16.05.2015 Views

D5 Annex report WP 3: ETIS Database methodology ... - ETIS plus

D5 Annex report WP 3: ETIS Database methodology ... - ETIS plus

D5 Annex report WP 3: ETIS Database methodology ... - ETIS plus

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>D5</strong> <strong>Annex</strong> <strong>report</strong> <strong>WP</strong> 3: <strong>ETIS</strong>­ <strong>Database</strong><br />

<strong>methodology</strong> development and database user<br />

manual– Freight transport demand V2.0<br />

CONTRACT N° : GMA2/2000/32051­SI2.335713 <strong>ETIS</strong>­BASE<br />

PROJECT N° : 2.1.1/9<br />

ACRONYM : <strong>ETIS</strong>­BASE<br />

TITLE : Core <strong>Database</strong> Development for the European Transport policy Information System (<strong>ETIS</strong>)<br />

PROJECT CO­ORDINATOR : NEA Transport Research and Training BV<br />

PARTNERS :<br />

­ Nouveaux Espaces de Transport en Europe Application de Recherche<br />

­ Istituto di studi per l’integrazione dei sistemi<br />

­ Universität Karlsruhe (TH)<br />

­ MDS Transmodal Limited<br />

­ MKmetric Gesellschaft Fuer systemplannug MBH<br />

­ Technical Research Centre of Finland<br />

­ Eidgenoessische Technische Hochschule Zuerich<br />

PROJECT START DATE: 1­12­2002<br />

Report reference number: R20030249<br />

Date of issue of this <strong>report</strong>: 27­05­2004<br />

DURATION : 33 Months<br />

Project funded by the European Community under the<br />

‘Competitive and Sustainable Growth’ Programme<br />

(1998­2002)


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

CONTENTS<br />

page<br />

1 INTRODUCTION ...............................................................................7<br />

2 OBJECTIVES AND STRATEGIC ASPECTS.....................................9<br />

3 SETTING THE FRAMEWORK ........................................................11<br />

3.1 Introduction...............................................................................................11<br />

3.2 Supporting indicators assigned to <strong>WP</strong> 3 Freight Transport Demand and<br />

method of calculation ................................................................................11<br />

3.3 Variables to be included in <strong>WP</strong> 3 Freight transport demand .......................12<br />

3.4 Variables required for calculation of supporting indicators that will not be<br />

included in the <strong>ETIS</strong> reference database.....................................................16<br />

4 THE FREIGHT OD DATA MODEL ..................................................19<br />

4.1 Introduction...............................................................................................19<br />

4.2 General structure .......................................................................................19<br />

4.3 Combining trade/transport/transhipment data sources ................................19<br />

4.4 Data availability and reliability; an example ..............................................20<br />

4.5 The top­down approach .............................................................................21<br />

4.5.1 Phase I The construction of a country­to­country matrix ............................21<br />

4.5.2 Phase II Including transhipment regions on the basis of transhipment<br />

statistics ....................................................................................................23<br />

4.5.3 Phase III Regional division of country­to­country totals.............................23<br />

4.5.4 Phase IV Incorporating domestic transport.................................................24<br />

4.6 Data gaps identified...................................................................................25<br />

4.7 Estimating data gaps..................................................................................26<br />

4.8 Adding information to the freight OD matrix .............................................26<br />

4.8.1 Cargo types and characteristics..................................................................26<br />

4.8.2 Containerisation ........................................................................................27<br />

4.8.3 Number of transport units (vehicles/vessels) by type..................................27<br />

4.8.4 Number of TEUs .......................................................................................28<br />

4.8.5 Transport performance information (tonne­km, vehicle­ km/vessel­km,<br />

TEU­ km)..................................................................................................29<br />

Document2<br />

27 May 2004<br />

3


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

5 TRANSPORT CHAIN FREIGHT O/D MATRIX – DATA NEEDS,<br />

DATA COLLECTION AND REMAINING DATA GAPS ..................31<br />

5.1 Introduction............................................................................................... 31<br />

5.2 Data needs................................................................................................. 31<br />

5.2.1 Trade data ................................................................................................. 31<br />

5.2.2 Transport data ........................................................................................... 32<br />

5.2.3 Maritime transhipment data .......................................................................33<br />

5.2.4 Container data ........................................................................................... 34<br />

5.2.5 Vehicle/vessel movement data...................................................................34<br />

5.2.6 Relation with other datasets in the <strong>ETIS</strong> reference database ....................... 34<br />

5.3 Data collection .......................................................................................... 36<br />

5.4 Remaining data gaps .................................................................................40<br />

5.4.1 Data available at Eurostat not available for the <strong>ETIS</strong> project ...................... 40<br />

5.4.2 Registration of mode of transport for intra­EU trade becomes optional in<br />

Eurostat trade statistics .............................................................................. 40<br />

5.4.3 Transhipment data of seaports and inland terminals ...................................40<br />

5.4.4 Data availability in the accession countries ................................................ 41<br />

6 DESCRIPTION OF THE <strong>WP</strong> 3 TESTING PHASE AND ITS<br />

RESULTS ........................................................................................43<br />

6.1 General description of the testing phase..................................................... 43<br />

6.2 Tested methodologies................................................................................ 43<br />

6.3 Lessons learned and changes compared to the original planned <strong>methodology</strong>64<br />

6.4 Output of the testing phase ........................................................................ 64<br />

6.5 Interpretation of the output data for the users .............................................65<br />

7 OBSERVATIONS UP TO PHASE 2.................................................67<br />

7.1 Introduction............................................................................................... 67<br />

7.2 Legal aspects and organisational aspects.................................................... 67<br />

7.3 Other remarks............................................................................................67<br />

ANNEX A: LIST OF INDICATORS CONSIDERED IN <strong>WP</strong> 3...................................69<br />

ANNEX B: INDICATOR COMPILATION TEMPLATES ..............................................77<br />

ANNEX C: EXPERIENCE FROM CONCERTED ACTION ON SHORT SEA SHIPPING<br />

PROJECT ............................................133<br />

ANNEX D: THEORETICAL ANALYSIS OF TRANSPORT CHAIN STRUCTURES IN<br />

DATABASES; CONSTRAINTS AND POSSIBILITIES ..........................149<br />

4<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

ANNEX E: NEWCRONOS TABLES (RELEVANT FOR <strong>ETIS</strong>)..................................165<br />

Document2<br />

5<br />

08 October 2009


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

1 INTRODUCTION<br />

The present <strong>report</strong> represents the <strong>WP</strong> 3 annex <strong>report</strong> of deliverable <strong>D5</strong> of the reference database<br />

development within the <strong>ETIS</strong> project 1 . The present document describes the <strong>ETIS</strong>­<strong>Database</strong><br />

<strong>methodology</strong> development and is the database user manual for the freight transport demand data<br />

set, which is being developed within <strong>WP</strong> 3.<br />

The objectives of the methodological <strong>report</strong> can be characterised as follows:<br />

· Determine the variables to be included<br />

· Describe the <strong>methodology</strong> to obtain the variables and to fill the remaining gaps<br />

· List the sources being used and document the status of data purchase<br />

· Describe the scope of and <strong>methodology</strong> for the pilot data set to be generated for freight<br />

transport demand indicators.<br />

· Describe the testing phase and its findings.<br />

This annex <strong>report</strong> and the annex <strong>report</strong>s of the other <strong>WP</strong>s are summarised in the synthesis <strong>report</strong><br />

of <strong>D5</strong>.<br />

1 The full title of the reference database part of <strong>ETIS</strong> project is <strong>ETIS</strong>­BASE ­ `Core <strong>Database</strong> Development for the European<br />

Transport policy Information System (<strong>ETIS</strong>)’<br />

Document2<br />

27 May 2004<br />

7


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

2 OBJECTIVES AND STRATEGIC ASPECTS<br />

The work on the <strong>ETIS</strong> reference database responds to key action 2, ‘Sustainable Mobility and<br />

Inter­modality’, objective 2.1 ‘Socio­economic scenarios for mobility of people and goods’,<br />

task 2.1.1/9, ‘Development of a European Transport policy Information System (<strong>ETIS</strong>) as a<br />

basis for transport planning and policy formulation’. The task is separated into three sub­tasks.<br />

The <strong>ETIS</strong> reference database addresses sub­task 2, ‘the development of a reference database for<br />

the modelling element’.<br />

The objectives of <strong>ETIS</strong> reference database are:<br />

1. To contribute to the building of a consensus view of the reference pan European transport<br />

modelling data set.<br />

2. To develop an open <strong>methodology</strong> to generate a version of such a set from existing<br />

international and national sources.<br />

3. To produce a first compilation of the data set by applying the <strong>methodology</strong> mentioned<br />

above, as on­line database.<br />

During the kick­off of the project it has been decided by the European Commission that within<br />

this project the work should focus on:<br />

1. the development of an <strong>ETIS</strong> for TEN­T policies,<br />

2. the procedures and data should face especially a monitoring of the TEN­T corridors,<br />

3. the geographic scope has to be adjusted to the forthcoming 10 new members,<br />

4. the PAN European scope has to be defined along the geographic hemispheres,<br />

5. the degree of detail in general can be reduced including a concentration upon a few<br />

indicators mentioned in the white paper,<br />

6. the results have to be available for further use within the G<strong>ETIS</strong> system,<br />

7. the work tasks and responsibilities have to be adjusted in respect of the new focus and the<br />

limited budget.<br />

The reference database of <strong>ETIS</strong> will serve a various series of TEN­T policy issues: the level of<br />

detail and the variables will be appropriate for this purpose. It will allow obtaining in an<br />

accurate way the performance and the impacts (environmental, economic) of transport, as well<br />

as the traffic at specific nodes or links of the networks. But the internal structure of the database<br />

will allow proceeding easily to any aggregation in order to get a compact view of transport<br />

performance (vehicle­kms etc) and effects (emissions level, energy consumption by mode etc).<br />

The work organisation of <strong>ETIS</strong> reference database is established in close co­operation with the<br />

external promotion 2 and contacts of <strong>ETIS</strong>. As part of this external promotion, support to the<br />

development of the <strong>ETIS</strong> reference database is an essential element.<br />

There are several aspects on which synchronisation of the two projects take place:<br />

2 Promotion work is covered by the <strong>ETIS</strong>­LINK project<br />

Document2<br />

27 May 2004<br />

9


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

· 4 workshops specifically related to the development of the <strong>ETIS</strong> reference database<br />

· Open Conferences in which dissemination of results takes place<br />

· Participation to the <strong>ETIS</strong> Steering group<br />

· Testing of the system with pilot users, i.e. also testing of results of the development of the<br />

<strong>ETIS</strong> reference database<br />

· Dealing with specific issues like legal and organisational aspects.<br />

In addition the <strong>ETIS</strong> reference database will be incorporated in the <strong>ETIS</strong> software tools 3 . The<br />

two most important outputs of the reference database development that serve as input to the<br />

system tools being developed are:<br />

1. The metadata concerning indicators and data sources serve.<br />

2. The final reference <strong>ETIS</strong> database<br />

Furthermore in order to make it possible for the software tool developers to continue their work<br />

while the reference database is being developed working material is being delivered which has<br />

also been used in the TEN­STAC project. Intermediate results from the reference database<br />

development will be delivered as soon as it comes available. <strong>ETIS</strong> reference database<br />

construction will, where possible, use the results of G<strong>ETIS</strong> (GIS data) and TEN­STAC<br />

(indicator definitions and use of a selection of the input data) and find co­operation where<br />

possible.<br />

The <strong>ETIS</strong> reference database project, <strong>ETIS</strong> promotion and external contacts project and <strong>ETIS</strong><br />

software tools project 4 have come up with a common and better harmonised focus during<br />

meetings from November 2003 up to January 2004 in order to be able to come up at the end of<br />

the three projects with one consistent and unique <strong>ETIS</strong> product. This product is a pilot and it is<br />

expected that one will continue to maintain and to develop the <strong>ETIS</strong> tool to the interest of the<br />

European transport policy.<br />

3 <strong>ETIS</strong> software tools are covered by the <strong>ETIS</strong>­AGENT project<br />

4 respectively <strong>ETIS</strong>­BASE, <strong>ETIS</strong>­LINK and <strong>ETIS</strong>­AGENT, the trilogy of <strong>ETIS</strong> projects<br />

10<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

3 SETTING THE FRAMEWORK<br />

3.1 Introduction<br />

In this chapter a full description of the key elements of the Work Package data and outcomes are<br />

provided. The emphasis, due to its importance in the overall database, is devoted to the<br />

indication of the scale and time scale dimensions of data collected and method of calculation of<br />

supporting indicators.<br />

3.2 Supporting indicators assigned to <strong>WP</strong> 3 Freight Transport Demand and<br />

method of calculation<br />

In table 3.1 the supporting indicators that are assigned to <strong>WP</strong> 3 have been listed. These are<br />

supporting indicators related to TEN policies and specifically related to freight transport<br />

demand. More information on these supporting indicators are listed in <strong>Annex</strong> A.<br />

Table 3.1:<br />

TEN policy supporting indicators considered in <strong>WP</strong> 3 Freight transport<br />

demand<br />

Domain ref. Definition<br />

Mobility 1.1.4<br />

Breakdown of journeys by origin/destination, proportion of long distance traffic using the<br />

TEN<br />

Mobility 1.5.1 Road freight volumes on TEN by cargo type<br />

Mobility 2.1.2<br />

Mobility 2.5.1 Freight volumes on TEN by train type<br />

Traffic volumes on the Trans­European rail network, by type (passenger / freight),<br />

including non fulfilled demand<br />

Mobility 4.1.2 Freight volumes on the inland waterway network<br />

Mobility 5.1.1 Port throughput (passengers, freight)<br />

Capacity 4.1.3 I/C factor of bridges and locks (intensity versus capacity) (waiting time)<br />

Utilisation 6.1.1 Traffic volumes served at the terminal<br />

Energy consumption 1.3.1 Estimated / measured energy consumption on roads<br />

Energy consumption 2.3.1 Estimated/measured energy consumption on railways<br />

Energy consumption 4.2.1 Estimated/Measured Energy Consumption on waterways<br />

Transport emissions 1.3.2<br />

Transport emissions 2.3.2<br />

Transport emissions 4.2.2<br />

Estimated /measured road transport emissions<br />

+corresponding marginal unit cost<br />

Estimated /measured rail transport emissions<br />

+corresponding marginal unit cost<br />

Estimated /measured Inland waterway transport emissions<br />

+corresponding marginal unit cost<br />

Transport noise 1.3.3 Noise levels generated by road transport + corresponding marginal unit cost<br />

Transport noise 2.3.3 Noise levels generated by rail transport +corresponding marginal unit cost<br />

Investments and return<br />

on capital<br />

3.2.2 Amount of investment made in the development and maintenance of air traffic control<br />

Document2<br />

27 May 2004<br />

11


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

In <strong>Annex</strong> B for each supporting indicator a template is included in which the method of<br />

calculation and required variables, modelling, model input and other characteristics are<br />

described.<br />

3.3 Variables to be included in <strong>WP</strong> 3 Freight transport demand<br />

In this section the variables are listed that are needed for the calculation of the supporting<br />

indicators described in section 3.2. Furthermore these variables are selected in such a way that<br />

the database will be flexible enough to cover also future changes to the indicators and to make it<br />

as useful as possible for modelling and forecasting. It should also provide a proper basis for<br />

comparison between modes and sub­markets. For this reason it should be mentioned that the<br />

requirements for the indicators listed have been considered as a minimum requirement and do<br />

not reflect one to one the selected variables to be considered in this work package.<br />

Freight OD matrix<br />

Many of the TEN policy related indicators listed in section 3.2 are directly or indirectly linked<br />

to OD transport flows. For instance information about transport volumes on specific links on the<br />

infrastructure network can be produced by applying an assignment where the assignment<br />

procedure needs as input the freight OD matrix. This also holds for many of the other indicators<br />

such as noise levels or energy consumption. Therefore the freight OD matrix has first priority.<br />

Once the freight OD matrix is available, other information and data can be added to it and<br />

indicators can be derived from it.<br />

It has been chosen to build a region to region (also cross border) transport chain structure in<br />

which change of mode and transhipment location are included. A theoretical analysis on the use<br />

of transport chains is given in annex D and in chapter 4 the methods to be used to construct the<br />

transport chain database are described.<br />

A record structure will be applied with two transhipment locations and three modes (from<br />

experience it appears that a record structure with more transhipment locations is not feasible,<br />

see also annex D). The transport chain matrix will then have the following structure:<br />

· Origin<br />

· Transhipment 1<br />

· Transhipment 2<br />

· Destination<br />

· Mode at origin<br />

· Mode between transhipment points (active mode)<br />

· Mode between transhipment points (passive mode)<br />

· Mode at destination<br />

· Commodity<br />

· Cargo type<br />

· Unitised / non­unitised<br />

12<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

· Other typology<br />

· Measuring unit<br />

For these variables the characteristics will be considered as listed in table 3.2.<br />

Table 3.2:<br />

Variables considered in the freight OD transport chain matrix<br />

Core countries<br />

EU­15, Norway, Switzerland, Estonia, Latvia, Lithuania, Poland, Czech<br />

Republic, Slovak Republic, Hungary, Slovenia, Malta, Cyprus<br />

Regional detail NUTS 2 or similar regional detail where no NUTS classification is valid (#)<br />

Country and country<br />

group detail<br />

Transhipment location<br />

Modes<br />

Commodities<br />

Cargo types<br />

Cargo characteristics<br />

Unitised<br />

Other Typologies<br />

Measuring units<br />

All European countries separate with exception of the smallest (like Andorra,<br />

Vatican, etc), Meda countries separate, USA, Rest North America, Middle<br />

and South America, Japan, Rest Asia, Rest Africa, Australia and New<br />

Zealand, Rest world<br />

Selection of Ports<br />

Road, rail, inland navigation, maritime, rest<br />

Distinction between active and passive mode (*) for maritime transport (if<br />

required data is available)<br />

NSTR 2 digits (as much as possible), if no data is available or when<br />

modelling becomes necessary aggregation to NSTR 1 digit<br />

General cargo, liquid bulk, semi bulk, dry bulk, vehicles, crude oil<br />

Hazardous, conditioned<br />

Yes/No, share of unitised flows in the total flows<br />

Vehicle/vessel types (definition depending on data availability)<br />

Values<br />

Tonnes<br />

Base year 2000<br />

Tonne­km<br />

Number of vehicles / vessels<br />

Vehicle­km / vessel­km<br />

TEU<br />

TEU­km<br />

(#)<br />

The NUTS2 classification forms the basis for the regional division. For some countries the<br />

NUTS2 classification has only one region (the country). If in such cases the NUTS3<br />

classification has more than one region and if data on this regional level of detail is available,<br />

the regional division according to the NUTS3 classification will be used as much as possible.<br />

(*)<br />

Definition of transport with active and passive mode: transport of goods using two modes of<br />

transport in combination, where one (passive) transport means is carried on another (active)<br />

transport means which provides traction and consumes energy (for instance truck (passive mode<br />

road) on a train (active mode rail)).<br />

Document2<br />

27 May 2004<br />

13


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Remarks<br />

Concerning the regional detail it can be remarked that NUTS2 (or NUTS3) can only be<br />

achieved where the data availability allows. In many cases estimation models can be applied<br />

requiring more aggregate regional data and/or socio­economic data. In the remaining cases the<br />

country level will be used.<br />

Concerning the modes it can be remarked that if sufficient data becomes available active and<br />

passive mode use can be applied. In case of transhipment several modes can be found in one<br />

transport chain.<br />

Transhipment in ports will be feasible (has been proven to work in NEAC and projects like<br />

INFOSTAT, MESUDEMO, CONCERTO and ALPNET) but the number of ports is depending<br />

on the availability of data in the ports.<br />

The commodities will be kept as much as possible at the NSTR 2 digit level. However, some<br />

data sources only have information on NSTR 1 digit level. Furthermore, existing estimation<br />

procedures are only available on a NSTR 1 digit level since otherwise no reliable result can be<br />

obtained.<br />

Also cargo type can not be collected from all sources. Here estimation by using translation<br />

tables will be needed.<br />

In order to be able to translate volumes into vehicle movements additional information is<br />

needed. Here information on vehicle/vessel type and empty vehicle movements will be used as<br />

input. For the empty vehicles/vessels it will be attempted to construct a database which will be<br />

separate from the transport chain database. Furthermore some general figures on loading factor<br />

is needed. The question of translating commodity flows expressed in tonnes into vehicle flows<br />

expressed in units follows on from the estimation of transport chains, as it depends upon the<br />

mode of transport being known. Estimates of vehicle numbers, particularly within the road<br />

mode provide an important input for calculations involving external effects.<br />

From this transport chain structure where the commodity is followed from origin to destination<br />

it can also become clear which flows are deep sea and which short sea. This is done by looking<br />

at the location of origin or transhipment and destination or transhipment.<br />

If the value is available in the database also the share of low and high valued goods can be<br />

identified.<br />

In an ideal situation where all required information is available the freight OD matrix as<br />

described above in this section can be completely filled. It is expected that in practice not all<br />

information will be available. Therefore the contents and the level of detail included in the final<br />

14<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

freight OD matrix depend on the availability of data and on the possibilities to estimate data<br />

gaps.<br />

Possibilities for extracting sub­databases<br />

Once such a transport chain has been developed it is possible to extract more specific subdatabases<br />

from it. Here are some examples listed.<br />

Trade matrices:<br />

· Origin of commodities<br />

· Destination of commodities<br />

· Mode at origin<br />

· Mode at destination<br />

· Commodity<br />

· Cargo type<br />

· Containerised<br />

· Other typology<br />

· Measuring unit<br />

or<br />

· Origin of commodities<br />

· Destination of commodities<br />

· Commodity<br />

· Cargo type<br />

· Containerised<br />

· Other typology<br />

· Measuring unit<br />

Transport matrices:<br />

· Origin of vehicles/vessels<br />

· Destination of vehicles/vessels<br />

· Mode<br />

· Commodity<br />

· Cargo type<br />

· Containerised<br />

· Other typology<br />

· Measuring unit<br />

Document2<br />

27 May 2004<br />

15


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Transhipment matrices:<br />

