />
(CSI) accuracy for multiuser MIMO than that for point-to-point MIMO. To acquire highlyaccurate<br />
CSI without incurring excessive overhead, limited feedback can be integrated with<br />
multiuser diversity. MIMO schemes can be broadly divided into two categories; the ones<br />
which have and use CSI at the transmitter which is called closed-loop (i.e., beam forming<br />
and/or pre-coding) and the other without CSI at the transmitter (i.e. space-time coding (STC)<br />
and spatial rate multiplexing (SM)) which is called open-loop [6].<br />
Space–time Coding [6]: This technique is used to transmit multiple copies <strong>of</strong> a data stream<br />
across a number <strong>of</strong> antennas and to exploit the various received versions <strong>of</strong> the data to<br />
improve the reliability <strong>of</strong> data-transfer. It combines all the copies <strong>of</strong> the received signal in an<br />
optimal way to extract maximum information as possible from each <strong>of</strong> them and does not<br />
require CSI, but extracts transmit diversity. Multiuser diversity refers to the degrees <strong>of</strong><br />
freedom due to independent fading in different users’ channels [5]..<br />
Spatial Multiplexing [6]: It multiplexes more than one symbol per channel use. The number<br />
<strong>of</strong> symbols transmitted per symbol use determines the spatial rate. This scheme allows<br />
transmission <strong>of</strong> multiple independent data streams on a set <strong>of</strong> transmit antennas without any<br />
CSI. However, the maximum number <strong>of</strong> streams that can be transmitted is limited by (Nr, Nt),<br />
where Nr is the number <strong>of</strong> receive antennas, Nt is the number <strong>of</strong> transmit antennas. In<br />
WiMAX, the spatial rate<br />
Beam forming [6]: This scheme uses CSIT to put energy only into the Eigen modes <strong>of</strong> the<br />
channel. It extracts both diversity and array gains. Since this scheme requires CSIT, it should<br />
be ensured that the delay between the CSIT estimates and its usage be smaller compared to<br />
the quasi-static interval <strong>of</strong> the channel (i.e., the period over which the channel remains<br />
constant).<br />
A comparison between the STC Alamouti, SM and SIMO is simulated in [6] for the same<br />
spectral efficiency. At 1% block error rate (BLER) the STC scheme has a SNR advantage <strong>of</strong><br />
approximately 1.0 dB and the spatial multiplexing has an advantage <strong>of</strong> approximately 1.5 dB<br />
compared to SIMO. In [8], the MIMO schemes are simulated for equal mean packet data call<br />
duration (EMPC-D) and equal mean packet data volume (EMPC-V). The spectral efficiency<br />
(h) and throughput performance for pedestrian B–3 model are given in the table 1.<br />
109
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
MIMO for 802.16j and 802.16m: A more efficient solution to extend coverage and enhance<br />
throughput in 802.16j and 802.16m is to deploy low-cost RSs in the network [5]. In relaybased<br />
wireless network, the transmission has two phases, namely from source to relay (SR)<br />
and from relay to final destination (RD). In this case the access RS or end user may receive<br />
the signal either an opportunistic fashion or from more than one superior RS in co-operative<br />
fashion [7].<br />
Frequency reuse<br />
The PFR takes the advantage <strong>of</strong> sub-channelization from OFDMA where FFR takes the<br />
advantage <strong>of</strong> cell saccharisation.<br />
PFR scheme: The key issue in PFR is how efficiently the sub-channel segmentation and the<br />
sub-channel are reused. In PFR [11], the Cell Edge Users (CEU) operate on a sub-frame zone<br />
with a fraction <strong>of</strong> all available sub-channels, while the inner cell users operate on a zone with<br />
all sub-channels available. Usually, CEU are operated with frequency reuse=3 (called the R3)<br />
zone and the cell center users (CCU) are operated in a R1 sub-frame zone. The transmission<br />
frame structure with time and frequency resource allocations for the R1 and R3 zones is<br />
shown in Fig.2. All the cells/sectors operated in the network have common boundaries. In<br />
PFR [11] scheme, the interference margins at the cell edge are effectively controlled for larger<br />
R3 zone sizes. In [12], there are five different algorithms (with or without frequency reuse)<br />
that are compared with either uniform Fixed Channel Allocation (FCA) or non uniform FCA.<br />
The main principle in the channel allocation in BS is the channel borrowing from CEUs to<br />
CCUs and from lower priority services to higher priority services <strong>of</strong> CEUs and CCUs. From<br />
their results [12], the performance <strong>of</strong> uniform-FCA scheme is much worse than non uniform-<br />
FCA.<br />
Figure 2 PFR-Transmission Frame [11]<br />
FFR scheme: In cell saccharisation, the BS, Omni directional antenna is replaced by several<br />
directional antennas each radiating within a specified sector. With hexagonal cell patterns, the<br />
most common sectoring schemes employ 120° or 60° sectors. Splitting each cell into three<br />
sectors multiplies the cell capacity by three, but since a separate channel is allocated to each<br />
sector, three channels are needed to serve the entire cell and the spectral efficiency remains<br />
unchanged. However, in practice, the radiation diagrams <strong>of</strong> the sector antennas used are<br />
exceeding the sector boundaries and there is a significant overlap region between adjacent<br />
sectors, which prohibits FFR.<br />
110
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
Figure 3- Cell Sectoring 120-deg hexagonal and 90-deg circle [10]<br />
One way to handle this problem is to reserve a group <strong>of</strong> subcarriers to serve users located in<br />
overlap regions, those are called CEU. The remaining subcarriers are reserved for the nonoverlap<br />
regions and they are reused in all sectors. Since only one part <strong>of</strong> the subcarriers is<br />
reused, this scheme ensures PFR. The FFR technique uses the existing MIMO concepts to<br />
serve the users located in the overlap regions. More specifically, while two users located in<br />
the overlap region <strong>of</strong> two sectors, say A and B, are served by these two sectors, the third<br />
sector C serves a user located in its overlap free region Thus, three users are simultaneously<br />
used and full frequency reuse is achieved. In [10], each cell is partitioned in to four sectors<br />
with different reuse (half rate code and full rate code) is shown in Fig.3. For interference<br />
reasons, 120° sectors are commonly used in hexagonal cell patterns. However, the number <strong>of</strong><br />
sectors per cell does not have to follow any rule when all <strong>of</strong> the sectors use the same<br />
frequency.<br />
3. MAC LAYER QOS IMPROVEMENTS<br />
Even though the spectrum efficiency and throughput capacity is improved by PHY layer,<br />
MAC layer plays the important role to improve the end user QoS/QoE. The standard [1, 2, 3<br />
and 4] defines only the functionalities <strong>of</strong> the MAC functions but the algorithm<br />
implementation in each block is vendor specific. The recent researches pay more attention to<br />
some <strong>of</strong> the blocks like scheduler, Radio Resource Management (RRM), Handover and multichannel<br />
HARQ IEEE 802.16 standards defines five different service classes and its QoS<br />
parameters, which are Unsolicited Grant Service (UGS), Extended real time Polling Service<br />
(E-rtPS), real time Polling Service (rtPS), non-real time Polling Service (nrtPS) and Best<br />
Effort (BE) services. Any end user application follows any one <strong>of</strong> the above five service<br />
classes. The MAC control function algorithm considers the type <strong>of</strong> service and its QoS<br />
parameters.<br />
A. Radio Resource Management (RRM)<br />
One <strong>of</strong> the main functions <strong>of</strong> the RRM is the Call Admission Control (CAC). Possible<br />
input parameters for the CAC module are: the connection type (downlink/uplink and<br />
new/handover), service class and QoS. The CAC implemented in [13, 14, 15, 16 and 17] are<br />
111
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
considered for our analysis. Performance metrics <strong>of</strong> the CAC algorithm is Call Blocking<br />
Probability (CBP), Call Dropping Probability (DP) and Bandwidth Utilization (BU) [15].<br />
The simple Quadra-Threshold Bandwidth Reservation (QTBR) CAC scheme is introduced<br />
in [16], which uses the different threshold for different service class. In that real time<br />