· Transhipment location<br />

· Origin of commodities<br />

· Destination of commodities<br />

· Mode incoming<br />

· Mode outgoing<br />

· Commodity<br />

· Cargo type<br />

· Containerised<br />

· Other typology<br />

· Measuring unit<br />

Or<br />

· Transhipment location<br />

· Incoming/outgoing/transit<br />

· Commodity<br />

· Cargo type<br />

· Containerised<br />

· Other typology<br />

· Measuring unit<br />

Time series<br />

Besides the OD matrix only aggregate terminal and port figures are needed as can be concluded<br />

from the indicator descriptions. Due to the complexity of the construction of the OD­matrix it is<br />

only possible to develop a base year 2000 OD matrix.<br />

Furthermore we will extract existing material from EUROSTAT describing other freight<br />

demand aggregate variables where a lot of data has already been harmonised. In annex E the<br />

relevant tables from New Cronos for <strong>ETIS</strong> have been listed.<br />

3.4 Variables required for calculation of supporting indicators that will not be included<br />

in the <strong>ETIS</strong> reference database<br />

During the test phase of the development of a freight O/D matrix two things have become clear<br />

in relation with the transhipment at inland terminals:<br />

· The required data is not available for inland terminals (this conclusion has already been<br />

drawn in other projects and it also became clear during the data collection in the <strong>ETIS</strong><br />

project);<br />

16<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

· Methods can be applied to fill the above mentioned data gap by estimating the data.<br />

These methods are too complex and too time consuming to fit in the scope of the <strong>ETIS</strong><br />

project.<br />

As a result, it has been decided to leave information about transhipment at inland terminals out<br />

of the freight O/D matrix. Information about transhipment included in the freight O/D database<br />

concerns solely transhipment at seaports.<br />

Document2<br />

27 May 2004<br />

17


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

4 THE FREIGHT OD DATA MODEL<br />

4.1 Introduction<br />

In this chapter the proposed freight OD transport chain data model is described. The exact<br />

structure of the type of model to be used is depending on the specific structure of the available<br />

data. For this reason not all details but merely the basic principles and available tools will be<br />

discussed. The exact model in full detail will be developed along the construction process in the<br />

remaining of the project. Some results obtained in the testing phase are already described in<br />

chapter 6.<br />

4.2 General structure<br />

The general structure of this model to be used can be split up into the following parts:<br />

1. Combining trade/transport/transhipment data sources<br />

2. Estimating data gaps<br />

3. Adding information to the OD matrix<br />

In the first part all available trade and transport data sources are being combined into one<br />

database by a top­down approach. This approach will have the same structure as the NEAC<br />

database construction model. It might be the case that some regional information is missing in<br />

transport or trade data or that no data are available at all for a specific geographical area. In this<br />

case estimation methods are required that can estimate data by modelling. Next procedures will<br />

be applied to add information to the OD matrix. Container information, cargo type information<br />

and characteristics on the OD flows will be estimated. Also estimation procedures will be<br />

applied for the translation from tonnes to vehicles/vessels and if possible by type, translating<br />

containerised tonnes into TEUs. Finally all this information will be used to determine transport<br />

performance information (tonnekm, vehiclekm/vesselkm and TEUkm). This will be described<br />

in more detail in the following sections.<br />

4.3 Combining trade/transport/transhipment data sources<br />

The first step in the top­down approach is to combine all collected trade/transport sources into<br />

one database. Here we use the trade data as the foundation of the whole database and refine this<br />

with information that can be found in the transport and transhipment data sources. In the<br />

INFOSTAT project a pilot database has been constructed with the top­down approach. As<br />

described in paragraph 6.4 there are some restrictions on data combinations. In INFOSTAT a<br />

database with two transhipment points has been constructed. A conclusion was that for all<br />

elements of the chain, data should be available to ensure quality of the results.<br />

Up to this point two transhipment points in the chain structure by combining sources seems to<br />

be the limit on the level of detail.<br />

Document2<br />

27 May 2004<br />

19


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

A top­down database construction method (see section 4.5) for combining<br />

trade/transport/transhipment data sources is proposed which is based on the NEAC as developed<br />

by NEA and which was also used in INFOSTAT.<br />

4.4 Data availability and reliability; an example<br />

The methods used in the top­down approach are very much dependent on the source data that<br />

are available. These sources can have very different structures (chain, O/D, aggregated flows<br />

through a node) and the transport flows described can be based on different definitions (trade,<br />

transport, only national carriers, etc). When developing a model we can choose different<br />

philosophies. We could for instance develop one robust model that requires the same level of<br />

detail for all data sources. The advantage is that the application of the model is relatively easy.<br />

The big disadvantage is that we will loose a lot of useful information since the model will be<br />

directed to the worst case of the available data sources. The other extreme is to develop a<br />

method for inclusion of each data source separately (NEAC database construction model). The<br />

advantage is that all information incorporated in the data sources can be used to the maximum.<br />

The disadvantage of this method is that it is very costly especially when many sources have to<br />

be considered and the reliability of transport flows may vary between different OD relations<br />

since some originate from observed data and others from estimation procedures.<br />

In the <strong>ETIS</strong> reference database model we have to take account of the fact that we will have to<br />

deal with quite a lot of data sources. We therefore have to look for an efficient top­down<br />

method that can make use of as much information in the sources as possible, but is not being<br />

held back by the poorest source.<br />

Each of the possible methods has its pros and cons. In any case it can be stated that only reliable<br />

information can be obtained from database construction if for all elements of the transport chain<br />

information is available. If this is not the case then some parts or elements of the database will<br />

be less reliable.<br />

Since different fragments of the total transport chain will be collected that have to be matched to<br />

each other it may be unsure whether the total constructed records will be reliable. In the<br />

INFOSTAT project (task 3 <strong>WP</strong> E) attempts have been made to include a second transhipment point<br />

in the record structure for flows of the European continent in relation with Norway. In this pilot<br />

case transhipment data was available for Belgium, the Netherlands, Germany and France on the<br />

continent and in Norway. A method was developed to combine a transport chain database for the<br />

continent with a transport chain database for Norway. Unfortunately by that time no port to port<br />

information was available which could be used to match the Norwegian ports and the ports on the<br />

continent. As a result the link between the ports in the eventual database was not reliable.<br />

Analysing either of the transhipment points in relation with all other variables except the other<br />

transhipment point is reliable however. The advantage of the eventual database is that all figures<br />

20<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

are consistent. In case the port to port data would have been available a multidimensional entropy<br />

method could have been applied to combine all sources.<br />

From this exercise we learned that combining sources is useful in any case to ensure consistency,<br />

but that the reliability is in the first place depending on availability of data for all elements of the<br />

chain.<br />

One of the main advantages of a top­down approach is that detail can be added to the database by<br />

inclusion of different sources on specific transhipment points or links without being held back by<br />

the non­availability of data on other transhipment points or links. A danger might be that in case the<br />

information for the construction of a transport chain is incomplete, like was the case in the example<br />

above, unreliable links can be included. In these cases it is important that these links can be<br />

identified in the database or that the main database can only be approached by trained experts and<br />

that sub­sets of the database constructed by aggregation of one or more variables will be available<br />

for a wider audience.<br />

4.5 The top­down approach<br />

The philosophy of the “transport chain principle” implicates that the transport flows are<br />

determined by the trade flows; we preserve the trade relation and follow the route of the<br />

transported goods. This means that, besides the origin and destination, the location of<br />

transhipment and the modes before and after transhipment are included. To do this we use a topdown<br />

approach, which means that we take the rough country­to­country trade information and<br />

refine this, step by step, using the various national data sources. This approach allows us to<br />

introduce the most disaggregated level permitted by the data sources available in each<br />

individual country, without being limited by the lack of data in other countries. By constructing<br />

the database using the top­down approach, it is possible to use a restricted number of sources with a<br />

maximum of regional detail. Additionally it is possible to give different parts of the database a<br />

different level of detail without introducing inconsistencies. This method is being materialised in an<br />

approach in which four phases can be distinguished.<br />

The phases in the top­down approach are the following:<br />

1. The building of a country­to­country matrix<br />

2. Including transhipment regions on the basis of transhipment statistics<br />

3. Regional division of country­to­country totals<br />

4. Incorporating domestic transport<br />

These phases will be described in the following paragraphs.<br />

4.5.1 Phase I The construction of a country­to­country matrix<br />

As a reference for the development of the database the EUROSTAT trade by mode (COMEXT) is<br />

used for the EU countries and national trade data sources for the other countries in the core area. In<br />

Document2<br />

27 May 2004<br />

21


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

cases where also no national sources can be found UN trade data will be used having the<br />

disadvantage that no mode information is included. In the trade data every trade relation is being<br />

registered from the export as well as the import side.<br />

These trade data sources, whose main function is to provide information relating to the value of<br />

trade, need to be modified before the information can be related to the volume of trade, even<br />

where volumes are recorded by the relevant customs authority. Corrections have to be applied<br />

due to incompatible measurement units (litres, kilograms, square metres), or due to data errors,<br />

often of significant magnitude. The first problem (incompatible units) can be solved by applying<br />

conversion ratios for the same commodity groups in instances where the ratio is known. The<br />

second problem (data outliers) can be solved by applying a smoothing technique to a time series<br />

of data. Such a technique has been applied with the MDS Transmodal trade forecasting model.<br />

The technique works by scanning the COMEXT data for a single trade flow (e.g. French exports<br />

of SITC 56 in tonnes to Italy) for quarterly time periods (of several years). The smoothing<br />

software samples four data points for each year and calculates the mean and the standard<br />

deviation for that year. It then compares each year's mean and standard deviation with all the<br />

others. Then if there are any years with unusual levels of variance, they are investigated by the<br />

software, and according to certain thresholds individual quarterly values may be marked as<br />

outliers and the software will replace them with interpolated values. It means that very erratic<br />

series will be left untouched, but erratic sequences within normally stable series will be<br />

corrected.<br />

Furthermore there will be adjustment required due to the resale of (bulk) commodities. For<br />

instance there is a significant amount of ores traded from the Netherlands to Germany according<br />

to COMEXT. These ores however do not originate from the Netherlands but are resold after<br />

being imported in the Netherlands, which implies that in fact this flow of ores between the<br />

Netherlands and Germany is just one link from the total transport chain. In the NEAC<br />

construction history many of these cases have been identified and can be traced back by<br />

comparing the NEAC forecast up to the year 2000 with the COMEXT 2000 data. The large<br />

outliers are being manually analysed. This procedure will be applied on top of the earlier<br />

mentioned procedure. For a future update of the <strong>ETIS</strong> reference database freight OD the OD<br />

matrix currently being developed can then be used for identification of outliers.<br />

22<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

The transport mode is registered at the border of the country in most trade sources and at the border<br />

of the EU for the extra EU trade in COMEXT; as a result it is possible to estimate the part that is<br />

transhipped onto another mode. When the trade statistics show that a flow leaves Spain by sea and<br />

enters Poland by road, it can be concluded that somewhere transhipment has taken place. In this<br />

phase all direct transport without transhipment and indirect transport with transhipment is<br />

registered.<br />

A difference in definition appears here since for the extra EU trade the mode is not anymore<br />

registered at the border of the countries but at the border of the EU. Specific solutions will be<br />

analysed amongst which the option of estimation of the country border mode by assignment in<br />

following steps.<br />

4.5.2 Phase II Including transhipment regions on the basis of transhipment<br />

statistics<br />

The identification of transhipment regions is taking place with the help of the available statistics<br />

originating from the national transhipment sources or ports and terminal. The inclusion of inland<br />

terminal information will be considered here as an experiment since no proof of concept is<br />

available.<br />

In this step two transhipment points will be included for intra <strong>ETIS</strong> reference database core area<br />

short sea flows. All collected port flows will be combined into one database. Here double countings<br />

have to be eliminated in the cases where for two ports transhipment data are available and where<br />

these ports have a connecting service; these flows are then registered at both ports.<br />

The port flows are then included in the trade database of step one taking account of all information<br />

included in the data (origin, destination, commodity, modes) and again removing all the double<br />

countings. This way the trade volumes on country to country level are the same as in step 1, but it is<br />

known whether transhipment takes place along the route, where this takes place and from what<br />

mode to what other mode.<br />

4.5.3 Phase III Regional division of country­to­country totals<br />

The first two steps resulted into information ranging from both the country of origin to the country<br />

of destination. The regional detail is added to the database by means of the different sources<br />

dividing the trade flows over the regions in a country. Countries for which this can be done we will<br />

call A countries. For the other countries to be called B countries, regionalisation can be performed<br />

by using domestic transport statistics (for instance New Cronos). These domestic transport statistics<br />

often only show us the total flows arriving or departing from a region. The remaining countries will<br />

be called C countries. For this last category estimation procedures will be applied which are<br />

developed in OD­ESTIM and which make use of socio­economic data (see section 6.6).<br />

Document2<br />

27 May 2004<br />

23


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

In annex C a description is included of a pilot exercise performed by MDS within the ‘Concerted<br />

Action Short Sea Shipping’ project where a method has been tested to estimate region­region flows<br />

from region­country data sources (A countries).<br />

4.5.4 Phase IV Incorporating domestic transport<br />

The previous phases are completely directed towards international transport. The fourth step<br />

concerns the inclusion of domestic transport in the database.<br />

A problem arises in the port regions when international and domestic transport is being connected.<br />

Transport that is transhipped in a port is registered twice; in the international trade statistics as well<br />

as in the domestic transport statistics. For example, a trade flow from Brussels via Antwerp by road<br />

to the USA by sea is registered in the domestic transport data as road from Brussels to Antwerp.<br />

A procedure will be applied to remove this type of double counting in the database. It reconstructs<br />

the complete chain by attaching domestic egress or access modes to the harbour for inland<br />

transport.<br />

It can be assumed that the transport between inland regions (without a sea port) is determined by<br />

domestic production and consumption patterns. Also it can be assumed that front or end transport<br />

connecting rail and inland waterways is taking place within the region of origin or destination (intra<br />

regional transport).<br />

The connection between the two sources of information is complicated by the fact that international<br />

trade flow statistics are based on documents that continuously register all trade passing the border,<br />

while domestic transport is based on surveys in a short period of time. When necessary, a<br />

correction has to be made for this fact in the direction of the trade statistics.<br />

The total top­down approach is schematised in figure 4.1.<br />

24<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Figure 4.1<br />

Top­down Approach<br />

4.6 Data gaps identified<br />

The top­down approach described above uses the available data sources to a maximum level.<br />

However, for some types of data no sources are available (on the complete EU level). Important<br />

data gaps identified are:<br />

· International region to region trade and transport flows<br />

· Transport mode on the European territory of intercontinental flows<br />

· Port transhipment data<br />

· Inland terminal transhipment data<br />

· Intermodal transport statistics<br />

· Container transport data<br />

· Commodity characteristics<br />

· Transport performance data (number of transport units, loading factors, number of<br />

loaded/empty trips)<br />

Since no complete data is available for these subjects (or data is available but not on the<br />

required level of detail or not for all required regions) estimation procedures have to be applied<br />

to fill these data gaps.<br />

Document2<br />

27 May 2004<br />

25


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

4.7 Estimating data gaps<br />

A broad collection of estimation models exists where socio­economic, network data and<br />

transport sector data serve as input. At this stage the OD­ESTIM project (4 th framework EC)<br />

will be used as reference. In this project different models are tested for estimation of transport<br />

data based on the four­stage model of generation, attraction, distribution and model­split.<br />

Models are developed for different levels of availability of data ranging from no transport data<br />

at all to estimation of only the modal­split. No models for estimation of inter­modal data or<br />

loading­units data are considered in OD­ESTIM. These models have proven to be successful as<br />

one of the tools for filling the remaining data gaps in the construction of the NEAC databases<br />

for Eastern Europe, the Russian Federation and Kazakhstan.<br />

In the top­down approach estimation procedures are applied to estimate the region to region<br />

flows. The results of the top­down approach can be used to make an assignment on the transport<br />

network. The assignment can be used to estimate missing data such as transport modes used and<br />

transhipment locations.<br />

Once the freight O/D matrix has been build from available data sources and data gaps have been<br />

filled, other data can be added.<br />

4.8 Adding information to the freight OD matrix<br />

After the freight OD matrix has been made available additional information that is not available<br />

in data sources can be added relatively easy by applying estimation procedures. For instance<br />

when the transport volume between an origin and a destination is known, transport performance<br />

information (expressed in tonne­kilometres) can be calculated by multiplying the volume by the<br />

distance between the regions. In this section an overview is given of how characteristics of the<br />

transport flows (cargo types, cargo characteristics, containerisation, number of TEUs, number of<br />

transport units) and transport performance information (tonne­km, vehicle­km/vessel­km, TEUkm)<br />

can be estimated. Along the project these estimation procedures will be further elaborated.<br />

4.8.1 Cargo types and characteristics<br />

Inclusion of cargo types and characteristics in the database has to be done by estimation since<br />

not all sources used in the top­down approach contain this type of information.<br />

Where the characteristics directly relate to the commodity itself , e.g. whether the goods are<br />

temperature controlled or ambient, or whether they are hazardous, it makes sense to relate these<br />

attributes to the commodity classification scheme, and to make the translation at an early stage<br />

of the processing so that this information is accessible in other stages.<br />

26<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

4.8.2 Containerisation<br />

Determining the containerisation is a real challenge due to the lack of data and the<br />

unreliability/incompleteness of the available material. An important source in which a container<br />

indicator is included is the COMEXT data. Another important source is the maritime statistics<br />

of EUROSTAT in which container information is available. The use of this source in the<br />

method will be analysed.<br />

The ability to observe and measure levels of containerisation within COMEXT is typically<br />

limited to extra­EU trade flows. However, intra­EU flows also need to be covered within <strong>ETIS</strong>.<br />

One possibility is to use a <strong>methodology</strong> developed by MDS Transmodal in which a broad<br />

segmentation is made by commodity according to the expected handling requirements of the<br />

products in question. The trade data is analysed at 5 digit SITC level (3500 different<br />

commodity descriptions) and they are assigned to two main cargo types:<br />

· Non­unitised;<br />

· Unitised.<br />

The non­unitised set, can typically be excluded from an analysis of containerisation. Instead, it<br />

can be further sub­divided into dry bulk, liquid bulk, semi bulk and general cargo categories.<br />

The ratios differ by corridor, so that for example, deep sea flows may have different<br />

characteristics compared to short sea.<br />

The unitised or “unitisable” set relates to commodities that are typically higher value goods that<br />

may be handled in containers, but which may also travel in road trailers. It is not possible to<br />

ascertain from a commodity description the actual mode of transport as this also depends upon<br />

the distance from origin to destination, the transport mode choices available, their costs, and in<br />

some cases upon the balance of containerised cargo in each direction. Therefore the assignment<br />

of the category ‘unitised’ a commodity is the limit of this process.<br />

Unitised flows can also converted into FEUs (forty foot equivalent container units), analogous<br />

to 40­44 tonne truck­loads. Again this is based upon the 5 digit SITC commodity, which<br />

improves the accuracy when there is a mix of commodity types within a single 2 digit<br />

definition. Like this, the volume traded can be expressed as unit loads without the actual mode<br />

being assigned.<br />

4.8.3 Number of transport units (vehicles/vessels) by type<br />

The determination of the number of transport units (number of vehicles for road, number of<br />

trains for rail and number of vessels for inland waterways and maritime transport) depends very<br />

much on the quality of the mode split data, and upon surveys providing profiles of vehicle type.<br />

The <strong>methodology</strong> depends upon knowing the mode of transport, the length of haul and the<br />

commodity.<br />

Document2<br />

27 May 2004<br />

27


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

For the rail and sea modes, the commodity classification can establish whether the goods are<br />

bulk or unitised. These headings can then be used to determine load factors in tonnes and FEUs<br />

respectively which will allow conversion to train numbers or ship numbers. In the case of the<br />

sea mode, there is a wide range of ship sizes, and in practice it may be easier to measure vessel<br />

flows (e.g. between ports) with reference to supply side data. For rail, provided that the mode<br />

share can be ascertained, conversions of tonnes or FEUs to train numbers could be usefully<br />

made.<br />

The road mode needs to be main focus of this procedure, as there is a wide variety of vehicle<br />

size possibilities, and there is no supply­side data (e.g. timetables) for road haulage services.<br />

The estimation needs to be informed by vehicle stock and survey data.<br />

There are differences in data availability by country, for instance in some national statistics up<br />

to 20 vehicle types are being distinguished where in other countries gaps exist. This data<br />

availability will be studied in the next phases in order to see whether improvements to the<br />

currently most used estimations are possible.<br />

In most cases the translation from tonnes to vehicles or vessels is done by one figure indicating<br />

the average number of tonnes per vehicle or vessel. Sometimes an estimate is made in this way<br />

of more than one type for instance van and truck.<br />

4.8.4 Number of TEUs<br />

Determining the number of TEUs will be done by estimation. Different determining factors are<br />

present in this respect amongst which the volume of the commodities and the weight which has<br />

to be compared with the maximum volume and weight of a TEU and of course the efficiency of<br />

transport. Tables are existing of the volume per weight of the different commodities.<br />

Different container types have different maximum weights like for instance:<br />

Max weight:<br />

· 20 foot container (1 TEU) with a maximum weight of 20 tonne and<br />

· 40 foot container (2 TEU) with a maximum weight of 23 tonne<br />

Often the TEU is determined by applying the average of 9 or 10 tonne per TEU. It will be seen<br />

whether this method can be refined with available data.<br />

MDS Transmodal have developed conversion tables, similar to those used to estimate other<br />

cargo characteristics to relate tonnes of a product at the 5 Digit SITC level to the expected<br />

number of 20’ or 40’ units, using estimated stowage factors (tonnes per 20’ and tonnes per 40’).<br />

28<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

4.8.5 Transport performance information (tonne­km, vehicle­ km/vessel­km, TEU­<br />

km)<br />

The estimation of transport performance information such as tonne­km, vehicle­km/vessel­km<br />

and TEU­km will be done in co­operation with the European transport network development<br />

since assignment on the network is required to determine the distance.<br />

Once the distance by mode has been determined between regions or by link in the transport<br />

network, transport performance information can be calculated by multiplying the transport<br />

volume information (in tonnes, in transport units and in TEUs) with the transport distance. This<br />

can be done for specific parts of the nework (for instance for the national territory of a country)<br />

or on a more aggregate level for total transport generated or attracted by a region or a country.<br />

Document2<br />

27 May 2004<br />

29


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

5 TRANSPORT CHAIN FREIGHT O/D MATRIX – DATA NEEDS, DATA<br />

COLLECTION AND REMAINING DATA GAPS<br />

5.1 Introduction<br />

In this chapter first the data needs for building a complete transport chain freight O/D matrix are<br />

described for an ideal situation (where all required data is available). Next, an overview of the<br />

data collection is given. Finally, data that is needed but that cannot be collected within the<br />

project is described, this is called the remaining data gaps.<br />

5.2 Data needs<br />

The exact structure of the data construction methods that are applied is dependent on the input<br />

to be found. In this section an overview is given of data needed in the database construction<br />

process in an ideal situation where all required data is available and all available data can be<br />

processed within certain limitations.<br />

In general different types of data need to be collected to be able to construct the freight transport<br />

database:<br />

· Trade data (movement of goods)<br />

· Transport data (movement of transport units)<br />

· Maritime transhipment data<br />

· Inland transhipment data<br />

· Container data<br />

· Vehicle/vessel movement data<br />

From the other datasets developed within <strong>ETIS</strong> the following is required:<br />

· Socio­economic data<br />

· Network data<br />

· Transport sector data<br />

Each of the data types will be discussed.<br />

5.2.1 Trade data<br />

Since trade is the origin of all transport movements this data is the basis of the top down<br />

approach to be applied. A consistent source for all countries is needed. This source preferably<br />

should contain information about:<br />

· Origin (region, country)<br />

Document2<br />

27 May 2004<br />

31


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

· Destination (region, country)<br />

· Commodities,<br />

· Modes (at the national border, EU border, at origin, at destination?),<br />

· Containerisation,<br />

· Weight<br />

· Value<br />

· Route information:<br />

• border crossing<br />

• transhipment sea ports<br />

• transhipment other terminal(s)<br />

These variables should be collected in as much detail as possible. The ideal situation would be<br />

that all information needed to describe transport would be available from the trade registrations.<br />

Unfortunately this is not expected to be found. It should be attempted to come as close as<br />

possible to the ideal situation. Elements like region to region information or transhipment<br />

locations are unfortunately only available in very limited and/or fragmented cases.<br />

The following priority list can be followed for origin destination data:<br />

1. Region to region<br />

2. Region to country<br />

3. Country to country<br />

In the top­down approach of the NEAC model country to country data has been used from the<br />

COMEXT database since this is consistent for all EU countries. Region to country trade data<br />

from national sources is then used to include the regional detail.<br />

The following sources are considered:<br />

· EUROSTAT COMEXT<br />

· National trade data for all core countries<br />

· UN trade data<br />

5.2.2 Transport data<br />

Transport data can be used for different purposes in the top­down approach and should be<br />

collected for all modes. All other variables mentioned under the trade flows should also be<br />

considered here as much as possible. The first distinction to be made is:<br />

· International transport<br />

· Domestic transport<br />

· Container or Loading units flows<br />

32<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

If trade data is available but excluding regional detail then regional international transport<br />

information can be used to estimate the regional detail. Here the following priority list can be<br />

used:<br />

1. Region to region<br />

2. Region to country<br />

3. Country to country<br />

International transport data without regional detail are not of use for a top­down database<br />

construction approach. Only in case of a database estimation procedure aggregated international<br />

transport data can be very useful. This has to be considered in case no other data can be found.<br />

Domestic transport data has the following priority list:<br />

1. Region to region<br />

2. Region incoming and region outgoing<br />

3. Country total<br />

The following transport data sources are considered:<br />

· EUROSTAT New Cronos<br />

· National transport data sources<br />

5.2.3 Maritime transhipment data<br />

This type of transhipment information is best known for the major ports or port countries in<br />

Europe. For these ports very reliable and complete information used to be available. After 1992<br />

the availability and quality of this type of data has decreased or the data are not available<br />

anymore (information about ports and modes have become optional for intra­EU flows in the<br />

Intrastat system). For other ports in the EU availability of data has always been limited. It has to<br />

be attempted to find alternative useful sources to fill this data gap.<br />

Currently Eurostat collects port­to­port data from the Member States according to the Maritime<br />

Directive (see ”Council directive 95/64/EC of 8 December 1995 on statistical returns in respect<br />

of carriage of goods and passengers by sea”). By the end of the testing method, a request has<br />

been made to EUROSTAT by DGTREN to make the data available for the construction of the<br />

<strong>ETIS</strong> reference database. The data have not been received yet at moment of writing. When the<br />

data becomes available, it will be used in the methods. This data contains no information about<br />

transhipment, but it does contain information about goods loaded and unloaded in ports in<br />

relation with partner ports.<br />

The following maritime transhipment data sources are considered:<br />

· National transhipment data sources<br />

· Ports<br />

Document2<br />

27 May 2004<br />

33


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

5.2.4 Container data<br />

Not only data is needed about vehicle movements (transport data) or the movements of the<br />

commodities in them (trade data) but also information on the containers is of interest.<br />

Containers can have different origins and destinations than the commodities in them and the<br />

vehicles transporting them. These data therefore are of at least comparable value for the <strong>ETIS</strong><br />

reference database. An example of such data sources is the Piers database that describes the<br />

movements of containers between the USA and Europe where multiple transhipment is<br />

included. It will be attempted to make a distinction between filled and empty containers.<br />

The following container data sources are considered:<br />

· EUROSTAT COMEXT (container indicator, only for extra­EU trade)<br />

· UIRR<br />

· ICF<br />

· Piers database<br />

· Ports<br />

· Inland terminals<br />

5.2.5 Vehicle/vessel movement data<br />

This type of information is available in several sources.<br />

The following vehicle/vessel movement data sources are considered:<br />

· EUROSTAT New Cronos<br />

· National transport statistics<br />

Special attention has to be paid to movements of empty vehicle/vessels<br />

5.2.6 Relation with other datasets in the <strong>ETIS</strong> reference database<br />

State of the art<br />

All state of the art information collected will be incorporated in the freight demand dataset.<br />

Socio­economic dataset<br />

Where data gaps remain estimation procedure must be applied. These estimation procedures for<br />

a large part rely on socio­economic input data. The required socio­economic data can be<br />

different for each model. The following data can be seen as frequently used.<br />

· Population by sex and age classes (0­4, 5­9, ..., 65­70, > 70)<br />

· Gross value added by three sectors:<br />

• Agriculture, fishery, forestry<br />

• Industry<br />

34<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

• Services<br />

· Total GDP and GDP/Capita<br />

· GDP by three sectors:<br />

• Agriculture, fishery, forestry<br />

• Industry<br />

• Services<br />

· GDP by other sectors:<br />

• Basic metals<br />

• Metal products<br />

• Chemical production; petrol products<br />

• Chemical production; other products<br />

• Mining & quarrying production<br />

• Construction<br />

• Electricity/Gas/Water production<br />

• Private final consumption<br />

• Food Consumption<br />

• Residential Construction<br />

· Fuel prices for unleaded fuel and diesel<br />

· Vectorized cards for NUTS­3 borders<br />

European transport network dataset<br />

Network data are mainly required for the freight demand dataset to determine the distance<br />

between regions that can be used to calculate for instance the transport performance in tonnekilometres.<br />

The freight O/D matrix will be used as input for the European transport network<br />

dataset to make assignments on the transport network in order to determine the intensities on<br />

each link of the network.<br />

Freight transport service and cost dataset<br />

Data about the transport sectors are needed for all countries/areas to be included in the system.<br />

Here variables like costs per hour and by kilometre are important but also restrictions on the<br />

commodities or weight to be transported.<br />

External effects dataset<br />

The freight O/D matrix will be used as input for the external effects dataset to calculate external<br />

effects between O/D relations or on network level (via assignments).<br />

Synthesis and dissemination<br />

The freight demand dataset will be used for the synthesis and dissemination.<br />

Document2<br />

27 May 2004<br />

35


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

5.3 Data collection<br />

Based on the description of the data needs for the transport chain freight O/D matrix in the<br />

previous sections many (potential) data providers have been contacted. For each of these data<br />

providers it has been investigated what data is available and under what conditions the data can<br />

be made available for use in the <strong>ETIS</strong> reference database project. An overview of the current<br />

status of the data collection from different data sources is given in <strong>Annex</strong> F. Tables 5.1, 5.2 and<br />

5.3 give an overview of available data by source and/or country for trade data, international<br />

transport data and domestic transport data respectively.<br />

If a field in the tables is indicated by ‘o’ this means that the availability of the data is known. If<br />

a field is empty this can mean either that the data is not available or that the availability of the<br />

data is unknown. It is noticed that this table presents the status of the data collection up to the<br />

test of the <strong>methodology</strong>; it gives a rough impression of the data availability. After the test of the<br />

<strong>methodology</strong>, the data collection will be continued where necessary in order to collect as much<br />

relevant information as possible within the limitations of the project.<br />

36<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Table 5.1:<br />

Overview of available trade data<br />

Trade data<br />

Type of data<br />

Sources country to region to transport commodity container transhipment volume trade<br />

country country mode type indicator location(s) in tonnes value<br />

Supranational<br />

sources<br />

UN Comtrade o o o (partly) o<br />

UN­ECE<br />

Common database o o o<br />

Comext o o o o o o<br />

National sources<br />

France o o o o o<br />

Belgium o o o o o<br />

Luxembourg<br />

Netherlands<br />

Germany o o o o o<br />

Italy<br />

United Kingdom o o* o o* o o<br />

Ireland o o o o<br />

Denmark<br />

Greece<br />

Portugal<br />

Spain o o o o o o o<br />

Norway o o o o<br />

Sweden<br />

Finland<br />

Switzerland o o o o<br />

Austria<br />

Poland o o o o o o<br />

Czech Rep. o o o o<br />

Slovak Rep. o o o o<br />

Hungary o o o o<br />

Estonia o o o o<br />

Latvia o o o o<br />

Lithuania o o o o o<br />

Slovenia o o o<br />

Malta<br />

Cyprus<br />

* Mode and transhipment information is known for UK trade with countries outside the EU.<br />

Table 5.2<br />

Overview of available international transport data<br />

Document2<br />

27 May 2004<br />

37


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

International<br />

transport<br />

data<br />

Type of data<br />

transhipment/<br />

Sources country to region to region to by transport commodity container border volume performance<br />

Supranational<br />

sources<br />

country country region region mode Type indicator location(s) in tonnes value in tonne­km<br />

NewCronos o o o o O o o o<br />

CAFT o o o o o o o<br />

UIRR o o o o<br />

ICF o o o o (TEU)<br />

National sources<br />

France o o o o o o o<br />

Belgium o o o o o o o o<br />

Luxembourg<br />

Netherlands o o o o o o<br />

Germany o o o o o o<br />

Italy o o o o o<br />

UK o o o o<br />

Ireland o o o o<br />

Denmark o o o o<br />

Greece<br />

Portugal<br />

Spain<br />

Norway<br />

Sweden o o o o o<br />

Finland o o o o<br />

Switzerland<br />

Austria o o o o<br />

Poland<br />

Czech Rep. o o o o<br />

Slovak Rep.<br />

Hungary<br />

Estonia<br />

Latvia<br />

Lithuania<br />

Slovenia<br />

Malta<br />

Cyprus<br />

38<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Table 5.3:<br />

Domestic<br />

transport data<br />

Overview of available domestic transport data<br />

Type of data<br />

Sources region to transport commodity container volume performance<br />

region<br />

by<br />

region mode type indicator in tonnes value in tonne­km<br />

Supranational<br />

sources<br />

NewCronos o o o o<br />

National sources<br />

France o o o o o o<br />

Belgium o o o o<br />

Luxembourg<br />

Netherlands o o o o<br />

Germany o o o o<br />

Italy o o o o<br />

United Kingdom o o o o<br />

Ireland o o o o<br />

Denmark o o o o<br />

Greece<br />

Portugal o o o o<br />

Spain o o o o<br />

Norway o o o o<br />

Sweden o o o o<br />

Finland o o o o<br />

Switzerland o o o o<br />

Austria o o o o<br />

Poland o o o o<br />

Czech Rep. o o o o<br />

Slovak Rep. o o o o<br />

Hungary o o o o<br />

Estonia o o o<br />

Latvia o o o<br />

Lithuania o o o<br />

Slovenia<br />

Malta<br />

Cyprus<br />

From these overviews of datasources and available data it becomes clear that besides data from<br />

Eurostat, data from many other data sources is required for the construction of the transport<br />

chain freight O/D matrix.<br />

Since the data collection is a very time consuming process, there are still negotiations going on<br />

about the conditions of the data delivery. The data collection process will therefore continue<br />

after the test phase has finished. For the test phase enough data had been collected in order to be<br />

able to test the different methods.<br />

Document2<br />

27 May 2004<br />

39


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

5.4 Remaining data gaps<br />

From the data collection process it becomes clear that not all data is available that is needed in<br />

an ideal situation. In this section the gaps between the needed data and the available data is<br />

described together with the consequences that follow from these gaps.<br />

5.4.1 Data available at Eurostat not available for the <strong>ETIS</strong> project<br />

In order to build for instance a region­to­region freight O/D matrix data is required on a regional<br />

level. Regional data is available in New Cronos from Eurostat, but this data has limited detail.<br />

Since the Member States deliver region to region transport data (for road transport) to Eurostat,<br />

a request has been made to Eurostat to ask for this data. Due to restrictions in the statistical law,<br />

Eurostat is not allowed to provide the data for use in the <strong>ETIS</strong> project. In response it has been<br />

attempted to collect these data from the Member States. Some Member States have delivered<br />

the data where others reply that the data has already been delivered to Eurostat and therefore<br />

does not have to be delivered again tot the European Commission. This illustrates that in some<br />

cases more structural organisational changes are needed for the data collection by the European<br />

Commission. During the second open conference of <strong>ETIS</strong> Eurostat has noted this problem and<br />

will discuss with the Member states what would be possible solutions.<br />

Since it is likely that the international road transport on region to region level (NUTS3) will not<br />

be made available in time to be used for the for the <strong>ETIS</strong> pilot by Eurostat, the international<br />

region to region transport flows on NUTS2 level will be estimated.<br />

5.4.2 Registration of mode of transport for intra­EU trade becomes optional in Eurostat<br />

trade statistics<br />

For the year 2000, the variable mode of transport is still included in the Comext trade by mode<br />

data (from Eurostat). From 1 January 2001, this variable will be still included in the data, but<br />

the variable will be less reliable. The reason for this is that the collection of mode of transport in<br />

intra­EU trade has become optional for Member States and applies only to those providers<br />

above a certain threshold. This has consequences for a future update of the <strong>ETIS</strong> reference<br />

database.<br />

5.4.3 Transhipment data of seaports and inland terminals<br />

In the data collection process it became clear that transhipment data is very dispersed. For<br />

transhipment in seaports data is available, for transhipment at inland terminals data is not<br />

available or only on a very aggregate level. Since methods to fill this data gap are complex and<br />

very time consuming to develop, it has been decided to leave transhipment at inland terminals<br />

out of the freight O/D matrix.<br />

40<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

5.4.4 Data availability in the accession countries<br />

Since these countries have become member of the EU from May 2004, the collection of data in<br />

these countries will change since they have to apply the Eurostat directives. In <strong>ETIS</strong> a<br />

repeatable method has to be developed, therefore it makes no sense to spend a lot of time and<br />

budget in using the available data for the accession countries and apply and develop methods<br />

and models to fill the data gaps while it is known that the data situation in these countries will<br />

change significantly in the near future. The data collection process in the accession countries<br />

was very time consuming, it is a hard job to get in contact with the right persons at the different<br />

institutes, data is not available on the required level of detail and limited response has been<br />

received. Because of these reasons it has been decided to make use of a combination of results<br />

from the ‘Traffic Forecast of the Ten Pan­European Transport Corridors of Helsinki’ project,<br />

the INTERMODA project, NEAC, the TEN­STAC project, UN data and data already available<br />

for the accession countries to estimate the transport flows within and in relation with the<br />

accession countries. Once the Eurostat directives are applied in the accession countries the same<br />

<strong>methodology</strong> as for the current EU countries can be applied in order to build the freight O/D<br />

matrix.<br />

Document2<br />

27 May 2004<br />

41


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

6 DESCRIPTION OF THE <strong>WP</strong> 3 TESTING PHASE AND ITS RESULTS<br />

6.1 General description of the testing phase<br />

In the testing phase the proposed methods have been tested in order to find out whether they are<br />

feasible and to determine whether and how the methods should be modified.<br />

In <strong>WP</strong> 3 in total 6 different methods are tested in this phase. In the first method a country­tocountry<br />

matrix in constructed. The second method determines where transhipment takes place<br />

along the transport chain. To store this information in the database, a record structure with two<br />

transhipment locations is introduced. Then in the third method the region­to­region flows are<br />

estimated and included in the database. In the fourth step (pure) domestic transport is included.<br />

After these four steps a complete region to region freight O/D transport chain matrix results. In<br />

method 5 cargo characteristics and transport unit information is added to the database. Finally in<br />

method 6 information about the transport performance (tonne­kilometres, number of vehicles,<br />

vehicle­kilometres, number of TEUs, TEU­kilometres) is included.<br />

6.2 Tested methodologies<br />

In this paragraph the applied methods are described.<br />

Method 1: Development of country­to­country matrix<br />

Data source: COMEXT database from Eurostat<br />

The COMEXT database, published by EUROSTAT contains detailed trade data harmonised for<br />

all EU member states. The database contains the variables described in table 6.1.<br />

Table 6.1 Variables included in the COMEXT database<br />

Variable<br />

Description<br />

DECLARANT Country Reporting the Trade : EU Member States<br />

PARTNER<br />

Trading Partner: All countries<br />

PRODUCT CODE Commodity Code, using the 8 Digit Combined Nomenclature (CN8)<br />

FLOW<br />

Import or Export<br />

STAT REGIME Statistical Regime: General or Special Trade<br />

PERIOD<br />

Time Period e.g. 12 Month Period<br />

VALUE<br />

‘000s of EURO<br />

QUANTITY Tonnes<br />

SUP_QUANTITY Additional Quantity, e.g. Square Metres<br />

Identification of outliers<br />

Document2<br />

27 May 2004<br />

43


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

The COMEXT database for the Year 2000 has been used for the testing phase of <strong>ETIS</strong>. The<br />

first test has considered whether it is necessary to apply additional error checking routines to<br />

avoid the inclusion of data errors in the subsequent O/D matrices. This has been achieved by<br />

comparing ‘smoothed’ data to the raw data to measure the impact of erratic values in the<br />

database.<br />

A technique for identifying outliers (errors or “erratics”) has been developed within the MDS<br />

Transmodal trade forecasting model, taking into account a long time series of trade data.<br />

During this process, the COMEXT data is converted into time series vectors for individual trade<br />

flows (e.g. French exports of SITC 56 in tonnes to Italy) for quarterly time periods covering<br />

approximately fifteen years. The smoothing software samples four data points for each year and<br />

calculates the mean and the standard deviation for that year. It then compares each year's mean<br />

and standard deviation with all the others. Then if there are any years with unusual levels of<br />

variance, they are investigated by the software, and according to certain thresholds individual<br />

quarterly values may be marked as outliers and the software will replace them with interpolated<br />

values. It means that normally erratic series will be left untouched, but erratic points or<br />

sequences within normally stable series will be changed. Every year, new data is collected, and<br />

the process is repeated, so it is possible that what is regarded as an outlier may change over time<br />

as the software learns more about the time series.<br />

The use of the smoothing algorithm can be illustrated, by comparing the smoothed data to the<br />

original data.<br />

In 2000, imports into EU countries amounted to 2.501 billion tonnes according to COMEXT.<br />

After smoothing the estimate was 2.391 billion tonnes, a change of only 4%. The largest<br />

absolute error in a single 2 digit SITC category is 13 million tonnes, for SITC 33, petroleum.<br />

However this is only a 2% difference within that category. The largest percentage error is for<br />

SITC 83, travel goods, with a 63% percent difference. However this only amounts to an<br />

absolute difference of 1.133 million tonnes. Most of the difference can be traced to a figure of<br />

1.079 million tonnes for travel goods between the UK and Germany.<br />

Looking at the same trade flow using German export data a total of 0.001 million tonnes can be<br />

found, a level that agrees more readily with the smoothed data. This difference (1000 times) is<br />

untypical however. Absolute differences are typically about 1 million tonnes per 2 digit SITC<br />

category, and relative differences are typically about 6%. It should also be noted that<br />

differences between smoothed and un­smoothed series do not necessarily imply that the unsmoothed<br />

series contains errors, only values that are unlikely to be repeated.<br />

At these levels, and given the scope of <strong>ETIS</strong>, the potential impacts of measurement errors are<br />

not alarming, particularly when the annual version of COMEXT is used. However, the example<br />

related above does suggest that a high level comparison between the trade data used within<br />

<strong>ETIS</strong> and the data based upon a smoothed quarterly time series will reveal a small number of<br />

44<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

important differences that can be corrected manually. The ability to compare counter­flows in<br />

intra EU data is also useful in this context.<br />

Conversion of commodity code<br />

The COMEXT database is published using the international 8 digit combined nomenclature<br />

(CN8) system. This has the advantage that the flows can be readily and unambiguously<br />

converted into other (more aggregated) systems such as the Standard International Trade<br />

Classification (SITC) and Standard Goods Classification for Transport Statistics/Revised<br />