applications are more prioritized. An adaptive CAC implemented in [17] used the minimum<br />
bandwidth requirement Bmin and the maximum bandwidth requirement Bmax input<br />
parameters and a simple Dynamic Guard Channel scheme. In this scheme the bandwidth is<br />
reserved for hand<strong>of</strong>fs by defining two levels <strong>of</strong> threshold, thmin and thmax, where thmin <<br />
thmax < B (total bandwidth). In [14], Adaptive Fuzzy-Based Admission Controller is used<br />
which considers the buffer occupancy based algorithm and the connection type. Token Bucket<br />
based CAC scheme is used in [15] and they compare the performance <strong>of</strong> existing handover<br />
prioritized CAC algorithm and Service class specific CAC algorithm. Their algorithm follows<br />
token bucket algorithm equation (1). Higher priority is given to the handover calls.<br />
Where, f = Duration <strong>of</strong> the frame in msec. CrtPS and CNRT = BW allocated to rtPS and<br />
nrtPS connections. ri and bi = Token arrival rate and bucket size in Kbps. mi = d/f, mi<br />
(integer); d = Maximum latency for rtPS in msec.<br />
In [12], the CAC considers the E2E delay as one the important parameter for call admission.<br />
Since the current network is based on multi-hop communication, E2E delay should be<br />
considered for any new CAC algorithm.<br />
Mobility management<br />
It scans neighbour BSs and decides whether the MS should perform a handover operation.<br />
In general, the handover can be broadly classified in to hard handover or s<strong>of</strong>t handover. The<br />
handover may occur with the same WiMAX network, called as horizontal hand<strong>of</strong>f or it may<br />
occur with different network, called as vertical hand<strong>of</strong>f. The QoS parameters considered<br />
during the handover are handover latency and packet drop.<br />
IEEE 802.16e [1] defines 3 types <strong>of</strong> handover, the hard Handover scheme which may be<br />
<strong>applied</strong> in the case <strong>of</strong> a low speed and it uses the mechanism break-before-make. Other two<br />
types are S<strong>of</strong>t Handover, Macro Diversity Handover (MDHO) and Fast Base Station<br />
Switching (FBSS) which may be <strong>applied</strong> in the case <strong>of</strong> a high speed and use the make-beforebreak.<br />
In our study, the Mobile IP (MIP) based handover performs well to provide more<br />
security and less delay [37]. In FBSS MS maintain the group <strong>of</strong> BSS nearer to it and select<br />
one <strong>of</strong> them as an Anchor BS among BSS, and communicates all traffic with only the Anchor<br />
BS. The mobile has the right to change Anchor BS when it wants as long as it is connected<br />
with. But these s<strong>of</strong>t handover can be <strong>applied</strong> only when the cell size is small to maintain more<br />
than two BSS in the MS diversity set or relay networks to maintain more than two relays in<br />
the MS diversity set.<br />
Hand<strong>of</strong>f with other wireless networks: The fourth generation network is based on All-IP<br />
and the emergence <strong>of</strong> multiple standards, such as WiMAX, 3GPP (LTE), Wi-Fi, Mobile-Fi<br />
(IEEE 802.20) and IEEE 802.15, it has become necessary to define a new interworking<br />
standard IEEE 802.21, which is being designed to facilitate handover among any wireless<br />
access technologies. IEEE 802.21 defines the Media Independent Handover (MIH) functions<br />
which facilitates messaging among IP and the various wireless link layer technologies, to<br />
select an appropriate network for handover and to set up new wireless links. In MIH, QoS at<br />
the IP layer can be interworked with QoS at the wireless link layers to provide the QoS<br />
112
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
required by the user application. Fig 4 shows the MIH function protocol stack. The three type<br />
<strong>of</strong> messages used are event notification from the same layer or other layer like signal<br />
threshold command service for processing the handover with any network and finally the<br />
information service to provide information to other layers.<br />
Figure 4- MIH Functions (Protocol Stack) [25]<br />
In [23] network initiated handover based on IEEE 802.21 is simulated. Both the Mobile<br />
Terminal Initiated Handover (MIHO) and Network Initiated Handover (NIHO) are simulated<br />
in [24]. The comparative analysis <strong>of</strong> alternative approaches that have been developed by IEEE,<br />
3GPP and others for handover <strong>of</strong> PSTN and Internet sessions among multiple wireless<br />
technologies including UMTS, WiFi, WiMAX, Mobile-Fi and WPANs discussed in [25]. The<br />
s<strong>of</strong>tware architecture which supports the seamless handover between WiMAX and UMTS<br />
network is discussed in [30].<br />
Scheduler<br />
It is one <strong>of</strong> the major blocks to ensure the QoS and it performs the scheduling <strong>of</strong> MAC<br />
control and data packets. The WiMAX schedulers can be classified into two main categories:<br />
channel-unaware schedulers and channel-aware schedulers. The channel-unaware schedulers<br />
are quite simple for design, which assumes that the channel is error free for packet<br />
transmissions. Basically, the channel-unaware scheduler makes the scheduling decision<br />
without any channel state information. However, in wireless environment, there is a high<br />
variability <strong>of</strong> radio link such as channel attenuation, fading and noise interference. The<br />
channel-aware schedulers, on the other hand, consider the channel state information while<br />
scheduling the packet.<br />
Since there is no single homogenous scheduling algorithm meets all the QoS requirements<br />
for different service classes, researchers are trying to find hybrid scheduling algorithms to<br />
satisfy QoS for different service classes. The hybrid scheduling algorithm combines more<br />
than one legacy homogeneous schedulers to satisfy the QoS. Three tiers scheduling also<br />
proposed in [20], but the complexity is increased.<br />
Channel-aware schedulers are kind <strong>of</strong> an opportunistic schedulers, whereby each MS is<br />
assigned a priority based on its channel quality and service status. In [18], some <strong>of</strong> the<br />
channel aware schedulers are analyzed for different QoS objective functions. The cross layer<br />
scheduling is analyzed in the next section.<br />
Hybrid automatic repeat-request (HARQ)<br />
It is critical for IEEE 802.16e OFDMA network, as it can significantly improve the<br />
reliability <strong>of</strong> wireless link. The IEEE 802.16e [1] standard defines the basic HARQ support.<br />
Since the data transmission is frame based and different OFDMA sub-channel are allocated<br />
for different users, multi-channel HARQ scheme is introduced to improve the performance. In<br />
113
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
order to adapt to various link conditions and to efficiently utilize available bandwidth, in the<br />
present design <strong>of</strong> multi-Channel (or N-process) HARQ operation, the MAC Service Data Unit<br />
(MSDU) as well as the PHY Service Data Unit (PSDU) is <strong>of</strong>ten divided into several<br />
fragments. With Multichannel HARQ, the transmission and retransmission <strong>of</strong> the fragments<br />
are taken care <strong>of</strong> in multiple HARQ channels that run in parallel as shown in Fig.5<br />
Figure 5- Multi-channel HARQ [26]<br />
IEEE 802.16m draft versions support the multi-channel HARQ for the IMT advanced<br />
WiMAX network. The maximum number <strong>of</strong> parallel HARQ channels defined in the current<br />
system will render a potential bottleneck and preponderantly limit the overall system capacity<br />
<strong>of</strong> next generation WiMAX network, wherein the wireless link is anticipated to support much<br />
higher data rate. The shared multichannel HARQ for IEEE 802.16j and IEEE 802.16m<br />
networks is an on going research. In order to enhance the throughput performance and<br />
improve the preparedness <strong>of</strong> multichannel HARQ protocol for adoption by future 802.16j and<br />
802.16m system, adaptive fragmentation and ACID field expansions are introduced in [26].<br />
Network coding<br />
HARQ is adopted to transmit data packets reliably. However, it sacrifices resilience in time<br />
varying channels, and it may under-utilize the wireless medium in the cases <strong>of</strong> multipath and<br />
multi-hop transmissions. On the other hand, random wireless network coding (NC) has been<br />
shown to be effective towards improving throughput in multi-hop wireless networks. NC can<br />
be implemented with multi-channel HARQ to improve the performance <strong>of</strong> transmission<br />