(NST/R). Conversion tables are published on EUROSTAT’s classification server. See<br />

http://europa.eu.int/comm/eurostat/ramon/.<br />

Selection between import and export registration<br />

The Comext data contains a for extra­EU trade one registration, the registration of import or<br />

export of the EU country. For intra­EU trade the Comext data contains two registrations for the<br />

same flow, once registered as export of the origin country and once registered as import of the<br />

destination country. In an ideal case, the transport volumes in both registrations are the same.<br />

Unfortunately, in many cases the registrations are not the same. The example described in the<br />

table below illustrates this.<br />

Table 6.2<br />

Origin country<br />

Example of differences in registration for the same flow<br />

Destination<br />

country<br />

Commodity Registration Transport volume in<br />

tonnes<br />

France The Netherlands Cereals Import registration the Netherlands 3611453<br />

France The Netherlands Cereals Export registration France 4252118<br />

In this example the export registration is about 640.000 tonnes higher than the import<br />

registration. In the database only one value will be included for the trade flow of cereals from<br />

France to the Netherlands, thus these two registrations have to be converted in a single<br />

registration. The trade flows are registered according to the INTRASTAT system. The rules that<br />

have to be followed in the INTRASTAT system give no indication that the import or the export<br />

registration is more reliable (in the past the import registration was considered to be more<br />

reliable). Since it cannot be decided what registration is more reliable, it is decided that both<br />

registrations are even reliable and therefore the average value of the import and the export<br />

registration is taken as the transport volume on this relation (in the example above the transport<br />

volume becomes 3931786 tonnes). In order to keep information about the difference between<br />

the import and the export registration, two variables are added to the data. One variable<br />

indicates whether the difference between the import and the export registration is more than<br />

500.000 tonnes (in the example given above this is the case), another variable indicates the<br />

relative difference between the average value and the import and export registrations (in the<br />

example above this percentage is 8%, indicating that the transport volume could actually be 8%<br />

lower or 8% higher). These indicators for the difference between import and export registration<br />

will be used in the second method for the identification of confusion between trade and<br />

transport.<br />

Document2<br />

27 May 2004<br />

45


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Recoding of variables and modification of record structure<br />

After the selection between import and export registration has been made, the origin country<br />

and the destination country variables are recoded into the <strong>ETIS</strong> reference database country<br />

classification. Besides, the record structure is modified:<br />

· Origin country (<strong>ETIS</strong> country classification)<br />

· Destination country (<strong>ETIS</strong> country classification)<br />

· NSTR 2 digit commodity classification<br />

· Value of the goods (in Euro)<br />

· Volume of the goods (in tonnes)<br />

· Indicator absolute difference between import and export registration<br />

· Indicator relative difference between import and export registration<br />

· Indicator for intra­EU or extra­EU trade<br />

Adding information about the mode of transport<br />

The used Comext database has the advantage that it includes information about the traded goods<br />

on a very detailed level which makes it possible to add information about the cargo<br />

characteristics as described above. A disadvantage of this database is that it includes no<br />

information about the mode of transport. Besides the Comext database used up to now, there is<br />

another Comext database called ‘Comext trade by mode’ that includes less detailed commodity<br />

information (on NSTR 3 digit level), but it includes information about the mode of transport.<br />

The definition of the transport mode included in the Comext trade by mode database is as<br />

follows:<br />

· For intra­EU trade the mode is defined as the active means of transport with which goods<br />

are presumed to enter/leave the statistical territory of the Member State for import/export;<br />

· For extra­EU trade the mode is defined as the active means of transport with which goods<br />

are presumed to enter/leave the statistical territory of the European Community for<br />

import/export.<br />

For extra­EU trade there is one registration with one modal­split that gives the mode of<br />

transport for the goods entering or leaving the European Community. For intra­EU trade the<br />

situation is more complex, there are two registrations with two modal­splits that give the mode<br />

of transport on the territory of the exporting Member State and the mode of transport on the<br />

territory of the importing Member State. In the same way as for the different registrations of the<br />

total trade volumes, the different registrations of the modal­split have to be converted into one<br />

registration of the modal split.<br />

The following method is applied:<br />

46<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

1. In the Comext trade by mode database the distinguished modes are recoded into the<br />

transport modes used in <strong>ETIS</strong> for freight transport (road, rail, inland waterways,<br />

maritime, rest);<br />

2. From the Comext trade by mode database for each country to country by commodity<br />

classification by cargo characteristics combination the modal­split is determined once in<br />

the export registration and once in the import registration, the trade volumes are scaled to<br />

the volumes as selected between the import and the export registration;<br />

3. A rule is applied to determine the origin mode – destination mode combinations:<br />

a. For each transport mode, the minimum volume is determined in the import and the<br />

export registration, for this trade volume the origin mode is equal to the destination<br />

mode. For instance if the export registration gives 1000 tonnes trade by road and<br />

the import registration gives 1500 tonnes trade by road, then the trade volume of<br />

the combination origin mode road – destination mode road becomes 1000 tonnes.<br />

b. For the trade volumes left after step a the origin mode and the destination mode<br />

cannot be the same. An order is determined in the modes of transport: maritime<br />

transport has the highest order, followed by inland waterways transport, rail<br />

transport, road transport and finally the rest category has the lowest order. The<br />

trade volume of the origin mode with the highest order is matched first with the<br />

destination mode with the highest order, then with the destination mode with the<br />

following order, etc. until the total trade volume of the origin mode is matched with<br />

the destination mode. Then for the origin mode with the next order the same<br />

procedure is applied etc. until all origin modes have been matched with destination<br />

modes.<br />

The example in the table below illustrates the effect of this procedure.<br />

Table 6.3<br />

Modes in export registration<br />

Example of procedure to match origin mode and destination mode<br />

Modes in import registration<br />

Road 2000 tonnes Road 1700 tonnes<br />

Rail 500 tonnes Rail 600 tonnes<br />

Inland waterways 500 tonnes Inland waterways 300 tonnes<br />

Maritime 1500 tonnes Maritime 1900 tonnes<br />

Rest 200 tonnes Rest 200 tonnes<br />

The method results in the following origin mode – destination mode combinations<br />

Origin mode Destination mode Transport volume Remark<br />

Road Road 1700 tonnes Direct transport<br />

Rail Rail 500 tonnes Direct transport<br />

Inland waterways Inland waterways 300 tonnes Direct transport<br />

Maritime Maritime 1500 tonnes Direct transport<br />

Rest Rest 200 tonnes Direct transport<br />

Inland waterways Maritime 200 tonnes Indication transhipment<br />

Road Maritime 200 tonnes Indication transhipment<br />

Road Rail 100 tonnes Indication transhipment<br />

In case the origin mode is different than the destination mode, this is interpretated as an<br />

indication that transhipment might have taken place along the route. In the next methods this<br />

information is used to include transhipment information to the database.<br />

Document2<br />

27 May 2004<br />

47


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

The last step of this method is the modification of the record structure. The record structure is<br />

modified in such a way that a transport chain with two transhipment locations fits in the<br />

structure:<br />

· Origin country (<strong>ETIS</strong> country classification)<br />

· First transhipment location (still empty in this phase)<br />

· Second transhipment location (still empty in this phase)<br />

· Destination country (<strong>ETIS</strong> country classification)<br />

· Mode at origin<br />

· Mode between transhipment locations<br />

· Mode at destination<br />

· NSTR 2 digit commodity classification<br />

· Value of the goods (in Euro)<br />

· Volume of the goods (in tonnes)<br />

· Indicator absolute difference between import and export registration<br />

· Indicator relative difference between import and export registration<br />

· Indicator for intra­EU or extra­EU trade<br />

In the second method the transhipment locations will be filled.<br />

Method 2: Estimation of transhipment<br />

In the first method the database structure has been set up in such a way that information for<br />

maximum two transhipment locations along the transport chain can be stored. In the first<br />

method, the transhipment locations are still empty. In the second method the transhipment<br />

locations will be estimated. Since it is not feasible to estimate transhipment at inland terminals<br />

in the <strong>ETIS</strong> project, only transhipment at seaports is included.<br />

For the estimation of the transhipment locations two types of flows are distinguished:<br />

· Transit flows; goods transhipped in a port with both origin and destination outside the<br />

country where the port is located (either mode before or mode after transhipment is sea<br />

transport);<br />

· Import and export by sea; Import: transport by sea from a country to a port (located in a<br />

different country) where the goods are transhipped after which they are transported by a<br />

land mode to the destination region (which is located in the same country as the port);<br />

Export: transport from a region by a land mode to a port (which is located in the same<br />

region as the origin region) where the goods are transhipped after which they are<br />

transported by sea to a country (which is different than the origin country).<br />

In this method first the transit flows are estimated and added to the O/D matrix, secondly the<br />

import and export by sea is estimated and added to the O/D matrix.<br />

48<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Transit flows<br />

Data about transit in seaports is available for ports in Belgium, the Netherlands and Germany,<br />

for ports in other countries no data is available for transit. There are some problems with this<br />

data (intra­EU transit is not included in the Belgium data, transit through the Netherlands is not<br />

collected anymore for the year 2000), but these gaps can be filled by applying estimation<br />

procedures in order to construct a complete and consistent transit database. This data has the<br />

following format:<br />

· Origin country (for instance Southern America)<br />

· Transhipment port (for instance Hamburg in Germany)<br />

· Destination country (for instance Austria)<br />

· Mode before transhipment (for instance sea transport)<br />

· Mode after transhipment (for instance road transport)<br />

· Commodity type (according to the NSTR2 or NSTR1 classification)<br />

· Volume of the goods (weight in tonnes)<br />

These transit flows have to included in the O/D matrix in such a way that double countings are<br />

avoided. This is done in three steps:<br />

1. Match transit flows with international trade flows;<br />

2. Make corrections for confusion between trade and transport flows;<br />

3. Add remaining transit flows.<br />

Match transit flows with international trade flows<br />

In the first step the transit flows through ports in Belgium, the Netherlands and Germany are<br />

matched with the international trade flows in the modified Comext database resulting from<br />

method 1.<br />

An example how these flows are matched is given in table 6.4.<br />

Table 6.4<br />

Origin<br />

region<br />

Mode<br />

origin<br />

Matching transit flows with international trade flows<br />

Transh.<br />

region<br />

Mode Transh. region Mode<br />

destination<br />

Destination<br />

region<br />

Volume<br />

(tonnes)<br />

Flow type<br />

USA sea Germany 25000 International<br />

USA sea Rotterdam<br />

(NL)<br />

rail Germany 15000 Transit<br />

In the Comext database it is known that 25.000 tonnes of a specific commodity is transported<br />

from the USA to Germany with mode sea (for extra­EU trade the mode of transport at the<br />

border of the Community). From the transit data it is known that 15.000 tonnes of a specific<br />

commodity is transported by sea from the USA to Rotterdam where the goods are transhipped<br />

onto mode rail and then transported to Germany. In this example all transit can be matched with<br />

the international trade, in the O/D matrix the international trade will be reduced by the transit<br />

flows (25.000 – 15.000 = 10.000 tonnes) and a record will be added to the O/D matrix for the<br />

transit through Rotterdam with a volume of 15.000 tonnes.<br />

Document2<br />

27 May 2004<br />

49


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

In this example the volume of the transit is lower than the international trade flow (between<br />

countries by commodity type). It also occurs that the transit flow can be matched, but that the<br />

transit flow is higher than the international trade flow. In such a case the total international trade<br />

flow is replaced by the transit flow (in order to include transhipment data in the O/D matrix).<br />

The remainder of the transit flow cannot be matched (country by country by commodity type),<br />

with these flows is dealt in the next two steps. Another possibility is that the transit flow cannot<br />

be matched at all (country by country by commodity type). These flows are also being dealt<br />

with in the next steps.<br />

From the total amount of about 190 million tonnes transit through Belgium, the Netherlands and<br />

Germany about 115 million tonnes can be matched with the international trade flows based on<br />

the country to country by commodity type relation.<br />

Make corrections for confusion between trade and transport flows<br />

In method 1 it has been noticed that for some flows large differences exist between import and<br />

export registration. In the previous step it appears that sometimes transit flows are larger or<br />

cannot be matched to international trade flows between the same countries and for the same<br />

commodity type. In this step these observations are further investigated in order to find a<br />

clarification and to see in what way the flows should be included in the O/D matrix.<br />

There is a relation between international trade flows with large differences between import and<br />

export registration and transit flows that cannot be matched with international trade flows. This<br />

relation is clarified by an example. In the Comext data there are two registrations for the trade<br />

of NSTR group 41 (iron ore) from the Netherlands to Germany: export registration of the<br />

Netherlands equals 8.9 million tonnes, import registration of Germany equals 26.9 million<br />

tonnes. There is a large gap between these registrations, besides, the Netherlands does not<br />

produce iron ore and thus the Netherlands cannot export these commodities. A possible<br />

explanation for this is that it does not concern a trade flow from the Netherlands to Germany,<br />

but a transit flow from a country via the Netherlands to Germany. The transit flows that cannot<br />

be matched to the international trade flows contains flows from Middle and Southern America<br />

via Rotterdam in the Netherlands to Germany. This confirms the idea that trade and transport<br />

are mixed up. The trade flow from Middle and Southern America via Rotterdam to Germany is<br />

registered by Germany as import from the Netherlands (the transport of the goods is considered<br />

as trade). This explains why the export registration of the Netherlands is much lower than the<br />

import registration of Germany (this flow is no trade of the Netherlands) and why the transit<br />

flow cannot be matched with the international trade data (the trade flow has a different origin<br />

country). This confusion between trade and transport is corrected in the O/D matrix by<br />

including the transit flows (in the example the flow from Middle and Southern America via<br />

Rotterdam to Germany including information about the transport modes used) and reducing the<br />

trade flow (trade from the Netherlands to Germany) with the same volume.<br />

50<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

This procedure is applied for all trade flows in the O/D matrix that have large differences<br />

between the import and the export registration as indicated by the variables indicating the<br />

absolute and the relative difference (see description method 1) and that can be linked to transit<br />

flows that have not been matched with international trade flows.<br />

In this step about 55 million tonnes of the total of 75 million tonnes that could not be matched in<br />

the first step is linked to international trade flows, simultaneously a correction for the confusion<br />

between trade and transport has been applied.<br />

Add remaining transit flows<br />

The remaining transit flows of about 20 million tonnes that could not be matched to<br />

international trade flows in the first and second step will be added to the O/D matrix.<br />

Finally, the commodity type information and the cargo characteristics information is added to<br />

the transit flows that have been included in the O/D matrix based on the distribution of the<br />

international trade flows.<br />

Import and export by sea<br />

In the test of the method data about import and export by sea for Spain and Germany is used in<br />

order to construct complete (intra­EU) maritime transport chains. For Spain trade data (source<br />

AEAT) is available including information about the border crossing region. Depending on the<br />

transport mode used for goods entering or leaving Spain, the border regions are equal to the port<br />

regions. The transport modes for the hinterland transport have been estimated based on<br />

domestic transport information. For Germany detailed data is available for German import and<br />

export via the ports of Hamburg and Bremen (although some missing elements have been<br />

estimated). Data about import and export via other German ports has been constructed by<br />

combining Eurostat port data with maritime data from SBA and national transport data from<br />

SBA and KBA. Import and export by sea data has the following format:<br />

· Origin region (NUTS2 region in case of export, otherwise country)<br />

· Transhipment region (NUTS2 region in the country)<br />

· Destination region (NUTS2 region in case of import, otherwise country)<br />

· Mode at origin<br />

· Mode at destination<br />

· Commodity type (NSTR2 or NSTR1 classification)<br />

· Transport volume (weight in tonnes)<br />

This data is used to estimate the origin/destination region, the transhipment region in the<br />

origin/destination region and the transport mode of the hinterland transport in the<br />

origin/destination country by commodity type (NSTR2 or NSTR1 classification) and by<br />

destination/origin country. Specific methods have been applied for different types of records in<br />

the database with two possible transhipment locations. It is noticed that ­ for instance for<br />

maritime transport from Spain to Germany ­ the port distribution in a country is estimated<br />

independently of the port distribution in the partner country (port distribution in Spain is<br />

Document2<br />

27 May 2004<br />

51


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

independently estimated of the port distribution in Germany, port distribution in Germany is<br />

independently estimated of the port distribution in Spain). After this method has been applied<br />

origin/destination regions (on NUTS2 level), transhipment regions (on NUTS2 level) and<br />

hinterland modes are available for all maritime transport leaving or entering Spain and<br />

Germany. Since the method has been applied for Spain and Germany, the complete maritime<br />

transport chain including all information on the required level of detail results from this method.<br />

Annual seaborne transport data Eurostat<br />

New Cronos from Eurostat contains (amongst others) the following maritime data: Annual<br />

seaborne transport between <strong>report</strong>ing countries/MCA (Main Coastal Area) and partner<br />

countries/MCA, by direction, type of cargo and nationality of vessel (in 1000 tonnes), year<br />

2000. Once all maritime transport chains have been estimated, the results will be validated (and<br />

eventually modified) using the seaborne transport data from Eurostat.<br />

Port­to­port transport data Eurostat<br />

Data about maritime transport with information on port­to­port level is collected by Eurostat<br />

(see ”Council directive 95/64/EC of 8 December 1995 on statistical returns in respect of carriage<br />

of goods and passengers by sea”). When this data becomes available (has already been<br />

requested by DGTREN), the method will be modified in order to include use of this data in<br />

order to produce maritime transport results that are more reliable (in the current method the port<br />

distribution on the origin/destination side is being estimated independently of the port<br />

distribution on the destination/origin side, the port­to­port data from Eurostat can be used to<br />

determine the exact port­to­port relations). When the port­to­port data becomes available, the<br />

validation with the seaborne transport data from Eurostat (described above) becomes redundant.<br />

With the port­to­port data a distinction can be made between an active mode and a passive<br />

mode for maritime transport.<br />

Method 3: Estimation of a region­to­region matrix<br />

Although Eurostat’s COMEXT database is limited in its spatial scope to country to country<br />

transactions, many European countries do calculate regional versions of their own trade statistics.<br />

The Concerted Action on Short Sea Shipping (CA­SSS) for DG­TREN, 1998/9, led by NTUA and<br />

ISL and in which NEA and MDS have been working on the statistical part, developed techniques<br />

for employing this existing regional data to build regional O/D matrices. These methods are being<br />

transferred to <strong>ETIS</strong>.<br />

The problem is a familiar one. The object is to construct an O/D matrix, where the cells represent<br />

the traffic ‘T’ between pairs of regions. The available data allows the row and column totals of the<br />

matrix to be measured, but an estimation procedure is required to fill the cells, for example:<br />

D 1 D 2 D 3 Total<br />

O 1 ΣO 1<br />

52<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

O 2 ΣO 2<br />

Total ΣD 1 ΣD 2 ΣD 3 ΣD 1..3 (= ΣO 1..2 )<br />

In this example, the sets of ‘O’s and ‘D’s belong to specific countries. Country to country totals are<br />

known, as are regions to/from country totals for a significant proportion of EU and accession<br />

countries.<br />

The solution attempted within CA­SSS was to derive a set of gravity models, one for each SITC<br />

commodity relating the volume of traffic in a given O/D cell to the distance between the origin and<br />

destination, and the size of the regions, measured by the total volumes of the commodity produced<br />

and consumed. Other measures of region ‘mass’ could be hypothesised, but in the CA­SSS study<br />

most were found to be closely correlated to the total import and export volumes.<br />

The functional form used was:<br />

T p = d n e ­md E p I p<br />

Where:<br />

T p<br />

d<br />

E p<br />

I p<br />

n,m<br />

= tonnes of product ‘p’ lifted<br />

= distance (kms)<br />

= Exports of product ‘p’<br />

= Imports of product ‘p’<br />

are the parameters to be estimated<br />

The use of a function that combines the exponential and power­n (where n is negative) curves<br />

provides greater control in circumstances where distance is close to zero, than would be the case if<br />

only the power­n curve were used. This is important if the parameters are being estimated for zone<br />

pairs where the distances are relatively high and then transferred to zone pairs where the distances<br />

could be small.<br />

The intention was to develop a technique that could estimate parameters for, potentially, a highly<br />

disaggregated set of commodities, and in which a number of non­linear functions could be tested<br />

and compared. The parameters would be estimated at the national level, for which large volumes<br />

of data are available, using regional information where possible. The same parameters would then<br />

be transferred to compute values at a region­region scale. So, for example, if at a national scale, the<br />

propensity to trade ( T p /E p I p ) for Portugal and Spain is higher than for Portugal and Finland, for a<br />

given product, it seems sensible to conclude that the propensity to trade for adjacent Portuguese and<br />

Spanish regions may be higher than for distant Portuguese and Spanish regions.<br />

It also appears sensible to distinguish between product groups, as, for example, any distance decay<br />

effects may be more significant for low value, high volume goods than for higher value densities.<br />

This is exemplified using results from the 1996/7 study, for two contrasting commodities.<br />

Document2<br />

27 May 2004<br />

53


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Distance Decay by Commodity<br />

120<br />

Propensity to Trade (Index)<br />

100<br />

80<br />

60<br />

40<br />

20<br />

Ores and Scrap<br />

Scientific Machinery<br />

­<br />

100<br />

200<br />

300<br />

400<br />

500<br />

600<br />

700<br />

800<br />

900<br />

1000<br />

1100<br />

1200<br />

1300<br />

1400<br />

1500<br />

1600<br />

1700<br />

1800<br />

1900<br />

2000<br />

Distance (Kms)<br />

The parameters calculated for these commodities indicate that distance decay is a relevant factor at<br />

the regional scale, with a high degree of impact between 200 and 1000 km. They also indicate that<br />

value density is a factor in determining the shape of the curve.<br />

The gravity model is used to estimate region­region traffics which in turn, are then used to seed the<br />

O/D matrix and a furnessing algorithm is then used to constrain the matrix to its (known) row and<br />

column totals.<br />

The estimation of the ‘n’ and ‘m’ parameters depends on a multidimensional optimisation<br />

technique known as the downhill simplex method, or “amoeba”. 5 This is a relatively simple search<br />

algorithm that can be applied to a wide range of functional forms, such that the calling routines can<br />

specify parameter ranges, evaluation criteria and numerical precision. It does not involve the use of<br />

derivatives or statistical methods such as regression analysis.<br />

A test was carried out in which a version of the gravity formula was used to generate some ‘real’<br />

data. A random number generator was then used to disturb these results within preset ranges, and<br />

these ‘sample’ points were then fed into a software routine that used the amoeba algorithm to detect<br />

the parameters.<br />

5<br />

Press WH, Teukolsky SA, Vetterling WT, Flannery BP, 1992, Numerical Recipes in C, Cambridge<br />

University Press<br />

54<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Tonnes Lifte d<br />

100000.0<br />

True Plus Error<br />

TRUE<br />

Estimated<br />

Tonnes<br />

10000.0<br />

50<br />

100<br />

150<br />

200<br />

250<br />

300<br />

350<br />

400<br />

450<br />

500<br />

550<br />

Km s<br />

600<br />

650<br />

700<br />

750<br />

800<br />

850<br />

900<br />

950<br />

1000<br />

The figure shows the ‘true’ relationship between distance and traffic generated with the solid line.<br />

The ‘sample’ points are shown as dots (note that the scale is logarithmic), and the error term is as<br />

much as <strong>plus</strong>/minus 50% of the actual. The broken line represents the curve that the amoeba<br />

algorithm estimates from the sample so that the sum of the errors vis­a­vis the sample points is<br />

minimized. This indicates a very good approximation of the original function, from only 20 sample<br />

points, which improves as the error term is reduced. The test implies that this simple estimation<br />

technique provides an accurate and computationally effective method for estimating gravity model<br />

parameters where the functions are non­linear, and where a large number of models for different<br />

commodities are required.<br />

Although parameters were calculated within CA­SSS it now makes sense to update these results<br />

with the method developed here and to use a new commodity classification system. Conversion<br />

tables could be used but recalculation is favoured allowing more base year data to be included in<br />

the estimation, and different functional forms to be compared.<br />

The matrix estimation method can be illustrated with trade data.<br />

The Year 2000 COMEXT database shows the following patterns of trade under the SITC heading<br />

‘89’ Miscellaneous Manufactures, for ten countries.<br />

Document2<br />

27 May 2004<br />

55


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Year 2000: Trade Volumes, Tonnes, SITC 89, Miscellaneous Manufactures<br />

F NL D IT UK IRL DK EL PT Total<br />

F 816,186<br />

NL 667,127<br />

D 1,403,972<br />

IT 960,777<br />

UK 519,952<br />

IRL 141,677<br />

DK 125,798<br />

EL 22,545<br />

PT 68,102<br />

Total 1,149,272 406,307 939,382 304,014 986,052 158,838 183,114 500,563 98,594 4,726,136<br />

In fact, all the cells in this table are known, but if region to region totals rather than country to<br />

country totals are sought, a method has to be found to complete the matrix. Therefore the test<br />

has been carried out at the higher level so that different estimation methods can be compared<br />

against known values.<br />

For reference, the actual matrix is:<br />

Year 2000: Trade Volumes, Tonnes, SITC 89, Miscellaneous Manufactures: Actuals<br />

F NL D IT UK IRL DK EL PT Total<br />

F 42,646 305,370 81,523 212,456 5,734 8,779 127,529 32,149 816,186<br />

NL 157,066 229,456 37,627 159,176 7,616 31,419 31,298 13,469 667,127<br />

D 444,615 245,475 129,615 363,540 17,501 106,381 71,863 24,982 1,403,972<br />

IT 359,208 37,894 193,248 130,637 5,911 14,713 202,331 16,835 960,777<br />

UK 121,061 56,900 92,344 37,424 120,762 18,461 63,197 9,803 519,952<br />

IRL 28,327 4,066 14,550 4,876 85,103 1,975 2,411 369 141,677<br />

DK 19,991 13,039 59,594 4,789 25,413 956 1,060 956 125,798<br />

EL 3,704 5,669 4,618 5,413 2,569 182 359 31 22,545<br />

PT 15,300 618 40,202 2,747 7,158 176 1,027 874 68,102<br />

Total 1,149,272 406,307 939,382 304,014 986,052 158,838 183,114 500,563 98,594 4,726,136<br />

The first possibility is to fill in the matrix using a purely mechanical Furness algorithm, which<br />

attempts to find a solution (out of many feasible solutions) without the need for any underlying<br />

model. It works by seeding the initial matrix with zeroes on the leading diagonal, and ones<br />

elsewhere, and then successively factoring up the rows and then the columns so that the sums in<br />

the rows and columns tend toward the known row and column totals. After a few cycles, the<br />

algorithm “succeeds”.<br />

A Solution by Furnessing at the 2 Digit Level<br />

F NL D IT UK IRL DK EL PT Total<br />

F 0 88,519 264,571 70,492 211,403 30,995 35,631 95,597 18,978 816,186<br />

NL 180,483 0 176,938 47,143 141,380 20,729 23,829 63,933 12,692 667,127<br />

D 461,227 151,284 0 120,475 361,300 52,972 60,896 163,382 32,435 1,403,972<br />

56<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

IT 255,312 83,743 250,297 0 199,998 29,323 33,709 90,440 17,954 960,777<br />

UK 160,429 52,621 157,278 41,905 0 18,425 21,182 56,829 11,282 519,952<br />

IRL 36,239 11,887 35,527 9,466 28,388 0 4,785 12,837 2,548 141,677<br />

DK 32,320 10,601 31,685 8,442 25,317 3,712 0 11,449 2,273 125,798<br />

EL 6,143 2,015 6,022 1,605 4,812 706 811 0 432 22,545<br />

PT 17,223 5,649 16,885 4,499 13,492 1,978 2,274 6,101 0 68,102<br />

Total 1,149,376 406,319 939,204 304,026 986,091 158,840 183,116 500,569 98,595 4,726,136<br />

The first improvement considered was to carry out the same process, but with further<br />

disaggregation by commodity. The data was therefore disaggregated into nine 3 digit SITC<br />

groups. As the algorithm is so straightforward, there is very little extra overhead involved in<br />

this step. The result is:<br />

A Solution by Furnessing at the 3 Digit Level<br />

F NL D IT UK IRL DK EL PT Total<br />

F 0 84,855 263,541 70,517 213,886 28,964 36,600 98,574 19,248 816,186<br />

NL 180,686 0 182,292 47,370 141,986 20,704 23,693 57,030 13,368 667,127<br />

D 481,226 161,631 0 118,699 378,145 56,627 61,486 114,592 31,567 1,403,972<br />

IT 233,890 77,001 232,976 0 184,096 27,567 31,779 155,660 17,807 960,777<br />

UK 164,494 52,001 164,700 41,359 0 18,234 22,102 45,829 11,233 519,952<br />

IRL 35,235 11,999 39,500 9,837 26,004 0 4,565 11,866 2,670 141,677<br />

DK 32,734 11,545 34,218 8,227 25,339 4,135 0 7,337 2,263 125,798<br />

EL 5,595 1,654 5,283 3,057 5,165 594 754 0 444 22,545<br />

PT 15,647 5,666 16,422 4,955 11,554 2,022 2,146 9,690 0 68,102<br />

Total 1,149,507 406,353 938,932 304,021 986,175 158,848 183,123 500,577 98,599 4,726,136<br />

The result is not dissimilar, but the overall error is lower, and it is now possible to generate<br />

tables at a higher level of detail, which may be advantageous for later stages of <strong>ETIS</strong>. For<br />

example:<br />

A Solution by Furnessing for SITC 891, Arms and Ammunition<br />

F NL D IT UK IRL DK EL PT Total<br />

F 0 18 396 2,052 1,285 25 131 235 138 4,280<br />

NL 290 0 165 855 535 10 55 98 58 2,065<br />

D 362 9 0 1,067 668 13 68 122 72 2,381<br />

IT 2,193 56 1,248 0 4,049 78 413 740 436 9,214<br />

UK 326 8 185 961 0 12 61 110 65 1,729<br />

IRL 0 0 0 0 0 0 0 0 0 1<br />

DK 57 1 32 168 105 2 0 19 11 396<br />

EL 638 16 363 1,882 1,179 23 120 0 127 4,349<br />

Document2<br />

27 May 2004<br />

57


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

PT 8 0 4 22 14 0 1 3 0 53<br />

Total 3,873 109 2,394 7,007 7,836 163 851 1,327 908 24,468<br />

By calculating the cell errors, and simply shading the cells involving neighbouring countries, it<br />

is fairly obvious that the errors are strongly correlated with distance, i.e. this method tends to<br />

under­estimate flows where the distance is small. It would follow that if the matrix is seeded<br />

with initial values that reflect distance, the estimation accuracy should improve.<br />

Analysis of Errors: Actual vs Estimated (based on 3 Digit Furnessing)<br />

F NL D IT UK IRL DK EL PT<br />

F 0 42,209 ­41,829 ­11,006 1,430 23,230 27,821 ­28,955 ­12,901<br />

NL 23,620 0 ­47,164 9,743 ­17,190 13,088 ­7,726 25,732 ­101<br />

D 36,611 ­83,844 0 ­10,916 14,605 39,126 ­44,895 42,729 6,585<br />

IT ­125,318 39,107 39,728 0 53,459 21,656 17,066 ­46,671 972<br />

UK 43,433 ­4,899 72,356 3,935 0 ­102,528 3,641 ­17,368 1,430<br />

IRL 6,908 7,933 24,950 4,961 ­59,099 0 2,590 9,455 2,301<br />

DK 12,743 ­1,494 ­25,376 3,438 ­74 3,179 0 6,277 1,307<br />

EL 1,891 ­4,015 665 ­2,356 2,596 412 395 0 413<br />

PT 347 5,048 ­23,780 2,208 4,396 1,846 1,119 8,816 0<br />

An allowance for distance has been introduced by using the gravity model described above.<br />

The amoeba algorithm was used to calculate ‘n’ and ‘m’ parameters, and then the gravity<br />

function was used to seed the matrix.<br />

A Solution by Furnessing a Seeded Matrix at the 2 Digit Level<br />

F NL D IT UK IRL DK EL PT Total<br />

F 0 63,087 257,750 87,545 263,764 25,400 26,260 75,374 17,006 816,186<br />

NL 131,527 0 240,474 30,627 171,076 16,474 17,032 48,887 11,030 667,127<br />

D 469,180 209,961 0 130,518 294,419 47,174 81,149 139,989 31,584 1,403,972<br />

IT 294,021 49,339 240,812 0 150,721 29,562 30,563 145,966 19,793 960,777<br />

UK 190,362 59,222 116,732 32,388 0 31,846 19,788 56,798 12,815 519,952<br />

IRL 24,128 10,026 22,094 6,282 60,295 0 4,173 11,978 2,703 141,677<br />

DK 18,144 9,218 44,234 6,784 27,205 4,359 0 12,935 2,918 125,798<br />

EL 4,646 1,445 4,740 3,847 4,850 1,116 1,154 0 747 22,545<br />

PT 17,409 4,069 12,369 6,033 13,656 2,912 3,011 8,643 0 68,102<br />

Total 1,149,417 406,366 939,204 304,023 985,986 158,844 183,131 500,570 98,596 4,726,136<br />

If the errors from this method are examined in the same way, the correlation between the error<br />

sign and the distance can be seen to be lower.<br />

Analysis of Errors (Tonnes): Actual vs Seeded Result<br />

F NL D IT UK IRL DK EL PT<br />

F 0 20,441 ­47,620 6,022 51,308 19,666 17,481 ­52,155 ­15,143<br />

NL ­25,539 0 11,018 ­7,000 11,900 8,858 ­14,387 17,589 ­2,439<br />

D 24,565 ­35,514 0 903 ­69,121 29,673 ­25,232 68,126 6,602<br />

IT ­65,187 11,445 47,564 0 20,084 23,651 15,850 ­56,365 2,958<br />

UK 69,301 2,322 24,388 ­5,036 0 ­88,916 1,327 ­6,399 3,012<br />

58<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

IRL ­4,199 5,960 7,544 1,406 ­24,808 0 2,198 9,567 2,334<br />

DK ­1,847 ­3,821 ­15,360 1,995 1,792 3,403 0 11,875 1,962<br />

EL 942 ­4,224 122 ­1,566 2,281 934 795 0 716<br />

PT 2,109 3,451 ­27,833 3,286 6,498 2,736 1,984 7,769 0<br />

The total error is lower using this method, and although the correlation with distance is still<br />

present, it is also lower.<br />

The conclusion is therefore that a combination of these techniques should be used. The data<br />

should be disaggregated by commodity, and by region where possible. The amoeba algorithm<br />

should be used to establish parameters for the gravity model, using known flows. These<br />

parameters should then be used to seed the region to region matrices, and the Furness algorithm<br />

to build up the O/D matrices.<br />

Method 4: Inclusion of domestic transport to the matrix<br />

In the fourth method domestic transport is included in the O/D matrix. Data about domestic<br />

transport flows is collected from national statistical offices and Eurostat. For a number of<br />

countries region­to­region by commodity type by transport mode information is available. For<br />

other countries only more aggregate data is available. In the latter case, data sources are<br />

combined and procedures are applied in order to estimate the domestic transport flows on the<br />

required level of detail.<br />

Once the domestic transport data is available, it can be added to the O/D matrix. However, if all<br />

domestic transport is added to the O/D matrix double countings are introduced because export<br />

and import flows by sea with hinterland transport in the origin/destination country already<br />

include domestic transport.<br />

Document2<br />

27 May 2004<br />

59


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

In table 6.5 this is illustrated by an example.<br />

Table 6.5<br />

Origin<br />

region<br />

Mode<br />

origin<br />

Illustration of double countings in domestic transport<br />

Transh.<br />

region<br />

Madrid (ES) road Pais Vasco<br />

(ES)<br />

Mode Transh. region Mode<br />

destination<br />

sea<br />

South­West<br />

(UK)<br />

rail<br />

Destination<br />

region<br />

West Midlands<br />

(UK)<br />

Madrid (ES) road Pais Vasco<br />

(ES)<br />

South­West<br />

(UK)<br />

rail<br />

West Midlands<br />

(UK)<br />

Volume<br />

(tonnes)<br />

Flow type<br />

2000 International<br />

30000 Domestic<br />

70000 Domestic<br />

This table shows an international transport flow from Spain to the UK by sea with hinterland<br />

transport in Spain and in the UK. The hinterland transport (road transport from origin region to<br />

the port region in Spain, rail transport from the port region to the destination region in the UK)<br />

that is part of an international transport chain is also registered in domestic transport. This<br />

means that if all domestic transport would be included in the O/D matrix, the hinterland flows<br />

would be included twice; once as part of an international transport chain and once as domestic<br />

transport. In order to include the pure domestic transport on the O/D matrix the domestic<br />

transport has to be reduced by the hinterland flows (specified by origin region, by destination<br />

region and by commodity group). In the example given above, the volume of 30.000 tonnes<br />

domestic road transport between Madrid and Pais Vasco and the volume of 70.000 tonnes<br />

domestic rail transport between region South­West and region West Midlands should be<br />

reduced by 2.000 tonnes.<br />

Method 5: Estimation of cargo characteristics and transport unit information<br />

Trade data is highly detailed with respect to commodity type, but this is not necessarily a useful<br />

way of segmenting demand within the freight sector. Instead it is necessary to segment traffic<br />

flows according to their handling characteristics. For example, temperature controlled goods are<br />

limited to certain transport modes, e.g. driver accompanied refrigerated trailer, refrigerated<br />

container, and to certain transport services, e.g. container ships with plug­in points.<br />

The challenge is that these characteristics have to be inferred from the commodity description,<br />

and cannot necessarily be drawn directly from trade data. The COMEXT database is published<br />

using the international 8 digit combined nomenclature (CN8) system. Because this database<br />

contains very detailed commodity information, the cargo characteristics are added to the<br />

database as one of the first steps where this detail is still available, afterwards when the data is<br />

aggregated to the NSTR 2 digit commodity level the cargo characteristics information stays<br />

available and accessible.<br />

The handling characteristics chosen for the testing phase of <strong>ETIS</strong> are:<br />

· Cargo Type/Mode of Appearance: Liquid Bulk, Dry Bulk, General Cargo, and Containers<br />

· Hazardous/Non Hazardous: e.g. flammable or toxic cargo<br />

60<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

· Temperature Controlled/Ambient: e.g. fruit<br />

· Containerised/Non­containerised<br />

· Transport Units: number of vehicles<br />

The main decision relates to the degree of commodity aggregation at which to make the<br />

estimation. If an aggregate approach is taken, the calculation procedure is relatively simple, and<br />

the conversion factors can be ‘hand­crafted’. However, if a disaggregate approach is taken, the<br />

conversion rules are far less variable from corridor to corridor, and across time periods.<br />

This can be shown in relation to a simplified case, where an aggregated product group ‘Metals’<br />

can be disaggregated into three types, each with different handling characteristics as far as rates<br />

of containerisation are concerned.<br />

Corridor 1<br />

Container Non Container. Container % Non Container %<br />

Metals 1500 1500 50% 50%<br />

Metals: Type 1 500 500 50% 50%<br />

Metals: Type 2 1000 0 100% 0%<br />

Metals: Type 3 0 1000 0% 100%<br />

Corridor 2<br />

Container Non Container. Container % Non Container %<br />

Metals 1600 2000 44% 56%<br />

Metals: Type 1 1500 1500 50% 50%<br />

Metals: Type 2 100 0 100% 0%<br />

Metals: Type 3 0 500 0% 100%<br />

By varying the product mix across the two corridors, the aggregate containerisation factors<br />

change even though the disaggregate factors do not. We can offer the hypothesis that handling<br />

factors depend upon:<br />

· The product type<br />

· The heterogeneity of the product type<br />

· The corridor, and<br />

· the balance of trade across the corridor<br />

The strength of these factors will vary, depending upon the factor considered. Of these, the key<br />

element which can be controlled is the degree of heterogeneity within the product category. By<br />

using the diasggregated form of the COMEXT database (using approximately 10,000 CN8<br />

headings) it is plausible to assume that for certain factors (e.g. hazardous goods, chilled goods)<br />

codes can be simply labelled as ‘X’ or ‘not X’ rather than, say ‘15% X’ or ‘85% not X’, where<br />

X is any given factor. This avoids having to adjust the factors depending on the underlying mix.<br />

Document2<br />

27 May 2004<br />

61


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

The corridor element is harder to control without specific data about individual country pairs.<br />

Of particular relevance is the extent of containerisation. The fact that for any given corridor, the<br />

balance of container flows may be different to the balance of cargo flows, implies that rates of<br />

containerisation will vary by direction. Furthermore, the absolute size of the flow is also<br />

important. Small flows of conventional cargoes may well be containerised, but there will be<br />

thresholds beyond which dedicated bulk transport services will be used.<br />

The approach that has been tested involves using the disaggregated (CN8) COMEXT data and<br />

applying look­up tables developed by MDS­Transmodal.<br />

Commodity Conversion Table<br />

Commodity Category Handling Mode of Stowage Containerisation<br />

CN8 NSTR SITC Description Haz Chill Appearance Per<br />

TEU<br />

4,031,099 139 2231 Yoghurt 0 1 General Cargo 17 22 100 100 95 100<br />

8,051,009 31 5711 Oranges 0 1 Semi Bulk 10 17 100 100 30 30<br />

15,191,100 182 43131 Fatty Acids 0 0 Liquid Bulk 12 22 90 90 0 90<br />

27,071,010 831 33522 Benzole 1 0 Dry Bulk 18 22 0 0 0 0<br />

27,090,000 310 33300 Petroleum Oils 1 0 Crude Oil 18 22 0 0 0 0<br />

44,103,200 976 63422 Particle Board 0 0 Semi Bulk 10 20 80 80 15 15<br />

85,281,099 931 76110 Colour TVs 0 0 General Cargo 6 11 100 100 100 100<br />

87,033,219 910 78120 Motor Vehicles 0 0 Vehicles 4 8 15 15 15 15<br />

Per<br />

FEU<br />

Intra<br />

Exp<br />

Intra<br />

Imp<br />

Extra<br />

Exp<br />

Extra<br />

Imp<br />

The table shows the key data fields for a diverse range of commodities. The trade data is linked<br />

through the index field (CN8) containing the eight digit product code, the maximum degree of<br />

product detail possible within multi­country trade data. Conversions can be made to NSTR and<br />

SITC coding systems, and the description at the five digit SITC level is also shown. The<br />

handling factors for hazardous cargo (HAZ) and temperature controlled (CHILL) is a simple<br />

boolean variable. Then the stowage factors are shown, giving conversion ratios for tonnes into<br />

unit loads: per 20 foot unit (TEU) and per 40 foot unit (FEU). From these factors it is possible<br />

to convert to lorry loads. Note that 1 FEU does not necessarily equal 2 TEUs, since many<br />

commodities ‘weigh­out’ before they ‘cube­out’, i.e. that they reach the maximum permitted<br />

weight for road haulage before they are full.<br />

Finally, the rates of containerisation are given. These are based partially on trade data records,<br />

but they have been simplified. Different rates are apparent for the simplified corridors: intra<br />

and extra EU, and import and export. Intra EU imports and exports are not differentiated<br />

because it would lead to ambiguities, but extra and intra flows will be different, and extra<br />

imports will be different from extra exports.<br />

Applying these containerisation factors will lead to an indicator of trade which needs to be<br />

defined carefully. Note that they will not show the total number of TEUs or container units<br />

handled in Europe, as they do not take into account trans­shipment or empty returns. It will<br />

show the estimated volume of containerised trade expressed as tonnes or TEU for a given<br />

62<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

European country. This is therefore an indirect measure of port traffic, but a direct measure of<br />

the underlying demand for transport services.<br />

The translation from trade data to the required quantities has been carried out within the <strong>ETIS</strong><br />

pilot using a simple Microsoft Access SQL query. Access has also been used to aggregate<br />

across the NST commodity definition.<br />

Contrary to international trade, for domestic transport less detail is available concerning the<br />

commodity type (at most information is available on NSTR2 digit level instead of the CN 8<br />

digit level). As a consequence the cargo characteristics and information about the number of<br />

vehicles have to be determined in a different way. Available information from Eurostat and<br />

national statistical offices is collected as much as possible. This data is used to estimate average<br />

figures that are used to estimate cargo characteristics and transport unit information for<br />

domestic transport.<br />

Method 6: Estimation of transport performance information<br />

Transport volume information, expressed in weight of the goods (tonnes) and expressed in value<br />

of the goods (euro), is included in the data right from the first step in the method. For transport<br />

performance information the variable number of TEUs is included from the first step of the<br />

method while information about the number of vehicles is included in the data in method 5.<br />