efficiency, which is called as shared multi-channel HARQ.<br />
Header Compression [27]<br />
Robust Header Compression (ROHC), a standard header compression scheme proposed by<br />
IETF, Bidirectional Optimistic-O mode provides higher compression efficiency and<br />
robustness compared with Payload Header Suppression (PHS) which is an optional header<br />
compression scheme for mobile WiMAX. However, if consecutive packet losses occur<br />
because <strong>of</strong> a Shadow Region <strong>of</strong> Mobile WiMAX, header decompression failure rate <strong>of</strong> the<br />
ROHC-O mode increases. Therefore, a complementary mechanism is proposed in [27] using<br />
the dynamic adjustment <strong>of</strong> an optimistic parameter, which has an effect on both the<br />
compression efficiency and the robustness, for improving the performance <strong>of</strong> ROHC - O<br />
mode.<br />
114
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
Framing<br />
The WiMAX uplink frame contains the random access channel for ranging and bandwidth<br />
request for BE service, Fast feedback channel to inform the channel conditions and user data<br />
channels. A proper frame allocation also leads to effective bandwidth usage <strong>of</strong> the system and<br />
some <strong>of</strong> it concentrates on the proper frame allocations to improve the QoS. To improve the<br />
system access success probability a mathematical basis for an accurate model is proposed in<br />
[28] ¸ that can be efficiently used to determine the resource allocation for each class. It also<br />
provides a useful analytical tool for contention systems that utilize random access for network<br />
entry, resource acquisition, and/or contention-based data transmission<br />
4. E2E AND CROSS LAYER QOS IMPROVEMENTS<br />
To assure the QoS/QoE for end user application it is not necessary to satisfy the QoS <strong>of</strong><br />
access network alone, the backhaul traffic in service provider’s connectivity network should<br />
satisfy the necessary QoS requirements. Here the E2E network considered for the analyses are;<br />
Homogeneous network and heterogeneous network.<br />
A. Homogeneous E2E QoS<br />
IEEE 802.16 networks provide mechanisms for QoS support at MAC level, but E2E QoS<br />
issues are not addressed in detail in the standards. Some kind <strong>of</strong> cross-layer work is needed to<br />
satisfy both the network and MAC layer QoS. Number <strong>of</strong> enhancements have been proposed<br />
to enable the <strong>of</strong>fering <strong>of</strong> different levels <strong>of</strong> QoS in IP networks including the integrated<br />
services (IntServ) architecture, and the differentiated service (DiffServ) architecture [29].<br />
IntServ is implemented by four components: the signalling protocol (e.g. RSVP), the<br />
admission control, the classifier and the packet scheduler. Furthermore, some rules are<br />
prescribed to classify DiffServ IP packets into different priority queues based on QoS<br />
indication bits in IP header. Therefore, the QoS architecture <strong>of</strong> WiMAX access network can<br />
support both IntServ and DiffServ. So the mapping rules are created between WiMAX MAC<br />
layer and IP layer.<br />
In traditional way, RSVP signalling message can be classified into a special high priority<br />
queue, and be transmitted in the second management connection. So the E2E resource is<br />
reserved for each connection to provide the network QoS. In the cross-layer approach, QoS<br />
mapping is created and the WiMAX MAC connection request message, Dynamic Service<br />
Addition (DSA) request carries the RSVP-PATH specific message. If the BS supports both<br />
MAC and network specific QoS, accepts the connection and informs it in the DSA response<br />
message. Since the communication in WiMAX network is connection-oriented; the<br />
application must establish the connection with the BS before data transmission. The message<br />
exchange for DSA and DSC can be deployed to carry QoS parameters <strong>of</strong> IntServ services for<br />
E2E resource reservation. For DiffServ services, on the other hand, a number <strong>of</strong> per-hop<br />
behaviors (PHBs) for different classes <strong>of</strong> aggregated traffic can be mapped into different<br />
connections directly.<br />
B. Heterogeneous E2E QoS<br />
Users shall be able to access its services through different access technologies, such as<br />
WLAN, WiMAX, UMTS and DVB technologies, from the same or different network<br />
operators, and to seamless move between different networks with active communications [31].<br />
The recent standardization efforts in the IETF for a new extensible IP signalling protocol suite<br />
(NSIS) describes the network layer signalling to provide network QoS. So the recent<br />
115
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
researches are trying to integrate both WiMAX E2E architecture defined by the WiMAX<br />
forum and NSIS signalling architecture. The QoS NSIS Signalling Layer Protocol (QoS-<br />
NSLP) is used to signal the application QoS requirements and request the resource allocation<br />
along the full path <strong>of</strong> data flows. The NSIS protocol supports the interoperability between<br />
different QoS-enabled domains defining distinct QoS-Models (QOSM) depending on the<br />
underlying network technologies. The QoS-Model defines the QoS parameters, the traffic<br />
descriptors and the methods to provide the desired QoS through the Resource Management<br />
Function (RMF) specification for the NSIS nodes <strong>of</strong> a specific domain. The QoS parameters<br />
and the traffic descriptors are encapsulated in an object called QSPEC and are interpreted by<br />
NSIS nodes according to the QOSM. Two different types <strong>of</strong> QSPECs are defined: Initiator<br />
and Local QSPEC. The former is created by the QoS NSIS Initiator (QNI) and the later is<br />
created at the edge node between two QoS enabled domains through a coherent mapping <strong>of</strong><br />
the Initiator QSPEC parameters.<br />
Other cross layer improvements<br />
In multiuser scenario, channel gains <strong>of</strong> a specific subcarrier vary from one wireless terminal<br />
to another wireless terminal. At any given time in a large network for each sub-channel there<br />
is a high probability that there is a user whose OFDMA subcarrier assigned to the users who<br />
see good channel gains on them. Maximization <strong>of</strong> system performance can be achieved by<br />
allocating the OFDMA subcarriers to the users with good channel condition. The performance<br />
<strong>of</strong> the MAC functions such as CAC, sub-channel allocation, schedulers and HARQ modules<br />
depends more on the PHY channel condition.<br />
In [33], priority function is defined for each connection admitted in the system, and BS<br />
update it dynamically depending on the wireless channel quality, QoS satisfaction and service<br />
priority across layers. The connection with the highest priority is scheduled each time. The<br />
occupation <strong>of</strong> physical resource is an important feedback for the CAC to admit the new<br />
connection. In [34] the cross layer based CAC and sub-channel assignment is simulated,<br />
which takes the PHY layer channel information. In [35], channel aware WF2Q+ based<br />
scheduling for point-to-multipoint (PMP) network is simulated, where the weights <strong>of</strong> the<br />
service class is dynamically varied based on the channel condition. On the other hand, Blind<br />
channel aware schedulers leads to fairness issue since those schedulers result in optimal<br />
network resource utilization, it does not take into account each user’s QoS perspective. To<br />
avoid this problem utility function is used to strike a balance between system capacity and<br />
fairness among users is called proportional fairness (PF). In [32] channel aware PF schedulers<br />
are simulated.<br />
The new WiMAX networks are based on MIMO and RS nodes. So the diversity and multihop<br />
communication should be considered for the upcoming researches. In [7], the scheduling<br />
and radio resource allocations performed at BS is based on the PHY channel conditions. The<br />
BS uses a look-up table developed for point-to-point flat fading links with a given SNR<br />
conditions. For each sub-channel and for each user, the BS calculates the post processing<br />
SNR with relays and calculates the E2E throughput for each cooperative diversity<br />
communication. Finally it selects the best one and uses the same for data transmission.<br />
Based on wireless channel condition, adaptive fragmentation is <strong>applied</strong> for multi-channel<br />
HARQ to improve the performance <strong>of</strong> normal multi-channel HARQ [26].<br />
In adaptive random network coding [36], the channel state information can be used to<br />
dynamically construct the packets that could be better matched to the fluctuating channel<br />
conditions, so as to obtain a higher throughput.<br />
116
QOS IMPROVEMENTS IN MOBILE WIMAX NETWORKS<br />
5. CONCLUSION<br />
This paper provides a comprehensive discussion <strong>of</strong> recent works in WiMAX networks to<br />
support the required QoS for multimedia applications. In traditional communication networks,<br />
the open systems interconnection (OSI) layer architecture has been widely adopted, and the<br />
performance optimization is conducted largely within each individual protocol layer. This<br />
may work well for the wired and fixed wireless networks. On the other hand due to high<br />
mobility and high data rate support nature <strong>of</strong> 4G networks, the conventional OSI architecture<br />
may not work well to support the real time applications. This leads to the Cross-layer design<br />
approaches for efficient utilization <strong>of</strong> the scarce radio resources with QoS provisioning. At the<br />
same time Cross-layer design approaches are critical in nature. This paper summarized in one<br />
place most <strong>of</strong> the efforts being conducted in the cross-layer designs and QoS provisioning in<br />
4G networks.<br />
REFERENCES<br />
[1] IEEE 802.16 -2009 “IEEE Standard for Local and metropolitan area networks Part 16: Air<br />
Interface for Broadband Wireless Access Systems”. IEEE press 2009.<br />
[2] IEEE 802.16j -2009 “IEEE Standard for Local and metropolitan area networks Part 16: Air<br />
Interface for Broadband Wireless Access Systems Amendment 1: Multiple Relay Specification”.<br />
IEEE press 2009.<br />
[3] IEEE 802.16m TGm “IEEE 802.16m System Description Document, 80216m-08-0034r1”, IEEE<br />
press 2009.<br />
[4] IEEE 802.16m TGm “IEEE 802.16m System Requirements Document, 80216m-07_002r9”,<br />
IEEE press 2009.<br />
[5] M.D. Katz, F.H.P. Fitzek;"WiMAX Evolution - Emerging Technologies and Applications", A<br />
John Wiley and Sons, Ltd, Publication.<br />
[6] S.P. Alex and L.M.A. Jalloul, “Performance Evaluation <strong>of</strong> MIMO in IEEE802.16e/WiMAX”,<br />
IEEE journl <strong>of</strong> selected topics in signal processing, vol. 2, 2008, pp 181-190.<br />
[7] Q. Li, X. Lin, J.Zhang and W.Roh, “Advancement <strong>of</strong> MIMO technology in WiMAX- from IEEE<br />
802.16d/e/j to 16m” IEEE Comm. Mag. vol. 47, issue 6, 2009, pp 100-107.<br />
[8] C.F.Ball, E.Humburg, S.Eder and L.Lacinak, “WiMax Capacity Enhancements introducing<br />
MIMO 2”, 16th IST mob. and wireless comm. summit, 2007, pp 1-5.<br />
[9] B.Can, H.Yanikomeroglu, F.A Onat, E.De Carvalho and H.Yomo, “Efficient Cooperative<br />
Diversity Schemes and Radio Resource Allocation for IEEE 802.16j”, Wireless Comm. and Net.<br />
Conf. 2008, pp 36-41.<br />
[10] H.Sari and S.Sezginer, “MIMO Techniques and Full Frequency Reuse in mobile wimax network”,<br />
Int’l Conf. on Telecomm., 2009, pp 8-12.<br />
[11] M.Hunukumbure, B.Upase, L.Sarperi, and S.Vadgama, “Advanced Techniques for Improving the<br />
QoS <strong>of</strong> the WiMAX Cell Edge User”, IEEE GLOBECOM, 2008, ECP-1082, pp 1-5.<br />
[12] Y. Qi, X. Zhong and J. Wang, “A Downlink Radio Resource Allocation Algorithm with<br />
Fractional Frequency Reuse and Guaranteed Diverse QoS for Multi-Cell WiMAX System”, 3rd<br />
Int’l Conf. on Comm. And Net., 2008, pp 289-294.<br />
[13] S.Lee, G.Narlikar, M.Pal, G.Wilfong and L.Zhang, “Admission Control for Multihop Wireless<br />
Backhaul Networks with QoS Support”, IEEE WCNC 2006, pp 92-97.<br />
[14] S.Ghazal, Y.H Aoul, J.Ben Othman and F.Nait-Abdesselam, “Applying a Self-Configuring<br />
Admission Control Algorithm in a New QoS Architecture for IEEE 802.16 Networks”, IEEE<br />
Symp. on Comp. and Comm., 2008, pp 1023-1028.<br />
[15] S.Kalikivayi, I.S Misra, and K.Saha, “Bandwidth and Delay Guaranteed Call Admission Control<br />
Scheme for QOS Provisioning in IEEE 802.16e Mobile WiMAX”, IEEE GLOBECOM 2008,<br />
ECP-246, pp 1-6.<br />
117
P. RENGARAJU, C.H. LUNG, A. SRINIVASAN, R.H.M. HAFEZ<br />
[16] H.Y Tung, K.F Tsang, L.T Lee and K.T Ko, “QoS for Mobile WiMAX Networks Call<br />
Admission Control and Bandwidth Allocation”, 5th IEEE Consumer Comm. and Net. Conf. 2008,<br />
pp 576-580.<br />
[17] B.Shafaq and K.Ratan, “Adaptive Connection Admission Control and Packet Scheduling for QoS<br />
Provisioning in Mobile WiMAX”, IEEE Int’l Conf. on signal proc.and Comm., 2007, pp 24-27.<br />
[18] So-In.Chakchai, R.Jain and A-K.Tamimi, “Scheduling in IEEE 802.16e Mobile WiMAX<br />
Networks: Key Issues and a Survey” IEEE Journl on Selected Areas in Comm., 2009, pp. 156 -<br />
171.<br />
[19] R.Perumalraja, C-H. Lung and S.Anand " QoS Assured simple Hybrid Uplink Scheduler for<br />
WiMAX networks", Paper submitted in Int’l Conf. on Comm. 2010.<br />
[20] M. Maode, L. Jinchang, S.K. Bose and C-N. Boon, “A Three-tier Framework and Scheduling to<br />
Support QoS Service in WiMAX”, 6 th Int’l Conf. on Info., Comm. and Signal Proc., 2007, pp. 1-5.<br />
[21] J.Jin, B.Li and T.Kong, “Is Random Network Coding Helpful in WiMAX?”, 27th IEEE Conf. on<br />
Comp. and Comm., 2008, pp 2162-2170.<br />
[22] T.Bchini, N.Tabbane, S.Tabbane, E.Chaput, and A.L Beylot, “QoS in IEEE 802.16e for VoIP and<br />
Video with mechanism <strong>of</strong> S<strong>of</strong>t Handover – FBSS in Highway”, 8th Int’l Conf. on Networks,<br />
2009, pp 88-93.<br />
[23] J-Y.Baek, D-J.Kim, Y-J.Suh, and Y-D.Chung “Network-initiated Handover Based on IEEE<br />
802.21 Framework for QoS Service Continuity in UMTS/802.16e Networks”, IEEE VTC Spring<br />
2008, pp 2157-2161.<br />
[24] V.Jesus, S.Sargento and R.L Aguiar, etc., “Mobility with QoS Support for Multi-Interface<br />
Terminals Combined User and Network Approach”, 12th IEEE Symp. on Comp. and Comm.,<br />
2007, pp 325-332.<br />
[25] D.J Wright, “Maintaining QoS during Handover among Multiple Wireless Access Technologies”,<br />
Int’l Conf. on Management <strong>of</strong> mobile Business, 2007, pp 1-10.<br />
[26] T.Zhifeng, A.Li, J.Zhang and T.Kuze, “Performance Improvement for Multichannel HARQ<br />
Protocol in Next Generation WiMAX network”, IEEE WCNC, Ap.2008, pp 2009-2014.<br />
[27] Zhuoqun Li, Walter Featherstone and Stavros Tzavidas "Shared Multi- Channel HARQ" IEEE<br />
C802.16m-09/1335.<br />
[28] J Kim, H Woo, H Lee and M Lee “Dynamic adjustment <strong>of</strong> optimistic parameter <strong>of</strong> ROHC for<br />
performance improvement”, Int’l Conf. on nfo Net., 2009, pp 1-3.<br />
[29] J Chen, W Jiao and Q Guo, “An Integrated QoS Control Architecture for IEEE 802.16 Broadband<br />
Wireless Access Systems”, IEEE GLOBCOM, Vol. 6, 2005, pp 3330-3335.<br />
[30] Anand S.V, “QoS based Handover layer for a Multi-RAT Mobile Terminal in UMTS and Wi-<br />
MAX Networks”, 3rd Comm. Sys. S<strong>of</strong>t. And middleware workshop, 2008, pp 472-479.<br />
[31] S Sargento and V Jesus, etc., “Context-Aware End-to-End QoS Architecture in Multi-technology<br />
Multi-interface Environments”, 16 th IST mob. and wireless comm. summit, 2007, pp 1-5.<br />
[32] A Ukil and J Sen, “QoS Aware Cross-Layer Optimized Resource Allocation in WiMAX<br />
Systems”, 1st Int’l Conf. on wireless Comm. Veh. Tech., Info. theory and Areo space., 2009, pp<br />
818-822.<br />
[33] Q Liu, X Wang and G.B Giannakis, “Cross-Layer Scheduler Design with QoS Support<br />
forWireless Access Networks”, Proc. <strong>of</strong> 2nd Int’l Conf. on QoS in Het. Wired/Wireless Networks,<br />
2005.<br />
[34] B.Xie, W.Zhou and J.Zeng, “A Novel Cross-Layer Design with QoS Guarantee for WiMAX<br />
System”, 3rd Int’l Conf. on pervasive computing and App., 2008, pp 835-840.<br />
[35] A.Lera, A.Molinaro and S.Pizzi, “Channel-Aware Scheduling for QoS and Fairness Provisioning<br />
in IEEE 802.16-WiMAX BWA systems”,IEEE Network Magzine, 2007, pp 34-41.<br />
[36] J.Jin and B.Li, “Adaptive Random Network Coding in WiMAX”, IEEE Int’l Conf. on Comm.,<br />
2008, 2576-2580.<br />
[37] R.Perumalraja, C-H. Lung and S.Anand " An Analysis on mobile WiMAX Security", IEEE<br />
toronto Int'l Conf. on Science and Tech. for humanity, 2009, pp 439-444.<br />
118
AHU J. <strong>of</strong> Engineering & Applied Sciences 3 (1) : 119-128 (2010)<br />
© 2009 ALHOSN University<br />
THE SOLAR-WIND-EV PARADIGM: A STRATEGIC POLICY<br />
FOR A SUSTAINABLE ENERGY FUTURE<br />
Zuwei Yu 1 , Tarik Aouam 2*<br />
1 Choren USA and College <strong>of</strong> Engineering, Purdue University<br />
2 Department <strong>of</strong> Industrial Engineering, ALHOSN University, P.O. Box: 38722, Abu Dhabi, UAE<br />