Transport performance information expressed in tonne­kilometres, vehicle­kilometres and TEUkilometres<br />

still has to be calculated.<br />

In order to be able to calculate these variables information is needed about the transport distance<br />

between regions (on NUTS2 level) by transport mode. Currently this data is available form<br />

transport networks available at NEA (from NEAC) that have also been used in the TEN­STAC<br />

project. After completion of <strong>ETIS</strong> <strong>WP</strong>5 ‘European transport network data’ this data will be<br />

available from <strong>WP</strong>5.<br />

For this calculation it is not only necessary to have distance information by mode between the<br />

origin and destination region, but also between transhipment locations and between<br />

origin/destination regions and transhipment locations. A special procedure will be applied for<br />

the intra­regional transport distances. For intra­regional transport no distance can be determined<br />

based on a transport network since the origin and the destination are the same. A procedure will<br />

be applied in which the distance for intra­regional transport will be determined by region based<br />

on a number of characteristics of the region (such as the area of a region and/or economic<br />

activity of a region).<br />

Once the transport distance data is available in the requested format, the transport performance<br />

information is calculated by multiplying the distance by the relevant variables.<br />

Validation and modification of results – use of CAFT and national sources<br />

Finally, when the final freight O/D matrix has been constructed, the results will be validated by<br />

comparing the results with CAFT data and with data from national sources (regional<br />

Document2<br />

27 May 2004<br />

63


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

distribution, loading factor, containerisation rate, etc.). In case this comparison shows large<br />

differences, the results will be modified in such a way that they match the data used for the<br />

validation.<br />

6.3 Lessons learned and changes compared to the original planned <strong>methodology</strong><br />

In general the <strong>methodology</strong> has been applied according to plan. Some minor changes have been<br />

made that are listed below.<br />

Intermodal transport<br />

For intermodal transport not enough information is available to include it in the freight O/D<br />

matrix. Because estimation methods are too time consuming to apply it has been decided to<br />

leave intermodal transport out of the database.<br />

Different method for the accession countries<br />

For the accession countries a different method will be applied. Available data from – amongst<br />

others – the Intermoda project, NEAC and TEN­STAC – will be used for these countries. Once<br />

the accession countries have become member of the EU and an update of the <strong>ETIS</strong> reference<br />

database will be made in the future, the same <strong>methodology</strong> can be applied for the accession<br />

countries as it is now applied for the current EU countries.<br />

Maritime transport data<br />

It appears that more maritime transport data is available than expected at first hand. From<br />

national statistics data is collected about maritime import and export flows by unloading/loading<br />

port in the country. Besides, Eurostat collects detailed port­to­port data.<br />

Continuation<br />

For the construction of a complete freight O/D matrix that will be part of the <strong>ETIS</strong> reference<br />

database, the tested methods will be applied.<br />

6.4 Output of the testing phase<br />

The output of the testing phase consists of a freight O/D transport chain database and a transport<br />

database. The transport chain database (origin and destination of the goods) describes trade<br />

flows containing the following variables:<br />

· Origin region or country (NUTS2 or <strong>ETIS</strong> country classification)<br />

· First transhipment location (NUTS2 classification)<br />

· Second transhipment location (NUTS2 classification)<br />

· Destination region or country (NUTS2 or <strong>ETIS</strong> country classification)<br />

· Mode at origin<br />

· Mode between transhipment locations<br />

64<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

· Mode at destination<br />

· NSTR 2 digit commodity classification<br />

· Indicator hazardous goods<br />

· Indicator conditioned goods<br />

· Manifestation of the goods<br />

· Value of the goods (in Euro)<br />

· Volume of the goods (in tonnes)<br />

· Volume of unitised goods (in tonnes)<br />

· Number of TEUs<br />

· Indicator absolute difference between import and export registration<br />

· Indicator relative difference between import and export registration<br />

· Indicator for intra­EU or extra­EU trade<br />

The transport database describes transport flows (origin and destination of the transport unit)<br />

containing the following variables:<br />

· Origin country (<strong>ETIS</strong> country classification)<br />

· Destination country (<strong>ETIS</strong> country classification)<br />

· Transport mode<br />

· NSTR 2 digit commodity classification<br />

· Indicator hazardous goods<br />

· Indicator conditioned goods<br />

· Manifestation of the goods<br />

· Transport distance<br />

· Value of the goods (in Euro)<br />

· Volume of the goods (in tonnes)<br />

· Tonne­kilometres<br />

· Number of transport units<br />

· Vehicle/vessel­kilometres<br />

· Volume of unitised goods (in tonnes)<br />

· Number of TEUs<br />

· TEU­kilometres<br />

· Indicator absolute difference between import and export registration<br />

· Indicator relative difference between import and export registration<br />

· Indicator for intra­EU or extra­EU trade<br />

6.5 Interpretation of the output data for the users<br />

As becomes clear from the section above, the results include very detailed data that are partly<br />

estimated. The more detailed the (estimated) data, the lower the reliability of the data. On the<br />

lowest level of detail, the results are very detailed. Consider for instance trade consisting of<br />

three transport legs: from region Madrid in Spain by road to port region Pais Vasco in Spain,<br />

then by sea to region Zuid­Holland in the Netherlands, and finally by inland waterways to<br />

Document2<br />

27 May 2004<br />

65


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

region Ruhrgebiet in Germany. The goods are characterised by commodity group and<br />

characteristics such as manifestation, conditioned, hazardous and/or unitised. The volume of the<br />

goods is expressed in tonnes, value, tonne­kilometres, number of vehicles/vessels,<br />

vehicle/vessel­kilometres, number of TEUs and TEU­kilometres. Because part of the<br />

information is estimated, on this level of detail the results should be interpreted in a proper way<br />

by transport experts. As the data is aggregated to a less detailed level, the results become more<br />

reliable and can be easier understand and interpreted.<br />

Detailed data can only be interpreted in the proper way by transport experts, in order to avoid<br />

problems with the interpretation of the results common users should only get access to<br />

aggregated OD data.<br />

The most detailed OD data can only be used by experts who are trained on the specific<br />

characteristics of the database. Developing a training program after the project might be an<br />

option. More aggregate sub­matrices will be developed for the final product that are easier to<br />

understand and interpreted by a broader audience.<br />

66<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

7 OBSERVATIONS UP TO PHASE 2<br />

7.1 Introduction<br />

This chapter describes the observations up to phase 2 of <strong>ETIS</strong> reference database on legal<br />

aspects and organisational aspects. These descriptions will be extended along the course of the<br />

project based on experiences.<br />

7.2 Legal aspects and organisational aspects<br />

EUROSTAT has provided data to <strong>ETIS</strong> free of charge for use within the project. No<br />

arrangements have been made yet on future use of the data by third parties outside the<br />

Commission. For now the data have to be considered restricted.<br />

Some data available at EUROSTAT (for instance international interregional road transport data)<br />

cannot be provided due to the statistical law or because of confidentiality reasons. For other data<br />

available at EUROSTAT (for instance the port­to­port data) a license has to be signed in which<br />

it is expressed that the data will only be used in this project.<br />

The conditions of data delivery from national statistical offices also differ from country to<br />

country. For some countries the data can be downloaded from the internet without any<br />

restrictions (for instance trade data of Spain), the only restriction is that the statistical office<br />

should be mentioned as source (for instance transport data in the Netherlands), a price has to be<br />

paid and a contract has to be signed for restricted use (for instance transport data in Germany) or<br />

in addition an official letter from the European Commission is required (for instance transport<br />

data France).<br />

As has also been discussed at the second open conference of <strong>ETIS</strong> it is recommended that a<br />

discussion is opened on how this data collected can be streamlined in the future as a follow up<br />

of the current <strong>ETIS</strong> projects. Should this be done through EUROSTAT or is a less formalised<br />

approach more appropriate. In any case it should be attempted to come to long lasting<br />

agreements on data delivery to avoid time consuming data collection procedures.<br />

7.3 Other remarks<br />

It has been observed in chapter 6 that the resulting OD matrix is very detailed which requires<br />

caution in its use. In case the most detailed level is used the reliability of the figures is in some<br />

cases relatively low due to the fact that some elements have been estimated. The user therefore<br />

has to be guided. One option is to provide only sub­matrices that do not contain all the variables<br />

in one matrix, to be used by the common users. The complete matrix can be used by expert<br />

users only which need to be trained to make the proper selections and interpretations.<br />

Document2<br />

27 May 2004<br />

67


ANNEX A: LIST OF INDICATORS CONSIDERED<br />

IN <strong>WP</strong> 3


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Mode Spatial scope Forecast Details to be provided<br />

Type of data<br />

required<br />

Data availability<br />

ref.<br />

Definition<br />

Measurement<br />

unit<br />

Road<br />

Rail<br />

Air<br />

Inland w.<br />

Sea<br />

Intermodal<br />

Network<br />

O­D<br />

Zone<br />

EU<br />

AC<br />

1.1.4<br />

Breakdown of journeys by<br />

origin/destination, proportion<br />

of long distance traffic using<br />

the TEN<br />

<strong>WP</strong>:3,4,5<br />

Number of<br />

percentage of<br />

total demand<br />

x TEN Y<br />

OD Matrices at the<br />

national level,<br />

roadside interviews,<br />

licence plate camera<br />

information (foreign<br />

plates).<br />

* *<br />

1.5.1<br />

2.1.2<br />

Road freight volumes on<br />

TEN by cargo type<br />

<strong>WP</strong>: 3,5<br />

Traffic volumes on the Trans­<br />

European rail network, by<br />

type (passenger / freight),<br />

including non fulfilled<br />

demand<br />

Freight vehiclekm<br />

Tonne km<br />

Number of trucks<br />

Number of trainkm<br />

by train type<br />

Passenger­km or<br />

tonne­km<br />

x TEN Y<br />

x TEN Y<br />

Journey type<br />

(domestic/international)<br />

Annual, weekly, daily,<br />

peaks<br />

Liquids, dry, bulk,<br />

general cargo, containers<br />

Hazardous, nonhazardous<br />

National/international<br />

journey<br />

Train type (high<br />

speed,…)<br />

Traffic surveys * *<br />

Rail traffic and<br />

passenger surveys<br />

Statistics on<br />

unfulfilled demand<br />

** **<br />

<strong>WP</strong>: 3,4,5<br />

Document2<br />

27 May 2004<br />

71


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Mode Spatial scope Forecast Details to be provided<br />

Type of data<br />

required<br />

Data availability<br />

ref.<br />

2.5.1<br />

Definition<br />

Freight volumes on TEN by<br />

train type<br />

<strong>WP</strong>: 3,5<br />

Measurement<br />

unit<br />

Freight train­km<br />

Tonne km<br />

(cross section<br />

volumes)<br />

Road<br />

Rail<br />

Air<br />

Inland w.<br />

Sea<br />

Intermodal<br />

Network<br />

O­D<br />

Zone<br />

x TEN Y<br />

Journey type<br />

(domestic/international)<br />

Train type (unitised,<br />

dedicated bulk, mixed<br />

bulk)<br />

Cargo type (liquids, dry,<br />

bulk, general cargo,<br />

containers)<br />

Risk level (hazardous,<br />

non­hazardous)<br />

EU<br />

AC<br />

Rail freight statistics ** **<br />

4.1.2<br />

Freight volumes on the inland<br />

waterway network<br />

<strong>WP</strong>: 3,5<br />

Tonnes, TEU<br />

Tonne­km, TEUkm<br />

x TEN Y<br />

Cargo type (Liquids, dry,<br />

bulk, general cargo,<br />

containers)<br />

Hazardous, nonhazardous<br />

Freight demand<br />

surveys<br />

* *<br />

5.1.1<br />

Port throughput (passengers,<br />

freight)<br />

<strong>WP</strong>: 3,4,5<br />

Number of units<br />

(tonnes,<br />

passengers, and<br />

vehicles), using<br />

an AADT type<br />

calculation<br />

Number of ship<br />

departures<br />

(AADT type<br />

calculation), by<br />

x<br />

x<br />

By port<br />

Country<br />

Y<br />

National, European,<br />

Intercontinental journey.<br />

Freight, Passengers,<br />

vehicle types<br />

Port statistics by<br />

categories listed above<br />

** *<br />

72<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Mode Spatial scope Forecast Details to be provided<br />

Type of data<br />

required<br />

Data availability<br />

ref.<br />

Definition<br />

Measurement<br />

unit<br />

ship type<br />

Number of<br />

containers (TEU)<br />

Peak period<br />

measure<br />

Road<br />

Rail<br />

Air<br />

Inland w.<br />

Sea<br />

Intermodal<br />

Network<br />

O­D<br />

Zone<br />

EU<br />

AC<br />

4.1.3<br />

I/C factor of bridges and locks<br />

(intensity versus capacity)<br />

(waiting time)<br />

<strong>WP</strong>: 3,4,5<br />

If I/C factor is<br />

above 0,5 / 0,6<br />

waiting time<br />

grows to<br />

unacceptable<br />

figures<br />

x x Y<br />

European<br />

classification system<br />

Design documentation,<br />

infrastructure<br />

characteristics<br />

** ?<br />

6.1.1<br />

Traffic volumes served at the<br />

terminal<br />

<strong>WP</strong>: 3,5<br />

Passengers<br />

Tonnes<br />

TEU<br />

x TEN Y<br />

Terminal type (road/rail,<br />

seaport/rail,….)<br />

Cargo type<br />

Terminal statistics * *<br />

1.3.1<br />

Estimated / measured energy<br />

consumption on roads<br />

<strong>WP</strong>: 3,4,8<br />

toe per vehiclekm<br />

x<br />

By<br />

road<br />

section<br />

Country<br />

Y<br />

Traffic volumes,<br />

estimated energy<br />

consumption data<br />

***<br />

(country)<br />

* (by road<br />

section)<br />

**<br />

(country)<br />

* (by road<br />

section)<br />

2.3.1<br />

Estimated/measured energy<br />

consumption on railways<br />

<strong>WP</strong>: 3,4,8<br />

toe ( tonnes oil<br />

equivalent) per<br />

train­km<br />

x<br />

By rail<br />

section<br />

Country<br />

Y<br />

Number of train­km<br />

Estimated energy<br />

consumption data<br />

*** **<br />

Document2<br />

27 May 2004<br />

73


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Mode Spatial scope Forecast Details to be provided<br />

Type of data<br />

required<br />

Data availability<br />

ref.<br />

4.2.1<br />

1.3.2<br />

2.3.2<br />

4.2.2<br />

Definition<br />

Estimated/Measured Energy<br />

Consumption on waterways<br />

<strong>WP</strong>: 3,4,8<br />

Estimated /measured road<br />

transport emissions<br />

+corresponding marginal unit<br />

cost<br />

<strong>WP</strong>: 3,4,8<br />

Estimated /measured rail<br />

transport emissions<br />

+corresponding marginal unit<br />

cost<br />

<strong>WP</strong>: 3,4,5,8<br />

Estimated /measured Inland<br />

waterway transport emissions<br />

+corresponding marginal unit<br />

cost<br />

<strong>WP</strong>: 3,4,5,8<br />

Measurement<br />

unit<br />

toe (tonnes oil<br />

equivalent) per<br />

tonne­km<br />

Tonnes of<br />

pollutant per<br />

million vehiclekm<br />

(CO2, NOx,<br />

VOC, SOx)<br />

Tonnes of<br />

pollutant per<br />

million train­km<br />

(CO2, NOx,<br />

VOC, SOx)<br />

Tonnes of<br />

pollutant per unit<br />

of traffic volume<br />

(million tonnekm)<br />

Road<br />

x<br />

Rail<br />

x<br />

Air<br />

Inland w.<br />

x<br />

x<br />

Sea<br />

Intermodal<br />

Network<br />

Main<br />

corrido<br />

rs<br />

By<br />

road<br />

section<br />

By rail<br />

section<br />

By<br />

waterw<br />

ay<br />

O­D<br />

Zone<br />

Country<br />

Country<br />

Country<br />

Y<br />

Y<br />

Y<br />

Y<br />

Traffic volumes and<br />

energy consumption<br />

data<br />

Traffic volumes,<br />

emission/pollution<br />

data<br />

Traffic volumes and<br />

emission/pollution<br />

data<br />

Percentage of<br />

network­km electrified<br />

Percentage of train­km<br />

electrically hauled<br />

Traffic volumes and<br />

emission/pollution<br />

data. (differentiation to<br />

inland waterways<br />

EU<br />

AC<br />

*** **<br />

**<br />

(country)<br />

* (by road<br />

section)<br />

*** **<br />

*** **<br />

**<br />

(country)<br />

* (by road<br />

section)<br />

74<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Mode Spatial scope Forecast Details to be provided<br />

Type of data<br />

required<br />

Data availability<br />

ref.<br />

1.3.3<br />

Definition<br />

Noise levels generated by<br />

road transport+corresponding<br />

marginal unit cost<br />

<strong>WP</strong>: 3,4,5,8<br />

Measurement<br />

unit<br />

Exposure to over<br />

a specified noise<br />

level in dB<br />

Road<br />

x<br />

Rail<br />

Air<br />

Inland w.<br />

Sea<br />

Intermodal<br />

Network<br />

By<br />

road<br />

section<br />

O­D<br />

Zone<br />

Y<br />

Traffic volumes, noise<br />

level data<br />

EU AC<br />

* (*)<br />

2.3.3<br />

Noise levels generated by rail<br />

transport +corresponding<br />

marginal unit cost<br />

<strong>WP</strong>: 3,4,5,8<br />

Exposure to over<br />

a specified noise<br />

level (L dn dB)<br />

x<br />

By rail<br />

section<br />

Y<br />

Traffic volumes, noise<br />

level data<br />

* (*)<br />

3.2.2<br />

Amount of investment made<br />

in the development and<br />

maintenance of air traffic<br />

control<br />

Million Euros<br />

x<br />

by<br />

airport<br />

Y<br />

National infrastructure<br />

programmes<br />

* *<br />

<strong>WP</strong>: 3,4,5<br />

Document2<br />

27 May 2004<br />

75


ANNEX B: INDICATOR COMPILATION<br />

TEMPLATES


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 1<br />

Ref. 1.1.4<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Breakdown of journeys by origin/destination (1.1.4a), proportion of long distance traffic using the TEN (1.1.4b)<br />

Journey: A movement of vehicle or vessel from a specified point of origin to a specified point of destination.<br />

Computation Method (Formula)<br />

1.1.4a<br />

T(i,j)<br />

1.1.4b<br />

V<br />

l<br />

( n )<br />

R ( n ) = or<br />

V ( n )<br />

R =<br />

å<br />

i , j Î L<br />

å<br />

i , j Î EU<br />

T ( i , j )<br />

T ( i , j )<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

T(i,j) – total number of journeys<br />

between origin i and destination j<br />

Trip generation model (OD matrix for<br />

passenger and freight movements)<br />

Method variable V2 i,j – NUTS2 zones NUTS2 zones database for Europe<br />

Method variable V3<br />

Method variable V4<br />

n ­ specific road link of TEN<br />

network<br />

R(n) – proportion of long­distance<br />

vehicles on the road link (n) of the<br />

TEN network<br />

G<strong>ETIS</strong> networks database for Europe<br />

Trip generation and assignment model<br />

(passenger and freight)<br />

Document2<br />

27 May 2004<br />

79


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V5<br />

V(n) – number of vehicles on the<br />

road link (n) of the TEN network<br />

Derived from traffic counts on the network<br />

(AADT) (for parts of the network)<br />

Trip generation and assignment model<br />

(passenger and freight)<br />

Method variable V6<br />

V l (n) ­ number of long distance<br />

vehicles on the road link (n) of the<br />

TEN network<br />

Trip generation and assignment model<br />

(passenger and freight)<br />

Method variable V7 l – distance category (long) The distance categories can be agreed<br />

(for example, long distance journey – exceeding<br />

300 km)<br />

Method variable V8 L ­ set of long distance OD relations NUTS2 zone combinations with a distance longer<br />

than the minimum threshold<br />

Method variable V9<br />

EU – total set of OD relations in the<br />

EU<br />

All NUTS2 zone combinations<br />

Remarks concerning method<br />

variable computation<br />

The same trip generation and assignment model can produce a complete variable set necessary to compute the indicator. This<br />

variable set can be split into variable set for passengers and variable set for freight.<br />

Model required to compute the<br />

method variables (listed above)<br />

Definition<br />

Trip generation and assignment model (passenger and freight)<br />

Model (component) analog<br />

For instance: VACLAV, SCENES, NEAC<br />

Model component C1 OD freight matrices (annual) annual number of freight vehicles that traveled between each OD<br />

pair, annual tons transported between each OD pair<br />

80<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C2 OD passenger matrices (annual) annual number of passengers and passenger vehicles (cars and<br />

buses) that traveled between each OD pair<br />

Model component C3 Distance matrices Distance for each OD pair per mode<br />

Model component C4 Networks and their attributes G<strong>ETIS</strong> networks<br />

Model component C5<br />

NUTS2 zones<br />

Model component C6 Traffic counts Vehicle counts on links<br />

Model component C7 Traffic assignment matrices Number of vehicles (long distance) per link per mode<br />

Model component C8 Survey data household survey, travel agencies survey, bus operators survey,<br />

road­side interviews<br />

Model component C9<br />

National and international trade, transport and social­economic<br />

data<br />

Remarks concerning the models More specific description of the reference models can be find in the material prepared by AJI­Europe<br />

(Tables_Final_Report_version_2000.xls)<br />

Document2<br />

27 May 2004<br />

81


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 2<br />

Ref. 1.5.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Road freight volumes on TEN by cargo type<br />