ABSTRACT: The main idea <strong>of</strong> the proposed strategic policy is to speed up the deployment <strong>of</strong> Electric<br />
Vehicles (EVs) that can be fueled by solar and wind power. The solar-wind-EV marriage is a bright<br />
paradigm for solving major energy and environmental problems thanks to recent EV technology<br />
breakthroughs. To quantify the benefits <strong>of</strong> such a policy, we formulate a leader-follower gaming model,<br />
whose solution results characterize a significant decrease <strong>of</strong> oil prices, when there is a high penetration <strong>of</strong><br />
EVs. As a result, US oil imports would be reduced by nearly 60% from current levels, energy security<br />
would be much improved, and CO 2 emissions would be reduced significantly. The economic impact <strong>of</strong> the<br />
proposed policy is significant.<br />
Keywords: Solar; wind power; EV; oil; energy security; gaming.<br />
1. INTRODUCTION<br />
Oil consumption and oil dependence have caused severe energy security, environmental, and<br />
economical problems. The U.S. generated 50% <strong>of</strong> the CO 2 emissions in the category <strong>of</strong> passenger<br />
cars and light trucks, with a huge trade deficit, and reduced energy security [40]. Therefore, there<br />
is an urgent need for effective policies. One potential policy is the fast deployment <strong>of</strong> EVs with<br />
batteries that can be charged using solar and wind power.<br />
The California Air Resources Board (CARB) mandated a set <strong>of</strong> targets in the early 1990s for<br />
adopting EVs. However, these targets were either altered or disbanded due to technical and<br />
political barriers. CARB voted, in March 2008, to give EVs and HFCVs (hydrogen fuel cell<br />
vehicles) parallel treatment. Hydrogen vehicles can also help achieve the desired results but they<br />
are costly and may contribute to ozone damage due to leakages. The U.S. hydrogen initiative has<br />
been questioned even since it was promoted in 2003 [1]. The argument has been that it may be<br />
too expensive to deploy hydrogen passenger cars and light trucks. Unfortunately, policy makers<br />
never paid attention to the argument. Today, the hydrogen infrastructure is still badly lacking<br />
even with heavy subsidies. Should we wait for the hydrogen economy to come with a huge cost<br />
burden? No. Even if the cost <strong>of</strong> such an economy is not too prohibiting, a hydrogen economy<br />
would still require many years to develop. The HFCVs have much lower energy efficiency than<br />
the EVs. On the other hand, EVs have become a reality after several breakthroughs in 2007,<br />
including advanced batteries with a range <strong>of</strong> up to 250 miles per charge, a much reduced recharge<br />
time and the battery safety features [2].<br />
The paper will quantitatively analyze the impact <strong>of</strong> large scale deployment <strong>of</strong> EVs on oil<br />
demand, prices, and energy security. A gaming model is formulated to simulate oil price<br />
movements and energy security improvements when various EV penetration rates are assumed.<br />
____________________________________<br />
* Corresponding Author. Tel.: +971 4070562<br />
E-mail : t.aouam@<strong>alhosn</strong>u.ae<br />
119
ZUWEI YU, TARIK AOUAM<br />
Figure 1. The annual average solar power in the continental U.S. (courtesy <strong>of</strong> NREL).<br />
An area <strong>of</strong> 100 square miles in the red zone <strong>of</strong> Figure 1 could provide the total US energy needs.<br />
In fact, such an area can provide 160,000 2 ×8kWh = 204,800,000 MWh/day <strong>of</strong> electricity. According<br />
to the 2004 EPRI-DOE Handbook: “Supplement <strong>of</strong> Energy Storage for Grid Connected Wind<br />
Generation Applications,” the U.S. peak demand was 808,000 MW in 2003, which may be converted<br />
to a peak electricity demand <strong>of</strong> no more than 820,000×24 = 19,200,000 MWh in 2008. Hence, the<br />
solar power in the 100 square mile area could provide the total U.S. energy needs.<br />
Wind resources are significant in the Great Plains; North Dakota alone has a wind power potential<br />
<strong>of</strong> 415,000 MW [13]. Some <strong>of</strong>fshore areas also have rich wind resources but with costs relatively<br />
high, [14]. The Great Plains can provide wind power at a price range around $60-80/MWh. Although<br />
wind is intermittent, wind power can be stored in large scale by the CAES (compressed air energy<br />
storage) technology [15]-[16]. Aquifer CAES systems can be used for efficient air compression, as<br />
pressure could be kept relatively constant to reduce excessive heat generation.<br />
2.2 EVS, PHEVs and HFCVs<br />
High performance zero emission EVs have been the dream cars for decades. California state<br />
legislators created the ZEV (zero emission vehicle) mandate back in 1990 after General Motors<br />
promised to mass-produce its battery-powered EV1 [17]. Unfortunately, battery development<br />
seriously lagged. Consequently, CARB repeatedly trimmed the quotas and allowed<br />
manufacturers to build a larger number <strong>of</strong> the so-called “partial zero-emissions vehicles.”<br />
The ZEV regulation came to a halt in 2003 when GM led a lawsuit arguing that CARB's ZEV<br />
mandate was regulating fuel efficiency, a power granted to the federal government. CARB settled<br />
the case by giving up the mandated EVs penetration rates and allowing automakers to research<br />
and commercialize HFCVs, beginning with the demonstration <strong>of</strong> fuel-cell cars by 2008. Partly<br />
due to the pressure from the automakers, CARB allowed fewer credits for EVs than for HFCVs<br />
until March 2008, when it voted again to give EVs and HFCVs parallel treatment.<br />
120
THE SOLAR-WIND-EV PARADIGM: A STRATEGIC POLICY FOR A SUSTAINABLE ENERGY FUTURE<br />
The first breakthrough related to EVs was the development <strong>of</strong> li-ion and li-ion polymer<br />
batteries that have a range around 100 - 250 miles per charge. The second is that recharge times<br />
are about 10 minutes and the third was the safety features in batteries to guard against<br />
temperature rise and explosion [2]. In addition, the nano ultracapacitor technology has reached<br />
the demonstration stage, which may support a drive range <strong>of</strong> 500 miles per charge with a charge<br />
time less than 10 minutes [18] – [19]. In fact, the Honda FCX fuel cell vehicle has a small<br />
ultracapacitor installed to boost the power for acceleration. Several EV models are summarized in<br />
Table 1. Even though the sporty Tesla is priced at $98,000, several firms have been designing<br />
mid-size EVs that can be sold at prices from $20,000 to below $50,000. As a contrast, there have<br />
been no HFCVs promoted for sale in this price range.<br />
Table 1. Selected EV Models [24].<br />
Model Range Charge Energy Battery Country Launch<br />
(miles) time (min.) (kWh)<br />
Cleanova-II 130 ~30 30 Li-ion France 2008<br />
Dodge Zeo 250 n/a 64 Li-ion U.S. n/a<br />
F6e 64 ~10 n/a Iron China 2009<br />
GM Volt 40 n/a 8 Li-ion U.S. 2010<br />
Hotzenblitz 218 n/a ~40 PLIB Switzerland 2008<br />
Lighting GT n/a 10 35 Li-ion UK 2009<br />
Phoenix SUT 130 10 35 Li-ion U.S. 2008<br />
250 10 70 (nano)<br />
Regane/Mixim n/a n/a n/a Li-ion France/Japan 2010<br />
Subaru G4e 125 15 n/a Li-ion Japan 2008<br />
Tesla 220 210 53 Li-ion U.S. 2008<br />
PLIB – Polymer Li-ion battery, also used as Li-Po sometime.<br />
PP – peak power<br />
Table 2 shows convincingly the efficiency advantages <strong>of</strong> EVs over HFCVs. It is assumed that<br />
hydrogen is produced at renewable power generation sites so electricity losses are zero, which is<br />
quite optimistic. The data is from various sources (e.g. [20] - [23]) and modified based on our<br />
best knowledge (e.g. 9.3% <strong>of</strong> electricity losses instead <strong>of</strong> 10% in the U.S.) The efficiency<br />
comparison is based on the percentages <strong>of</strong> the same amount <strong>of</strong> electricity being delivered to the<br />
wheels <strong>of</strong> different vehicles. EVs are clearly twice as efficient as HFCVs. Notice that our<br />
estimate <strong>of</strong> the HFCV efficiency is consistent with other estimates. For example, a target<br />
efficiency <strong>of</strong> 42% is set for the fuel cell hybrid vehicles in the Toyota FCHV BOOK.<br />
Table 2. Estimated Energy Efficiency for EVs and HFCVs (%).<br />
Vehicle Power H 2 Power Battery Fuel cell Drive train Overall<br />
to H 2 * Compression<br />
T&D**<br />
or liquefaction<br />
EV 89-92.4 85-92 90-95 68-81<br />
HFCV 75-85 90-92 87-90 *** 55-60 90-95 29-40<br />
(gas)<br />
HFCV 75-85 65-70 94 *** 55-60 90-95 22.7-26<br />
(liquid)<br />
* Electrolysis is assumed and power supply losses are included. The upper limit is the target in the long run while 75% is attainable now.<br />
** In power T&D (transmission & distribution), the average efficiency is 90.7% and the average loss is about 9.3% .<br />
*** Most HFCV models also have small batteries installed to improve efficiency. The batteries also have charge/discharge losses which are not<br />
considered in this table due to their smaller sizes.<br />
121
ZUWEI YU, TARIK AOUAM<br />
Based on the results above, EVs would only need about half <strong>of</strong> the solar and wind power<br />
compared to the HFCVs for the same driving distance and same weight. That is, HFCVs would<br />
be twice as expensive as the EVs in terms <strong>of</strong> the renewable power infrastructure. As shown in<br />
Table 1, typical EVs (e.g., Tesla and Phoenix SUT) may need 0.25 – 0.3 kWh for driving one<br />
mile. If they are charged overnight, the cost <strong>of</strong> 1.0 kWh may be no more than 10 cents, resulting<br />
in a cost <strong>of</strong> about 3 cents per mile <strong>of</strong> EV driving. The cost <strong>of</strong> gasoline driven vehicles for the<br />
same one mile would then be 360 cents/21 = 17 cents, given that the gasoline price is $3.6/gallon<br />
and the average MPG is 21 (21 is the national average for cars and light trucks).<br />
Other advantages <strong>of</strong> EVs may include: much lower maintenance time and cost, overnight<br />
charging to smooth out the power load, greater safety factor than the HFCVs, minimization <strong>of</strong><br />
water use, no water vapor produced, etc. Results from the French postal service La Poste EV test<br />
program show that EVs can reduce operating costs significantly [25]. As a result, La Poste<br />
decided to replace its old fleet with EVs. However EVs have notable disadvantages, including<br />
their susceptibility to power outages and currently high investment cost. The first issue may not<br />
be an important concern because most families can buy one EV and one PHEV, while the second<br />
issue can be resolved by mass production and battery standardization. It should be noted that EVs<br />
are more reliable than PHEVs in the use <strong>of</strong> electricity from renewable sources, as there is no<br />
guarantee that PHEVs will be plugged-in due to various reasons including inconvenience and the<br />
possibility <strong>of</strong> gasoline price going down. In this paper, PHEVs are used for hedging the risks <strong>of</strong><br />
power outage rather than a reliable means to reduce oil consumption.<br />
3. THE LEADER-FOLLOWER GAMING MODEL<br />
The model is an extension <strong>of</strong> what we have developed previously for electricity analysis [10]. It<br />
is a 2-level leader-follower gaming model where the leaders’ Cournot gaming is in Level-1 while<br />
the fringe producers’ pr<strong>of</strong>it maximization is in Level-2, as described below.<br />
To facilitate the presentation, we first adopt the notations below:<br />
f fringe producer index<br />
s leader producer index<br />
t time period (in year)<br />
g strategic storage owner index (usually governments)<br />
i oil production field (or plant) index<br />
maximum oil production limit<br />
availability <strong>of</strong> oil production<br />
cost <strong>of</strong> oil production<br />
intercept <strong>of</strong> the linear (inverse) demand function<br />
slope <strong>of</strong> the linear demand function<br />
oil price<br />
pr<strong>of</strong>it <strong>of</strong> producer s<br />
oil produced by plant i<br />
oil sale by s<br />
Pmax( t, s, i)<br />
avail( t, s, i)<br />
Cv( t, s)<br />
a0( t)<br />
b( t)<br />
Price( t)<br />
( s)<br />
op(, t s,)<br />
i<br />
q(, t s)<br />
122
THE SOLAR-WIND-EV PARADIGM: A STRATEGIC POLICY FOR A SUSTAINABLE ENERGY FUTURE<br />
The mathematical model for the leaders’ individual pr<strong>of</strong>it maximization in Level-1 can be<br />
summarized below:<br />
Minimize - ( s)<br />
<br />
<br />
<br />
a0() t b()( t qts (, ) qt (, f)) qts (, ) Cvtsioptsi (, ,) (, ,), st<br />
,<br />
<br />
s f <br />
i<br />
(1)<br />
optsi (, ,) availtsiP (, ,) max(, tsi ,), tsi<br />
, ,<br />
(2)<br />
qts (, ) optsi (, ,),<br />
ts<br />
,<br />
(3)<br />
i<br />
op( t, s, i) 0 t, s,<br />
i<br />
(4)<br />
qts ( , ) 0 ts ,<br />
(5)<br />
The Cournot strategy. (6)<br />
where q ’s are quantities from the fringe producers. Notice that (6) is not a regular constraint but<br />
rather a requirement that the Cournot quantity strategy will be used for developing a set <strong>of</strong><br />
pseudo Karush-Kuhn-Tucker (PKKT) first order conditions in the MCP formulation.<br />
The fringe producers’ pr<strong>of</strong>it maximization problem in Level 2 is:<br />
Max price() t q(, t f ) Cv(, t f ,) i op(, t f ,), i t,<br />
f<br />
(7)<br />
i<br />
s.t.<br />
op(t, f,i) avail(, t f ,)Pmax(, i t f ,), i t, f , i<br />
(8)<br />
qt (, f) opt (, f,) i,<br />
t,<br />
f<br />
(9)<br />
i<br />
op( t, f , i) 0 t, f , i<br />
(10)<br />
qt (, f) 0 t,<br />
f<br />
(11)<br />
Notice that price(t) is calculated from the solution <strong>of</strong> the Cournot-Nash gaming in Level-1 and<br />
fed to (7) as a fixed quantity.<br />
The solution procedure: An iterative method is developed that iterates between the leaders’<br />
MCP problem and the followers’ simple pr<strong>of</strong>it maximization problem. The MCP method by itself<br />
is verified to be an efficient solution procedure as describe in [12] – [13]. The method involves<br />
the conversion <strong>of</strong> the leaders’ problem (level-1) into a set <strong>of</strong> MCP conditions (or the PKKT<br />
conditions, as mentioned earlier), which can be found in our earlier studies and will not be<br />
repeated, [12]. The solution to the MCP equations can be obtained through the use <strong>of</strong> an MCP<br />
solver, for example, PATH or any others that work. The details <strong>of</strong> these MCP solvers can be<br />
found in reference [41]. The nodal prices, price(i), are calculated after the MCP equations are<br />
solved. The fringe producers’ problem is solved using a regular pr<strong>of</strong>it maximization solver. The<br />
whole solution process is an iterative one that can be summarized below.<br />
Step 1: Set fringe producers’ quantities qt (, f) 0.8 Pmax(, t f,)<br />
i.<br />
i<br />
Step 2: Solve the leaders’ problem using PATH.<br />
Step 3: Calculate price(t).<br />
Step 4: Check for convergence. If price(t) is within a preset tolerance from the price(t) <strong>of</strong> the<br />
last iteration, stop. Else, continue.<br />
Step 5: Solve pr<strong>of</strong>it maximization for all fringe producers. Go to Step 2.<br />
123
ZUWEI YU, TARIK AOUAM<br />
Step 6: Stop.<br />
In this study, the convergence <strong>of</strong> the proposed iterative method for solving the 2-level model<br />
is very fast, <strong>of</strong>ten requiring only three iterations or so to reach the equilibrium nodal prices.<br />
4. CASE STUDIES<br />
In the first case study, California is assumed to lead the nation in adopting EVs with a 1%<br />
penetration in 2010 and a 2% in 2011, followed by a 3% nationwide EV penetration in 2012. The<br />
second case study examines a 5% EV penetration nationwide plus a 3% PHEV penetration. The<br />
third case, where OPEC and Eurasia form a coalition and no EVs and PHEVs are deployed, will<br />
serve as a reference scenario. Deployment <strong>of</strong> EVs could occur prior to large scale solar-wind<br />
power projects in regions where coal power is limited, such as California. Natural gas power<br />
plants can be used for charging EVs and reducing CO 2 emissions compared to conventional<br />
gasoline vehicles. The overall CO 2 reduction could be around 27% if PHEVs are charged by the<br />
existing power systems [26]. Conventional oil capacity expansion is taken from recent EIA<br />
studies including new fields discovered. The U.S. ethanol production is assumed to be in line<br />
with the U.S. House Bill HR-6 passed in late 2007 [27]. Oil from the Canadian oil sands is<br />
assumed to be in line with the projection by the Canadian Association <strong>of</strong> Petroleum Producers<br />
[28]. The CTL fuels such as the CTL diesel, gasoline, naphtha, DME (dimethyl ether) and<br />
methanol are based on the data in [29]. Bi<strong>of</strong>uels <strong>of</strong> other countries are from [30].<br />
4.1 Two leaders with a 3% annually EV penetration<br />
In this section, the reference “oil demand growth” is taken from recent EIA studies [31]-[32]<br />
but with some modifications. In the EIA studies, the annual oil demand growth rate <strong>of</strong> the U.S. is<br />
about 0.7% in 2000 – 2007, and about 5-6% for China (may decline to about 2.9% after 2020). It<br />