Road freight volumes: In Tonne km A movement of a road vehicle from a specified point of origin to a specified point of<br />

destination.<br />

Cargo type to be defined depending on data availability<br />

P(c) c = 1, 2, ……n<br />

P(c) = å T ij ( c ) * D ij ( c )<br />

i , j<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

P(c) ­ road freight volumes of cargo<br />

type c on TEN (in ton­km)<br />

OD matrix model of tons divided by<br />

cargo type + estimated distance matrix<br />

Method variable V2<br />

c – specific cargo type<br />

(c= 1, 2, …,n)<br />

Cargo type classification (to be defined)<br />

Method variable V3 i, j – origin i and destination j<br />

(NUTS2 zones)<br />

NUTS2 zones database for Europe<br />

82<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V4 Tij (c)<br />

Method variable V5 Dij (c)<br />

­ total tons of cargo type c<br />

transported between origin i and<br />

destination j<br />

­ total distance covered by<br />

cargo type c transported between<br />

origin i and destination j<br />

OD matrix model of tons per cargo<br />

type<br />

estimated distance matrix<br />

Remarks concerning method<br />

variable computation<br />

Definition<br />

Model (component) analog<br />

Model required to compute the<br />

method variables (listed above)<br />

Freight OD matrix model<br />

For instance SCENES, NEAC<br />

Model component C1 OD freight matrix Annual total freight (number of tones) transported between OD<br />

(domestic, international)<br />

Model component C2<br />

Estimated break­down of freight into cargo types generated by<br />

each zone<br />

Estimates obtained from trade and transport statistics (domestic<br />

and international), traffic survey<br />

Model component C3 Estimated break­down of freight into cargo types Estimates obtained from trade and transport statistics (domestic<br />

and international), traffic survey<br />

Model component C4<br />

Model component C5<br />

cargo type­based OD freight matrices<br />

NUTS2 zones<br />

Document2<br />

27 May 2004<br />

83


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C6 Distance matrices TEN­based road distances between NUTS2 zones and internal<br />

distances per zone<br />

Model component C7 Traffic survey data Traffic counts by freight vehicle category on TEN links<br />

Model component C8<br />

National and international trade and transport data<br />

Remarks concerning the models More specific description of the reference models can be find in the material prepared by AJI­Europe<br />

(Tables_Final_Report_version_2000.xls)<br />

84<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 3<br />

Ref. 2.1.2<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Traffic volumes on the Trans­European rail network, by type (passenger 2.1.2a / freight 2.1.2b), including non fulfilled demand<br />

Rail passenger kilometer: Unit of measure representing the transport of one rail passenger by rail over a distance of one kilometer.<br />

Tonne­km by rail: Unit of measure of goods transport which represents the transport of one tonne of goods by rail over a distance of<br />

one kilometer.<br />

Computation Method (Formula)<br />

(2.1.2a)<br />

(2.1.2b)<br />

P(m)<br />

P(f)<br />

P(m) = å N<br />

i , j<br />

P(f) = å T<br />

i , j<br />

ij<br />

ij<br />

* D ij ( m )<br />

* D ij ( f )<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Document2<br />

27 May 2004<br />

85


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V1<br />

P(m) ­ annual passenger traffic<br />

volumes (passenger­km) on the<br />

Trans­European rail network<br />

Trip generation and modal split model<br />

(aggregated figure of passenger­km on<br />

TEN­rail)<br />

Method variable V2 P(f) ­ annual freight traffic<br />

volumes (ton­km) on the Trans­<br />

European rail network<br />

Trip generation and modal split model<br />

(aggregated figure of tkm on TEN­rail)<br />

Method variable V3 m, f – traffic type (passenger/freight) Agreed traffic type categories (passenger/freight)<br />

Method variable V4<br />

i,j – NUTS3 zones (for passenger)<br />

i,j – NUTS2 zones (for freight)<br />

NUTS3 database for Europe<br />

NUTS2 database for Europe<br />

Method variable V5<br />

N ij ­ annual number of passengers<br />

transported on Trans­European rail<br />

between origin zone i and destination<br />

zone j<br />

Trip generation and modal split model<br />

(OD matrices on NUTS3 level)<br />

Method variable V6 Dij ( m )<br />

­ total average distance<br />

between zone i and zone j on TEN<br />

rail (passenger)<br />

Calculation of average distances between NUTS3<br />

zones reachable on TEN­rail (G<strong>ETIS</strong>­rail)<br />

Method variable V7<br />

T ij ­ annual freight (tons)<br />

transported between zone i and zone<br />

j on TEN­rail<br />

OD matrices on NUTS2 level<br />

86<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V8 Dij ( f )<br />

­ total average distance<br />

between zone i and zone j on TEN<br />

rail (freight)<br />

Calculation of average distances between NUTS2<br />

zones reachable on TEN­rail (G<strong>ETIS</strong>­rail)<br />

Remarks concerning method<br />

variable computation<br />

Special attention to the following aspects: non­fulfilled demand; occupancy in relation to capacity and in line with this empty train<br />

or wagons.<br />

Model required to compute the<br />

method variables (listed above)<br />

Definition<br />

Trip generation and modal split model (passenger)<br />

OD matrix model<br />

Model (component) analog<br />

For instance VACLAV (passenger)<br />

For instance NEAC, SCENES (freight)<br />

Model component C1 OD passenger matrix (annual) Number of passengers carried or passenger trips made between<br />

(and within) NUTS3 zones on TEN­rail<br />

Model component C2 OD freight matrix (annual) Freight carried between OD zones on TEN­rail<br />

Model component C3 Distance matrices Average Distances between OD zones calculated for TEN­rail<br />

(passenger/freight)<br />

Model component C4 Rail passenger services matrices Times schedules, routes, travel times of passenger trains serving<br />

TEN­rail obtained from operators<br />

Model component C5<br />

OD matrices showing LOS for passenger transportation by rail<br />

(total travel time by rail, number of transfers, waiting times etc)<br />

in relation to other modes (road)<br />

Rail services traffic allocation over TEN network;<br />

Generalised costs for rail and other modes (road) per OD pair<br />

Document2<br />

27 May 2004<br />

87


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C6 OD matrices showing transport chains for freight Transport chain information for freight flows between NUTS2<br />

zones (including transshipment points­zones)<br />

Model component C7 rail networks and their attributes G<strong>ETIS</strong> networks<br />

Model component C8<br />

Model component C9<br />

Ticketing information, trade and transport statistics<br />

NUTS2/NUTS3 zones<br />

Remarks concerning the models<br />

88<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 4<br />

Ref. 2.5.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Freight volumes on TEN by train type<br />

Train type: Goods train: Train for the carriage of goods composed of one or more wagons and, possibly, vans moving either empty<br />

or under load.<br />

F(t) = P ( t , r , c ) / L ( t , r , c )<br />

( 2.5.1a)<br />

P(t, c, r)<br />

= å T ij ( t , c , r ) * D ij ( t , c , r )<br />

i , j<br />

(2.5.1b)<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

F(t) – Freight train­km traveled<br />

annually by train type t on TEN­rail<br />

Freight train service schedules and routes<br />

database; G<strong>ETIS</strong>­rail network<br />

Method variable V2 t – train type Agreed train types: unitized, dedicated bulk,<br />

mixed bulk<br />

Method variable V3 i,j – NUTS2 zones NUTS2 zones database for Europe<br />

Method variable V4 L(t,r,c) Loading per train in (tonnes per train)<br />

Document2<br />

27 May 2004<br />

89


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V5<br />

Method variable V6<br />

Dij (t) ­ Average distance of trains<br />

of train type t between zone I and<br />

zone j on TEN­rail<br />

P(t,c,r) – freight volumes on TEN by<br />

train type t, cargo type c and risk<br />

level r<br />

Freight train served routes database; G<strong>ETIS</strong>­rail<br />

network<br />

OD matrices model on NUTS2 level<br />

Method variable V7 c – cargo type Agreed cargo types: liquid, dry, bulk, general<br />

cargo, (container possibly separate variable<br />

separate )<br />

Method variable V8 r – risk level Agreed risk levels : hazardous, non­hazardous<br />

Method variable V9 T ij ( t , c , r ) ­ total annual tons of<br />

cargo type c with risk level r<br />

transported by train type t on TENrail<br />

between zone I and zone y.<br />

Method variable V10<br />

D ij ( t , c , r ) ­ average distance<br />

between zone I and zone j on TENrail<br />

accessible for trains of train type<br />

t carrying cargo of cargo type c of<br />

risk level r.<br />

Freight train routes database; G<strong>ETIS</strong> network (rail<br />

+ terminals)<br />

OD matrix model on NUTS2 level<br />

Remarks concerning method<br />

variable computation<br />

90<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Definition<br />

Model (component) analog<br />

Model required to compute the<br />

method variables (listed above)<br />

OD matrix model<br />

For instance NEAC or SCENES<br />

Model component C1 OD freight matrix (annual) Freight carried between OD zones on TEN­rail<br />

Model component C2 Distance matrices Average Distances between OD zones calculated for TEN­rail<br />

per train type (freight).<br />

[P.S. It is possible that different train types can be directed to take different<br />

routes (i.d. resulting in different distances between zone i and zone j)].<br />

Average distances between OD zones reachable by other modes<br />

(competing)<br />

Model component C3 Rail freight train services matrices Times schedules, routes, travel times of freight trains on TENrail<br />

obtained from operators<br />

Model component C4 Modal split matrices Generalised costs for rail and other modes per OD pair<br />

Model component C5 OD matrices showing transport chains for freight Transport chain information for freight flows between NUTS2<br />

zones (including transshipment points­zones)<br />

Model component C6<br />

rail networks and their attributes, including the information<br />

whether the rail link is allowed for transportation of hazardeous<br />

goods (i.e, nuclear waste)<br />

G<strong>ETIS</strong> networks (rail)<br />

Document2<br />

27 May 2004<br />

91


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C7 Terminals networks G<strong>ETIS</strong> networks (terminals)<br />

Model component C8<br />

Model component C9<br />

trade and transport statistics<br />

NUTS2 zones<br />

Remarks concerning the models<br />

92<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 5<br />

Ref. 4.1.2<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Freight volumes on the inland waterway network<br />

Freight volume.­ Unit of measure representing the movement of one tonne available in an IWT freight vessel when performing the services for which it is<br />

primarily intended over one kilometer.<br />

Computation Method (Formula)<br />

P(f)<br />

=<br />

å<br />

i , j<br />

T ij * D ij<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

P(f) – annual freight volumes (ton­km) on the<br />

Trans­European inland waterways network<br />

aggregated figure of tkm on TEN­inland<br />

waterways as computed in the method<br />

Method variable V2 i,j – NUTS2 zones having TEN­iww NUTS2 zones database for Europe<br />

Method variable V3<br />

T ij<br />

­ annual freight (tons) transported<br />

between zone i and zone j on TEN­iww<br />

OD matrices and assignment model on NUTS2<br />

level<br />

Method variable V4<br />

D ij ­ total average distance between zone i<br />

and zone j on TEN­ iww<br />

G<strong>ETIS</strong> networks (iww)<br />

Document2<br />

27 May 2004<br />

93


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Remarks concerning method variable<br />

computation<br />

Model required to compute the method<br />

variables (listed above)<br />

Definition<br />

OD matrix and assignment model (freight)<br />

Model (component) analog<br />

for instance, SCENES, NEAC, VACLAV<br />

Model component C1 OD freight matrices (annual) Freight carried between OD zones on TEN­iww<br />

Model component C2 Distance matrices Average distances between OD zones reachable by TEN­iww and other<br />

competing modes<br />

Model component C3 Speed matrices Average speeds per competing mode between relevant NUTS2 zones<br />

Model component C4 Generalised costs matrices Generalized costs per competing mode for each OD<br />

Model component C5 OD matrices showing transport chains for freight within Europe Transport chain information for freight flows between all NUTS2 zones pairs<br />

Model component C6 iww networks G<strong>ETIS</strong> networks<br />

Model component C7<br />

Model component C8<br />

NUTS2 zones<br />

Trade and transport statistics<br />

Remarks concerning the models<br />

94<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 6<br />

Ref. 5.1.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation<br />

(Formula)<br />

Method<br />

Port throughput (passengers, freight)<br />

Port throughput (passengers, freight): Amount of the flow of traffic through a port (passengers, gross weight of goods: this includes the<br />

tonnage of goods carried, including packaging but excluding the tare weight of transport units. ) at a given time<br />

(5.1.1a ) N e = å<br />

Ne ( p )<br />

p<br />

N d = å<br />

p<br />

Nd ( p )<br />

(5.1.1b)<br />

WL<br />

WU<br />

=<br />

å<br />

p<br />

= å<br />

p<br />

W L ( p )<br />

W U ( p )<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

N e ­ annual number of passengers<br />

embarked at ports<br />

Aggregated port statistics<br />

Document2<br />

27 May 2004<br />

95


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V2<br />

N d – annual number of passengers<br />

disembarked at ports<br />

Aggregated port statistics<br />

Method variable V3 p – specific port List of TEN­ports<br />

Method variable V4<br />

Method variable V5<br />

N e (p)– annual number of passengers<br />

embarked at port p<br />

N d (p)– annual number of passengers<br />

disembarked at port p<br />

Port statistics<br />

Port statistics<br />

Method variable V6<br />

Method variable V7<br />

W L – gross weight of goods loaded<br />

at ports including packaging but<br />

excluding the tare weight of transport<br />

units (annual)<br />

W U – gross weight of goods<br />

unloaded including packaging but<br />

excluding the tare weight of transport<br />

units (annual)<br />

Aggregated port statistics<br />

Aggregated port statistics<br />

Method variable V8<br />

W L (p) – gross weight of goods<br />

loaded at port p including packaging<br />

but excluding the tare weight of<br />

transport units (annual)<br />

Derived from port statistics<br />

96<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V9<br />

W U (p) ­ gross weight of goods<br />

unloaded at port p including<br />

packaging but excluding the tare<br />

weight of transport units (annual)<br />

Derived from port statistics<br />

Remarks concerning<br />

method variable<br />

computation<br />

Document2<br />

27 May 2004<br />

97


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 7<br />

Ref. 4.1.3<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

I/C factor of bridges and locks (intensity versus capacity) (waiting time)<br />

I/C factor: relationship between intensity (actual traffic of ships) and capacity (theoretical number of ships per week)<br />

S(k)=I(k)/C(k) k= 1,…,n<br />

W L (k)<br />

W b (k)<br />

k=1,…,n<br />

k=1,…,n<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

Method variable V2<br />

Method variable V3<br />

S(k)­Service level of bridge or lock k<br />

I(k) Intensity in terms of number of<br />

ships per week at bridge or lock k<br />

C(k) Capacity in terms of theoretical<br />

number of ships per week of bridge<br />

or lock k<br />

Assignment model, OD data model<br />

Inland navigation capacity model<br />

Method variable V4 W L (k) waiting time at lock k Inland navigation waiting time model<br />

Method variable V5 W b (k) waiting time at bridge k Inland navigation waiting time model<br />

98<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V6<br />

Method variable V7<br />

Method variable V8<br />

Method variable V9<br />

Remarks concerning method<br />

variable computation<br />

Model required to compute the<br />

method variables (listed above)<br />

Definition<br />

Inland navigation waiting time model<br />

Assignment model, OD model<br />

Inland navigation capacity model<br />

for instance, SCENES, NEAC<br />

Model (component) analog<br />

Model component C1 OD freight matrix (annual) by vessel type Freight carried between OD zones on TEN­iww<br />

Model component C2<br />

OD matrices showing transport chains for freight between<br />

Europe<br />

Transport chain information for freight flows between NUTS2<br />

zones pairs in the catchment area<br />

Model component C3 Commercial speed matrices Average commercial speeds per competing mode between<br />

relevant NUTS2 zones<br />

Model component C4 iww networks (including ports, locks and bridges) G<strong>ETIS</strong> networks with relevant attributes<br />

Document2<br />

27 May 2004<br />

99


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C5<br />

NUTS2 zones<br />

Model component C6 Inventory of locks and bridges including technical<br />

specifications: depth and width at entrance and exit, entrance<br />

time, operation time, number of locks, length, opening times.<br />

Model component C7<br />

Model component C8<br />

Model component C9<br />

Model component C10<br />

Traffic survey data<br />

Port statistics, trade and transport statistics<br />

I/C factor by lock and bridge<br />

OD showing empty vessel movements<br />

Remarks concerning the models<br />

100<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 8<br />

Ref. 6.1.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Traffic volumes served at the terminal<br />

Traffic volume: Number of passengers (embarked/disembarked) and freight (tonne) (loaded/unloaded)<br />

(6.1.1a ) N = å ( N<br />

e<br />

( p ) + N<br />

d<br />

( p ) - N<br />

t<br />

( p ))<br />

p<br />

Ne = å<br />

p<br />

N d = å<br />

p<br />

N t<br />

= å<br />

p<br />

Ne<br />

( p )<br />

N t<br />

Nd ( p )<br />

( p )<br />

(6.1.1b) W(p) = W ( p ) + W ( p ) W ( p )<br />

L U<br />

-<br />

W = W<br />

L<br />

+ W<br />

U<br />

- W<br />

t<br />

WL = å W L ( p )<br />

p<br />

WU = å W U ( p )<br />

p<br />

W t<br />

= å W<br />

t<br />

( p )<br />

p<br />

t<br />

Document2<br />

27 May 2004<br />

101


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1<br />

Method variable V2<br />

Method variable V3<br />

Method variable V4<br />

N ­ annual number of passengers at<br />

terminals<br />

N d – annual number of passengers<br />

disembarked at terminals<br />

N e ­ annual number of passengers<br />

embarked at terminals<br />

N t – annual number of passengers<br />

transiting at terminals<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Method variable V5 p – specific terminal List of TEN­ terminal<br />

Method variable V6<br />

Method variable V7<br />

Method variable V8<br />

Method variable V9<br />

N (p) ­ annual number of passengers<br />

at terminal p<br />

N d (p)– annual number of passengers<br />

disembarked at terminal p<br />

N e (p) ­ annual number of<br />

passengers embarked at terminal p<br />

N t (p) – annual number of passengers<br />

transiting at terminal p<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

102<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V10<br />

Method variable V11<br />

W – gross weight of goods at<br />

terminal including packaging but<br />

excluding the tare weight of<br />

transport units (annual)<br />

W U ­ gross weight of goods<br />

unloaded including packaging but<br />

excluding the tare weight of<br />

transport units (annual)<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Method variable V12<br />

Method variable V13<br />

W L – gross weight of goods loaded<br />

at terminal including packaging but<br />

excluding the tare weight of<br />

transport units (annual)<br />

W t ­ gross weight of goods<br />

transiting including packaging but<br />

excluding the tare weight of<br />

transport units (annual)<br />

Aggregated terminal statistics<br />

Aggregated terminal statistics<br />

Method variable V14<br />

Method variable V15<br />

W(p) – gross weight of goods at<br />

port p including packaging but<br />

excluding the tare weight of<br />

transport units (annual)<br />

W U (p) ­ gross weight of goods<br />

unloaded at port p including<br />

packaging but excluding the tare<br />

Derived from terminal statistics<br />

Derived from terminal statistics<br />

Document2<br />

27 May 2004<br />

103


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

weight of transport units (annual)<br />

Method variable V16<br />

Method variable V17<br />

W L (p) – gross weight of goods<br />

loaded at port p including packaging<br />

but excluding the tare weight of<br />

transport units (annual)<br />

W t (p) ­ gross weight of goods<br />

transiting at port p including<br />

packaging but excluding the tare<br />

weight of transport units (annual)<br />

Derived from terminal statistics<br />

Derived from terminal statistics<br />

Remarks concerning method<br />

variable computation<br />

Definition<br />

Model (component) analog<br />

Model required to compute the<br />

method variables (listed above)<br />

The indicator can be directly calculated from terminal statistics<br />

depending on availability. Models might be used where needed.<br />

Model component C1 ­<br />

Model component C2 ­<br />

Model component C3<br />

104<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Model component C4<br />

Model component C5<br />

Model component C6<br />

Model component C7<br />

Model component C8<br />

Model component C9<br />

Remarks concerning the models<br />

Document2<br />

27 May 2004<br />

105


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 9<br />

Ref. 1.3.1<br />

Definition<br />

Estimated / measured energy consumption on roads<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

toe per vehicle­km for road transport<br />

Consumption calculation examples from COMMUTE model, Commute software <strong>report</strong>, page 8 ­14<br />

PASSENGER transport ( see specific formulas in <strong>Annex</strong> I):<br />

E total = E hot + E cold<br />

The consumption equations are similar to emissions equations, E can be understood as consumption cp. ref<br />

template 1.3.2. The factors can be derived from HBEFA Hand Book Emission factors for road Transport<br />

Hot running consumption<br />

Ehot= c*m<br />

m<br />

= n × l<br />

The hot consumption factor e has to be corrected because of the emission / consumption relevant influences and is calculated from<br />

an equation having the form:<br />

x<br />

å<br />

e c = ( ps × a s ) × ( pl × a l ) × e<br />

i, v , k i, v , s i, v , k , s<br />

i, v , u i, v , k , u i, v , k<br />

s = 1 u = 1<br />

z<br />

å<br />

In the case of passenger cars, where load is considered not to affect emissions and consumption, this equation is further simplified:<br />

106<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

x<br />

å<br />

e c = ( ps × a s ) × e<br />

i, v , k i, v , s i, v , k , s i, v , k<br />

s = 1<br />

For the calculation of the corrected hot consumption of the vehicle fleet, the equation for the basic hot emissions has to be<br />

converted to:<br />

i = categories<br />

æ<br />

EC = å ç n × l × p × ec<br />

è<br />

w<br />

å ( )<br />

hot , k i i i , v i , v , k<br />

i = 1<br />

v = 1<br />

ö<br />

÷<br />

ø<br />

Effectively, the product n i .l i in the above equation corresponds to “vehicle­kilometre” (number of vehicles by travel distance) for<br />

each vehicle category i.<br />

For gasoline (with and without catalyst) and diesel cars (without catalyst), the formula for excess consumption) is in the following<br />

form:<br />

excess consumption= w×[f(V)+g(T)­1]×h(d)<br />

d = d / d c<br />

engine start temperature (in °C) for starts at an intermediate temperature<br />

It is assumed that the excess emission of diesel cars with oxidation catalyst is proportional to the excess emission of diesel cars<br />