is assumed that the U.S. oil demand growth is about 0.2% in 2008 – 2012 due to economy slow<br />
down. We modify China’s growth rate so that it declines from 5% in 2010 to about 1% in 2022<br />
with equal annual decrements. We use lower oil demand growth rates than the EIA data in later<br />
years due to our belief that persistently high oil prices would slow down oil demand in the long<br />
run. EVs and PHEVs are considered separately as extra oil demand reduction.<br />
We assume a policy <strong>of</strong> 1% EV penetration in the sales <strong>of</strong> new cars and new light trucks in<br />
2010 and 2% in 2011 in California, followed by a 3% penetration in the U.S. and other OECD<br />
countries starting from 2012. We assume that the fast industrializing countries (e.g. China and<br />
India) would do the same but with a 3-year delay. The 2008 U.S. new car and light truck sales are<br />
around 16 million, as forecasted by the industry, about one million less than in 2005. We forecast<br />
that this number will be around 18 million in 2012 due to population increase. According to [33]-<br />
[34], the average driving distance for U.S. cars and light trucks is about 12,000 MPY (miles per<br />
year), with each consuming about 570 gallons per year (GPY) <strong>of</strong> gasoline on average. All <strong>of</strong> the<br />
factors are considered in the construction <strong>of</strong> the linear demand functions for the planning horizon<br />
starting from 2009. Notice that the fuel use <strong>of</strong> EVs is 0, which equals gasoline (or diesel) savings<br />
<strong>of</strong> 570 GPY in 2010.<br />
The results are summarized in Figure 2. One can see that if OPEC and Eurasia act like cartels<br />
and there is only an annual 3% EV penetration in the selected regions, oil prices would remain<br />
relatively high for the next 10 years or so. The reasons are multiple: high demand growth in<br />
developing countries for at least the next 8 years, limited capacity expansion from non-cartel<br />
124
THE SOLAR-WIND-EV PARADIGM: A STRATEGIC POLICY FOR A SUSTAINABLE ENERGY FUTURE<br />
players, and little capacity expansion from the leaders. This case shows that a low EV penetration<br />
rate is not enough to bring down oil prices in the short run. As a result, oil demand remains<br />
relatively high, and energy security is still an issue for the U.S. for a while.<br />
4.2 Two leaders with 5% EV and 3% PHEV penetration<br />
In this case, we assumed an additional annual 3% PHEV penetration on top <strong>of</strong> the 5% EV<br />
penetration as described in case III.2. According to [33], each PHEV may consume 226.8 GPY <strong>of</strong><br />
gasoline by 2012, with an estimated fuel savings <strong>of</strong> 570-226.8 = 343.2 GPY. An extra 3% fuel<br />
reduction per year is also assumed for PHEVs beginning from 2012. For example, if 20 million<br />
cars and light trucks are sold in the U.S. in 2018, the gasoline savings from the 3% annual PHEV<br />
penetration would be 20million*(1.03^7-1)*343.2*(1.03^6)= 122,899 bpd (barrels per day) <strong>of</strong><br />
gasoline. The results are illustrated in Figure 2. The U.S. cumulative oil savings from both a 5%<br />
EV and a 3% PHEV penetration are illustrated in Figure 3. The total U.S. gasoline saving would<br />
be nearly 7.6 mbpd in 2024, which translates to at least 8.0 mbdp <strong>of</strong> crude oil saved. The savings<br />
would be about 60% <strong>of</strong> the current U.S. oil import (13.5 mbpd). Based on [40], one gallon <strong>of</strong><br />
gasoline emits about 5.3 pounds <strong>of</strong> CO 2 in the U.S. The CO 2 reduction from reduced oil import<br />
and consumption is then nearly 800,000 tons <strong>of</strong> CO 2 by 2024. Notice that we assume a 75%<br />
PHEV plug-in rate when oil prices are below $70/bbl (i.e., 25% <strong>of</strong> the PHEVs would not be<br />
plugged-in). Therefore, oil savings from PHEVs would be reduced somewhat after 2016.<br />
We can see that due to the additional PHEV penetration in new car and light truck sales<br />
starting from 2012, oil prices are further reduced. By the end <strong>of</strong> the study horizon, prices would<br />
be around the lower $50/bbl (or even below $50/bbl by 2022). The prices at this level would be<br />
welcomed by the consumers, who would also be assumed to have a positive attitude toward fuel<br />
efficiency (i.e., most <strong>of</strong> the consumers would be assumed to purchase EVs and PHEVs around<br />
2020).<br />
4.3 Coalition between OPEC and Eurasia without EVs and PHEVs<br />
In this case study, we assume that OPEC would form a coalition with Eurasia so that the two<br />
behave like a single cartel. We assume that there would be no EVs and PHEVs deployment. This<br />
scenario will serve as a reference and will show characterize extreme oil prices as depicted in<br />
Figure 2. We can see that if OPEC and Eurasia form a single cartel to coordinate their output, and<br />
if there is no EV and PHEV penetration while hybrid penetration is at the current level, the world<br />
oil prices would be over $160/bbl by 2022.<br />
US oil sales.<br />
125
ZUWEI YU, TARIK AOUAM<br />
$/bbl (2007$)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
III.1<br />
III.2<br />
III.3<br />
40<br />
20<br />
0<br />
2008 2010 2012 2014 2016 2018 2020 2022<br />
Year<br />
Figure 2. Plot <strong>of</strong> the simulated prices (The 2008 price is from EIA [32]).<br />
8000000<br />
bpd (oil import reduction)<br />
7000000<br />
6000000<br />
5000000<br />
4000000<br />
3000000<br />
2000000<br />
1000000<br />
0<br />
2010<br />
2011<br />
2012<br />
2013<br />
2014<br />
2015<br />
2016<br />
2017<br />
Year<br />
Figure 3. U.S. oil savings in case III.2.<br />
2018<br />
2019<br />
2020<br />
2021<br />
2022<br />
2023<br />
2024<br />
More case studies and scenarios can be done. However, except for extreme cases such as the<br />
closure <strong>of</strong> the Hormuz Straight, most other likely scenarios would fall between Cases III.1 and<br />
III.3. The price movement curves would fall between the curves <strong>of</strong> Cases III.1 and III.3.<br />
5. CONCLUSION<br />
U.S. policy makers have failed to include EVs in the national energy policy. As a result, federal<br />
support for EV research and development has been almost non-existent. Fortunately, private<br />
sectors have been leading R&D and production <strong>of</strong> EVs through venture investments over the<br />
126
THE SOLAR-WIND-EV PARADIGM: A STRATEGIC POLICY FOR A SUSTAINABLE ENERGY FUTURE<br />
past several years. It is time to revisit national and international energy policy in the era <strong>of</strong> global<br />
warming, increasing oil demand and production gap, and energy security. Several policy issues<br />
should be reevaluated.<br />
First, we notice that the infrastructure for EVs is not as massive as for HFCVs. The majority<br />
<strong>of</strong> the current power network can be used to charge EVs overnight due to a much lower power<br />
demand during the night than during the day. Parking lot charging facilities may be constructed<br />
near places such as <strong>of</strong>fices, shopping centers, sport and game areas, and even gas stations. These<br />
actions require regulatory reforms in the utility sector.<br />
Second, policy makers should help facilitate the research, demonstration and deployment <strong>of</strong><br />
EVs. The U.S. may need to redirect some funding from HFCVs to EVs for project demonstration.<br />
Third, bi<strong>of</strong>uel quotas should be reduced. In fact, bi<strong>of</strong>uel production has been causing food<br />
prices to rise globally without solving much <strong>of</strong> the U.S. oil problems. Some recent studies show<br />
that bi<strong>of</strong>uel may damage the environment. Expanded land cultivation causes carbon dioxide to<br />
escape from the soil into the atmosphere, leaving a carbon deficit from 17 years to more than 420<br />
years depending on the land cleared for ethanol production [37]. Several OECD countries have<br />
now either cut back or been considering cancelling subsidies to some bi<strong>of</strong>uels due to new<br />
findings on bi<strong>of</strong>uel-related problems [39].<br />
Finally, government should subsidize EVs. One <strong>of</strong> the issues policy should address is how to<br />
overcome the car market entry barriers.<br />
6. REFERENCES<br />
[1] D.W. Keith, A.E. Farrell, “Rethinking <strong>of</strong> Hydrogen Cars,” Science, Vol. 301, no. 5631, July 18, 2003.<br />
[2] D. Schneider, “Who’s Resuscitating the Electric Car?” American Scientist Magazine, Sept-Oct., 2007.<br />
[3] J. Grinffin, “OPEC Behavior: A Test <strong>of</strong> Alternative Hypothesis,” The American Economic Review,<br />
Vol. 75 1985, pp. 954-963.<br />
[4] S. Gülen, “Is OPEC Cartel? Evidence from Cointegration and Causality Tests,” The Energy Journal,<br />
April 1996.<br />
[5] S. Shaffer, “Stable Cartel with a Cournot Fringe,” Southern Economic Journal, Vol. 61, 1995, pp. 744-<br />
754.<br />
[6] E. Diamantoudi, “Stable Cartel Revisited,” Economic Theory, vol. 26(4), 2005, pp. 907-921.<br />
[7] A. Tuzuner, Z. Yu, “A Stochastic Gaming Framework for Analyzing Market Potentials and Risks <strong>of</strong><br />
CTL,” in Proceedings <strong>of</strong> the Pittsburgh Coal Conference, Sept. 2006.<br />
[8] Z. Yu, B. Bowen, “Integrated Energy Resource Planning Approach to the U.S. Energy Security<br />
Issues,” Oil, Gas, Energy & Law, Nov. 2007.<br />
[9] Z. Yu, F.T. Sparrow, T. Morin, G. Nderitu, "A Stackelberg Price Leadership Model for Deregulated<br />
Electricity Markets," Proceedings <strong>of</strong> the IEEE PES Winter Meeting, Singapore, January 2000.<br />
[10] Z. Yu, P. Preckel, F.T. Sparrow, D. Nderitu, “A Strategic Power Market Model with CO2 Limits,”<br />
International Journal <strong>of</strong> Electric Power & Energy Systems, Vol. 23, no. 6, 2001.<br />
[11] B. F. Hobbs, “Linear complementarity models <strong>of</strong> Nash–Cournot competition in bilateral and<br />
POOLCO power markets,” IEEE Trans. Power Syst., vol. 16, no. 2, pp. 194–202, May 2001.<br />
[12] Penn Well, “Solar Power - Utility-scale sun power,” Power Engineering, June 2007.<br />
[13] Pacific Northwest National Laboratory, “An Assessment <strong>of</strong> the Available Windy Land Area and<br />
Wind Energy Potential in the Contiguous United States,” a reported prepared for the Department <strong>of</strong><br />
Energy (DOE), 1991.<br />
[14] PACE, “Assessment <strong>of</strong> Offshore Wind Power Resources,” prepared for Long Island Power<br />
Authority, August 22, 2007.<br />
127
ZUWEI YU, TARIK AOUAM<br />
[15] P. Vadasz, “Compressed Air Energy Storage: Optimal Performance and Techno-Economical<br />
Indices,” Int. J. Applied Thermodynamics, Vol.2, No.2, pp. 69-80, June-1999.<br />
[16] J. Greenblatt, S. Succar, D. Denkenberger, R. Williams, “Toward optimization <strong>of</strong> a wind/compressed<br />
air energy storage (CAES) power system,” Electric Power Conference, Baltimore, MD, March 30-<br />
April 1, 2004.<br />
[17] P. Fairley, “California to Rule On Fate <strong>of</strong> EVs,” IEEE Spectrum, Nov. 2007.<br />
[18] J. Schindall, “The Charge <strong>of</strong> Ultracapacitors,” IEEE Spectrum, Nov. 2007, pp. 42-46.<br />
[19] A. Westervelt, B. Brun, “Two Batteries,” published by The Sustainable Industries, Jan. 4, 2008.<br />
[20] U. Bossel, “Efficiency <strong>of</strong> Hydrogen Fuel Cell, Diesel-SOFC-Hybrid & Battery Electric Vehicles,”<br />
European Fuel Cell Forum, Oct. 20, 2003.<br />
[21] K. Wipke, et al., “Learning Demonstration Interim Progress Report –Summer 2007,” NREL/TP-560-<br />
41848, July 2007.<br />
[22] Stephen Eaves, and James Eaves, “A cost comparison <strong>of</strong> fuel-cell and battery electric vehicles,”<br />
Journal <strong>of</strong> Power Sources, Vol. 130, Issues 1-2, May 3 2004, pp. 208-212.<br />
[23] GAO, “Hydrogen Fuel Initiative,” Government Accountability Office: Report to Congressional<br />
Requesters, GAO-08-305, Jan. 2008.<br />
[24] Z. Yu, “Wind power and transportation: A bridge to energy security,” Proceedings <strong>of</strong> the IEEE PES<br />
eit conference, May 18-21, 2008, Iowa State University.<br />
[25] D. Guthier-Villars, “Electric Cars Gather Speed,” The Wall Street Journal, March 4, 2008.<br />
[26] M. Kintner-Meyer, K. Schneider, R. Pratt, “IMPACTS ASSESSMENT OF PLUG-IN HYBRID<br />
VEHICLES ON ELECTRIC UTILITIES AND REGIONAL U.S. POWER GRIDS,” a report prepared<br />
for DOE, Pacific Northwest National Laboratory, Nov. 2007.<br />
[27] HR 6, “The Energy Independence and Security Act <strong>of</strong> 2007,” the U.S. House <strong>of</strong> Representatives,<br />
Dec. 2007.<br />
[28] CAPP, “Canadian Oil Sands Outlook,” a report prepared by the Canadian Association <strong>of</strong> Petroleum<br />
Production, Sept. 2007.<br />
[29] Z. Yu, T. Aouam, “A Monte Carlo Gaming Model for Analyzing the 21C World Oil Markets with<br />
PHEVs,” a book chapter accepted by the NOVA Science Publishers, to appear in 2008.<br />
[30] “Bi<strong>of</strong>uel Market Worldwide,” a report prepared by Biotechnology, July 2006.<br />
[31] EIA, “The 2007 Annual Energy Outlook,” released in June 2007.<br />
[32] EIA, “The Short-Term Energy Outlook 2008,” released in Jan. 2008.<br />
[33] EPRI (Electric Power Research Institute), “Environmental Assessment <strong>of</strong> Plug-In Hybrid Electric<br />
Vehicles,” a report prepared for DOE, July 2007.<br />
[34] NREL (National Renewable Energy Laboratory), “Field Operations Program - Overview <strong>of</strong><br />
Advanced Technology Transportation,” a report prepared for the U.S. DOE, July 2000.<br />
[35] J. Olsen, “Electric Car System Planned in Denmark,” The Associated Press, March 27, 2008.<br />
[36] T.K. Tromp, R. Shia, M. Allen, J.M. Eiler, and Y.L. Yung, “Potential Environmental Impact <strong>of</strong> a<br />
Hydrogen Economy on the Stratosphere,” Science 13 June 2003: Vol. 300. no. 5626, pp. 1740 – 1742<br />
[37] J. Fargione, J. Hill, D. Tilman, S. Polasky, P. Hawthorne, “Land Clearing and the Bi<strong>of</strong>uel Carbon<br />
Debt,” Science online publication, Feb. 7, 2008.<br />
[38] P.J. Crutzen, A.R. Mosier, K.A. Smith, and W. Winiwarter, “N 2 O release from agro-bi<strong>of</strong>uel<br />
production negates global warming reduction by replacing fossil fuels,” Atmospheric Chemistry and<br />
Physics Discussion, Vol. 7, 11191-11205, 2007.<br />
[39] E. Rosenthal, “Europe, Cutting Bi<strong>of</strong>uel Subsidies, Redirects Aid to Stress Greenest Options,” The<br />
New York Times, Jan. 22, 2008.<br />
[40] J. DeCicco, F. Fung, “Global Warming on the Road – The Climate Impact <strong>of</strong> American’s<br />
Automobiles,” Environmental Defense, 2006.<br />
[41] GAMS – The Solver Manuals, GAMS Development Corporation, 2004.<br />
128
ALHOSN UNIVERSITY JOURNAL<br />
OF ENGINEERING AND APPLIED SCIENCES<br />
GENERAL INFORMATION<br />
AHU <strong>journal</strong> <strong>of</strong> Engineering and Applied Sciences<br />
is a refereed <strong>journal</strong>, published bi annually by<br />
ALHOSN University. The aim <strong>of</strong> <strong>of</strong> the <strong>journal</strong><br />
is to emphasize the use <strong>of</strong> <strong>engineering</strong> design and<br />
analysis, to publish a high quality research, to<br />
strike a balance between research and application<br />
and to promote international dialogue and<br />
collaboration. The <strong>journal</strong> welcomes scholarly<br />
submissions <strong>of</strong> original and innovative articles<br />
in all fields <strong>of</strong> <strong>engineering</strong> and <strong>applied</strong> <strong>sciences</strong>.<br />
These include mutually but not exclusively,<br />
articles pertaining to the reporting <strong>of</strong> advances in<br />
theory, techniques, methodologies, applications,<br />
short communication, review articles and book<br />
reviews.<br />
The <strong>journal</strong> is published bi annually in June<br />
andDecember each year. Special issues may be<br />
published from time to time. A guest editor may<br />
be invited to edit a special issue and he/she will be<br />
responsible for the quality <strong>of</strong> all papers selected<br />
for publication.<br />
Submission <strong>of</strong> manuscript for publication will be<br />
considered to imply a tacit understanding that it<br />
has not been previously published is not currently<br />
considered for publication and will not be sent for<br />
publication elsewhere.<br />
3. Articles should be served in word format and<br />
emailed to the editor with a cover page stating<br />
name, email address and where applicable,<br />
academic affiliation. If a submission includes<br />
graphics and pictures, this should be served<br />
in jpg format and sent as a separate attachment<br />
along the submission.<br />
4. Articles must be doubled spaced and in a<br />
single column format.<br />
5. Instructions on the preparation <strong>of</strong> tables,<br />
figures, references and abstracts should<br />
follow the IEEE guidelines for publication.<br />
at: http://www.ieee.org/publications_standards/<br />
publications/authors/authors_<strong>journal</strong><br />
6. The author <strong>of</strong> articles will be supplied with<br />
one free hard copy <strong>of</strong> the <strong>journal</strong> in which his/<br />
her paper has appeared by airmail soon after<br />
the <strong>journal</strong> is published<br />
129
130