Document2<br />

27 May 2004<br />

107


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

without catalyst, according to the following equation:<br />

excess consumption for diesel cars with catalyst<br />

excess consumption for diesel cars without catalyst<br />

= a<br />

Cold correction<br />

E c =<br />

å<br />

i<br />

( i )<br />

æ cm s, v<br />

ç tf i × × w i ×<br />

è 100<br />

å å å<br />

j<br />

k<br />

m<br />

é p × p × p<br />

ê<br />

ë 10 6 × d<br />

j k m<br />

m<br />

d m<br />

× ( f ( V j ) + g ( T k ) - 1 ) × h<br />

æ ö ù ö<br />

ç ÷ ú ÷<br />

è d c (V j ) ø û ø<br />

the following parameters are model internal data:<br />

cm (s, vi) percentage of mileage recorded under cold start or intermediate temperature<br />

conditions for season s and overall speed vi of vehicle type i (%)<br />

wi reference excess emission for vehicle type i (g)<br />

j speed class with a cold engine<br />

k class of start­up engine temperature<br />

m trip length class<br />

pj percentage of the trips travelled at speed j with a cold engine, for the overall<br />

average speed considered (%)<br />

pk percentage of the trips travelled with a start­up engine temperature Tk (%)<br />

pm percentage of trips started with a cold engine and distance dm, for speed Vj with a cold engine (%)<br />

dm average distance of the trips under cold start conditions of class m (km)<br />

Vj average speed with a cold engine corresponding to class j (km/h)<br />

Tk average start­up temperature of class k (°C)<br />

108<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

f, g, h and dc are provided as coefficients relevant for calculating cold emissions<br />

Examples from e.g. UNITE :<br />

corresponding unit costs<br />

Examples from RECORDIT:<br />

FREIGHT transport including marginal costs<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of<br />

the model<br />

Method variable V1<br />

Method variable V2<br />

Method variable V3<br />

Method variable V4<br />

Method variable V5<br />

Method variable V6<br />

Method variable V7<br />

E total= Total energy consumption<br />

E hot= Hot running consumption<br />

e= the hot consumption factor (in g/km)<br />

m= the activity, in distance travelled per time unit<br />

(usually in km/year)<br />

n= is the number of vehicles in each category<br />

l= is the average distance travelled by the average<br />

vehicle of each category over the time unit (in km/year)<br />

ec= the corrected hot consumption factor for the vehicle fleet, in<br />

consideration of the influence of gradient and in the case of<br />

catalyst vehicles of mileage<br />

Document2<br />

27 May 2004<br />

109


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V8<br />

Method variable V9<br />

Method variable V10<br />

Method variable V11<br />

Method variable V12<br />

Method variable V13<br />

Method variable V14<br />

Method variable V15<br />

Method variable V16<br />

Method variable V17<br />

Method variable V18<br />

Method variable V19<br />

Method variable V20<br />

Method variable V21<br />

Method variable V22<br />

Method variable V23<br />

Method variable V24<br />

Method variable V25<br />

i=the number of vehicle categories<br />

v=the road type, practically signifying the mean speed<br />

k=the number of pollutants<br />

s= the gradient class<br />

x= the number of gradient classes<br />

u=the load class<br />

ps= the percentage of annual distance travelled at gradient class<br />

αs= the correction factor for gradient class s<br />

pl=the percentage of annual distance travelled at load<br />

class u<br />

αl= the correction factor for load class u<br />

EC= the corrected hot consumption of the vehicle fleet<br />

n= the number of vehicles<br />

l= the average annual distance travelled<br />

w= number of road types (urban, non­urban, motorway)<br />

p= the percentage of annual distance travelled<br />

excess consumption for a trip, expressed in g<br />

ω= reference excess consumption (at 20 °C and 20 km/h)<br />

V= means speed during the cold period (in km/h)<br />

110<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V26<br />

Method variable V27<br />

Method variable V28<br />

Method variable V29<br />

Method variable V30<br />

Method variable V31<br />

Method variable V32<br />

Method variable V33<br />

Method variable V34<br />

Method variable V35<br />

Method variable V36<br />

T= ambient temperature (in °C) for cold starts<br />

δ= undimensioned distance<br />

d= travelled distance<br />

d c = cold distance<br />

E cold= Cold correction consumption<br />

Ec= traffic excess emissions with a cold engine over 1 km<br />

for a given pollutant (in g)<br />

tfi= traffic flow for the studied vehicle type i (in km · veh)<br />

vi= traffic overall average speed for the studied vehicle type<br />

i (km/h)<br />

s,= season (winter, summer, middle)<br />

Road type*<br />

Road gradient distribution*<br />

Method variable V37 Temperature* Model default from COMMUTE can be used if case<br />

specific data not available<br />

Method variable V38 Traffic amount, m= total annual mileage of gasoline<br />

powered vehicles of category j*<br />

Method variable V39 Average trip length* Updated model default from COMMUTE<br />

Method variable V40 Average speed* Updated model default from COMMUTE can be used<br />

if case specific data not available<br />

Document2<br />

27 May 2004<br />

111


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V41 Average load* Updated model default from COMMUTE can be used<br />

if case specific data not available<br />

Method variable V42 Average occupancy* Updated model default from COMMUTE<br />

Method variable V43 Fleet composition* Updated model default from COMMUTE<br />

Method variable V44<br />

Type of loading unit (Swap Body Class A or 20 feet<br />

container)*<br />

Method variable V45 Emission/ consumption factors COMMUTE/ MEET<br />

Method variable V46 Emission Standard (Euro I, II, III, IV) Old vehicles?<br />

Method variable V47 Loading factor (Loaded vkm/total vkm)* <strong>WP</strong> 3<br />

Method variable V48<br />

Chot= Hot running consumption<br />

112<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 10<br />

Ref. 2.3.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Estimated/measured energy consumption of railways<br />

Energy Consumption of rail within a link or area divided between train characteristics of freight and passenger transport<br />

COMMUTE model for passenger transport<br />

RECORDIT model for freight transport<br />

COMMUTE:<br />

where:<br />

v<br />

'<br />

E ( N + 1 ) 2<br />

max<br />

D h<br />

@<br />

+ . . g<br />

L<br />

B + B v +<br />

ave B v 2 +<br />

0 1 2 ave<br />

2<br />

L<br />

g<br />

gravitational constant<br />

Variable definition<br />

Variable computation method<br />

Method variable V1 E’ the energy consumption in<br />

KJ/tonne­km*<br />

Method variable V2 L trip length (km)*<br />

Method variable V3 Dh change in height (gradient) (m)*<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Energy consumption in KJ/tonne­km<br />

Method variable V4 N number of intermediate stops* default values from COMMUTE<br />

Method variable V5 V ave average speed (m/s)* default values from COMMUTE<br />

Document2<br />

27 May 2004<br />

113


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V6 V max maximum speed (m/s) * default values from COMMUTE<br />

Method variable V7 B0 constant ­ equating to rolling<br />

resistance<br />

Method variable V8 B1 constant ­ equating to friction<br />

resistance<br />

Method variable V9 B2 constant ­ equating to cross<br />

Method variable V10<br />

sectional aerodynamic resistance<br />

Capacity (max. number of loading<br />

units)*<br />

default values from COMMUTE<br />

default values from COMMUTE<br />

default values from COMMUTE<br />

Method variable V11<br />

Method variable V12<br />

Method variable V13<br />

Method variable V14<br />

Method variable V15<br />

Method variable V16<br />

Loading factor (number of loading units<br />

carried/capacity)*<br />

Loading Unit Weights*<br />

Number of stops<br />

Number of wagons*<br />

Type of loading unit*<br />

Average segment speed*<br />

114<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 11<br />

Ref. 4.2.1<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Estimated/ measured energy consumption on waterways<br />

Energy Consumption of rail within a link or area divided between ship categories<br />

COMMUTE<br />

LINKFUELCONSUMPTION Link , FuelType =<br />

æ SHIPMOVEMENTS TypeShip , Sizeclass * FUELCONSUMPTION TypeShip , Sizeclass , FuelType * LINKLENGTH ö Link<br />

å ç<br />

OPERATINGSPEED<br />

TypeShip , Sizeclass è<br />

TypeShip , Sizeclass * 1,<br />

852<br />

÷<br />

ø<br />

( * *<br />

)<br />

PORTFUELCONSUMPTION Node , FuelType = å PORTCALLS TypeShip , Sizeclass FUELCONSUMPTIONPORT TypeShip , Sizeclass , FuelType PORTSTAYTIME Node , TypeShip , Sizeclass<br />

TypeShip , Sizeclass<br />

Fuel Consumption= C jk = K * Gross Tonnage + L [ton/day]<br />

Remark<br />

Inland waterways data poorly<br />

available<br />

Variable definition<br />

Variable computation method<br />

Method variable V1<br />

Cjk ­ fuel consumption at full power<br />

(t/day) as function of gross tonnage<br />

(GT)<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Energy consumption in KJ/tonne­km<br />

Method variable V2 j ­ fuel type Energy consumption in KJ/tonne­km<br />

Document2<br />

27 May 2004<br />

115


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V3<br />

k ­ ship class<br />

Method variable V4 K ­ variable from MEET table C)<br />

depending on ship type<br />

Method variable V5 L ­ variable from MEET table C)<br />

depending on ship type<br />

default values from COMMUTE<br />

116<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 12<br />

Ref. 1.3.2<br />

Definition<br />

Estimated/ measured road transport emissions + corresponding marginal cost<br />

<strong>ETIS</strong> Glossary<br />

Transport emissions are estimated for different vehicle composition on a link type divided by passenger and freight transport.<br />

Computation Method (Formula) Emission calculation examples from COMMUTE model, Commute software <strong>report</strong>, page 8 ­14<br />

PASSENGER transport ( see specific formulas in <strong>Annex</strong> I):<br />

E total = E hot + E cold + E evap<br />

(E evap only for VOC)<br />

The factors can be derived from HBEFA Hand Book Emission factors for road Transport<br />

Hot running emissions<br />

E<br />

hot<br />

= e × m<br />

m<br />

= n × l<br />

The hot emission/consumption factor e has to be corrected because of the emission / consumption relevant influences and is<br />

calculated from an equation having the form:<br />

x<br />

å<br />

e c = ( ps × a s ) × ( pl × a l ) × e<br />

i, v , k i, v , s i, v , k , s<br />

i, v , u i, v , k , u i, v , k<br />

s = 1 u = 1<br />

z<br />

å<br />

In the case of passenger cars, where load is considered not to affect emissions and consumption, this equation is further simplified:<br />

Document2<br />

27 May 2004<br />

117


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

x<br />

å<br />

e c = ( ps × a s ) × e<br />

i, v , k i, v , s i, v , k , s i, v , k<br />

s = 1<br />

For the calculation of the corrected hot emissions / consumption of the vehicle fleet, the equation for the basic hot emissions has to<br />

be converted to:<br />

i = categories<br />

æ<br />

EC = å ç n × l × p × ec<br />

è<br />

w<br />

å ( )<br />

hot , k i i i , v i , v , k<br />

i = 1<br />

v = 1<br />

ö<br />

÷<br />

ø<br />

Effectively, the product n i .l i in the above equation corresponds to “vehicle­kilometre” (number of vehicles by travel distance) for<br />

each vehicle category i.<br />

For gasoline (with and without catalyst) and diesel cars (without catalyst), the formula for excess emission (or excess consumption)<br />

is in the following form:<br />

excess emissions= w×[f(V)+g(T)­1]×h(d)<br />

engine start temperature (in °C) for starts at an intermediate temperature<br />

d = d / d c<br />

It is assumed that the excess emission of diesel cars with oxidation catalyst is proportional to the excess emission of diesel cars<br />

118<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

without catalyst, according to the following equation:<br />

excess emission for diesel cars with catalyst<br />

excess emission for diesel cars without catalyst<br />

= a<br />

Cold correction<br />

E c =<br />

å<br />

i<br />

( i )<br />

æ cm s, v<br />

ç tf i × × w i ×<br />

è 100<br />

å å å<br />

j<br />

k<br />

m<br />

é p × p × p<br />

ê<br />

ë 10 6 × d<br />

j k m<br />

m<br />

d m<br />

× ( f ( V j ) + g ( T k ) - 1 ) × h<br />

æ ö ù ö<br />

ç ÷ ú ÷<br />

è d c (V j ) ø û ø<br />

the following parameters are model internal data:<br />

cm (s, vi) percentage of mileage recorded under cold start or intermediate temperature<br />

conditions for season s and overall speed vi of vehicle type i (%)<br />

wi reference excess emission for vehicle type i (g)<br />

j speed class with a cold engine<br />

k class of start­up engine temperature<br />

m trip length class<br />

pj percentage of the trips travelled at speed j with a cold engine, for the overall<br />

average speed considered (%)<br />

pk percentage of the trips travelled with a start­up engine temperature Tk (%)<br />

pm percentage of trips started with a cold engine and distance dm, for speed Vj with a cold engine (%)<br />

dm average distance of the trips under cold start conditions of class m (km)<br />

Vj average speed with a cold engine corresponding to class j (km/h)<br />

Tk average start­up temperature of class k (°C)<br />

Document2<br />

27 May 2004<br />

119


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

f, g, h and dc are provided as coefficients relevant for calculating cold emissions<br />

Evaporative correction<br />

E evap, VOC, j = 365. a j (e d + S c + S fi ) + R<br />

and:<br />

x = v j / (365. l trip )<br />

S c = (l­q) (p . x . e s, hot + w . x . e s, warm )<br />

S fi = q . e fi . x<br />

R = m j (p . e r, hot + w . e r, warm )<br />

l trip<br />

average trip length (km)<br />

Examples from e.g. UNITE :<br />

corresponding unit costs<br />

Examples from RECORDIT:<br />

FREIGHT transport including marginal costs<br />

120<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of<br />

the model<br />

Method variable V1<br />

Method variable V2<br />

Method variable V3<br />

Method variable V4<br />

Method variable V5<br />

Method variable V6<br />

Method variable V7<br />

Method variable V8<br />

Method variable V9<br />

Method variable V10<br />

Method variable V11<br />

Method variable V12<br />

E total= Total emissions<br />

E hot= Hot running emissions<br />

e= the hot emission factor (in g/km)<br />

m= the activity, in distance travelled per time unit<br />

(usually in km/year)<br />

n= is the number of vehicles in each category<br />

l= is the average distance travelled by the average<br />

vehicle of each category over the time unit (in<br />

km/year)<br />

ec= the corrected hot emission / consumption factor for the<br />

vehicle fleet, in consideration of the influence of gradient<br />

and in the case of catalyst vehicles of mileage<br />

i=the number of vehicle categories<br />

v=the road type, practically signifying the mean<br />

speed<br />

k=the number of pollutants<br />

s= the gradient class<br />

x= the number of gradient classes<br />

Document2<br />

27 May 2004<br />

121


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V13<br />

Method variable V14<br />

Method variable V15<br />

Method variable V16<br />

Method variable V17<br />

Method variable V18<br />

Method variable V19<br />

Method variable V20<br />

Method variable V21<br />

Method variable V22<br />

Method variable V23<br />

Method variable V24<br />

Method variable V25<br />

Method variable V26<br />

Method variable V27<br />

Method variable V28<br />

u=the load class<br />

ps= the percentage of annual distance travelled at gradient<br />

class s<br />

αs= the correction factor for gradient class s<br />

pl=the percentage of annual distance travelled at<br />

load class u<br />

αl= the correction factor for load class u<br />

EC= the corrected hot emission/consumption of<br />

the vehicle fleet<br />

n= the number of vehicles<br />

l= the average annual distance travelled<br />

w= number of road types (urban, non­urban,<br />

motorway)<br />

p= the percentage of annual distance travelled<br />

excess emission for a trip, expressed in g<br />

ω= reference excess emission (at 20 °C and 20 km/h)<br />

V= means speed during the cold period (in km/h)<br />

T= ambient temperature (in °C) for cold starts<br />

δ= undimensioned distance<br />

d= travelled distance<br />

122<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V29<br />

d c = cold distance<br />

Method variable V30<br />

E cold= Cold correction emissions<br />

Method variable V31 Ec= traffic excess emissions with a cold engine over 1<br />

km for a given pollutant (in g)<br />

Method variable V32 tfi= traffic flow for the studied vehicle type i (in km ·<br />

veh)<br />

Method variable V33<br />

Method variable V34<br />

Method variable V35<br />

Method variable V36<br />

Method variable V37<br />

vi= traffic overall average speed for the studied<br />

vehicle type i (km/h)<br />

s,= season (winter, summer, middle)<br />

E evap= Evaporative correction emissions<br />

E evap, VOC, j = VOC emissions due to evaporative losses<br />

caused by vehicle category j<br />

a j = number of gasoline vehicles of category j<br />

Method variable V38 e d = mean emission factor for diurnal losses of<br />

gasoline powered vehicles equipped with metal tanks,<br />

depending on average monthly ambient temperature,<br />

Method variable V39<br />

Method variable V40<br />

Method variable V41<br />

temperature variation and fuel volatility (RVP)<br />

S c = average hot and warm soak emission factor of<br />

gasoline powered vehicles equipped with carburettor<br />

S fi = average hot and warm soak emission factor of<br />

gasoline powered vehicles equipped with fuel<br />

injection<br />

R= hot and warm running losses<br />

Document2<br />

27 May 2004<br />

123


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V42<br />

Method variable V43<br />

Method variable V44<br />

Method variable V45<br />

Method variable V47<br />

Method variable V48<br />

q= fraction of gasoline powered vehicles with fuel<br />

injection<br />

p= fraction of trips finished with hot engine<br />

(dependent on average monthly ambient temperature)<br />

x= mean number of trips of a vehicle per day, average<br />

over the year<br />

w= fraction of trips finished with cold or warm engine<br />

(shorter trips) or with catalyst below its light­off<br />

temperature<br />

Road type*<br />

Road gradient distribution*<br />

Method variable V49 Temperature* Model default from COMMUTE can be used if case specific<br />

data not available<br />

Method variable V50<br />

Traffic amount, m= total annual mileage of gasoline<br />

powered vehicles of category j*<br />

Method variable V51 Average trip length* Updated model default from COMMUTE<br />

Method variable V52 Average speed* Updated model default from COMMUTE can be used if case<br />

specific data not available<br />

Method variable V53 Average load* Updated model default from COMMUTE can be used if case<br />

specific data not available<br />

Method variable V54 Average occupancy* Updated model default from COMMUTE<br />

124<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V55 Fleet composition* Updated model default from COMMUTE<br />

Method variable V56<br />

Type of loading unit (Swap Body Class A or 20 feet<br />

container)*<br />

Method variable V57 Emission/ consumption factors COMMUTE/ MEET<br />

Method variable V58 Emission Standard (Euro I, II, III, IV) Old vehicles?<br />

Method variable V59 Loading factor (Loaded vkm/total vkm)* <strong>WP</strong> 3<br />

Method variable V60<br />

Chot= Hot running consumption<br />

Document2<br />

27 May 2004<br />

125


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 13<br />

Ref. 2.3.2<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Estimated/measured rail transport emissions + corresponding unit cost<br />

Emissions of rail transport within a link or area<br />

COMMUTE model for passenger transport<br />

The emissions are calculated from energy consumption according to MEET tables. (COMMUTE software <strong>report</strong> p. 19­20 )<br />

Energy consumption equation is shown in Indicator Compilation Table 2.3.1<br />

E= emission factor * Energy Consumption<br />

The sulphur dioxide emissions from diesel combustion are heavily dependent on the sulphur content of the fuel. The fuel specific<br />

sulphur emission (FSSO2) may be derived from the following expression:<br />

FSSO2 = 20 x %S<br />

where %S is the mass weight percent of sulphur in fuel. The value in the above table indicates a range of fuel sulphur content from<br />

0.05 to 0.5%.<br />

RECORDIT model for freight transport<br />

Variable definition<br />

Variable computation method<br />

Method variable V1<br />

FSSO2= sulphur emissions<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

126<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V2<br />

%S = the mass weight percent of<br />

sulphur in fuel<br />

Method variable V3 Electrical power generation –<br />

European average/Country specific<br />

Updated tables from MEET/COMMUTE<br />

Method variable V4 Diesel railway emissions ­ Updated MEET/COMMUTE database<br />

Method variable V5 Degree of electrification Updated MEET/COMMUTE database<br />

Method variable V6 Emission factors MEET/COMMUTE<br />

Method variable V7 Consumption factors MEET/COMMUTE<br />

Document2<br />

27 May 2004<br />

127


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 14<br />

Ref. 4.2.2<br />

Definition<br />

Estimated/measured Inland waterway transport emissions<br />

<strong>ETIS</strong> Glossary<br />

Estimated waterway transport emissions divided between ship categories, links or ports<br />

Computation Method The COMMUTE equations are for sea. The port emissions for inland waterways could be assessed similarly.<br />

(Formula)<br />

Emissions, except SO 2<br />

LINKEMISSIONS<br />

Link , Emissiontype<br />

=<br />

æ SHIPMOVEMENTS , * FUELCONSUMPTION , * EMISSIONFACTOR , , * LINKLENGTH<br />

å ç<br />

è<br />

OPERATINGSPEED TypeShip , Sizeclass * 1,<br />

852<br />

TypeShip , Sizeclass<br />

TypeShip Sizeclass TypeShip Sizeclass TypeShip Sizeclass Emissiontype Link<br />

ö<br />

÷<br />

ø<br />

( * * *<br />

)<br />

PORTEMISSIONS = å PORTCALLS FUELCONSUMPTIONPORT EMISSIONFACTOR PORTSTAYTIME<br />

Emissions SO2<br />

LINKEMISSIONS<br />

Node , Emissiontype TypeShip , Sizeclass TypeShip , Sizeclass TypeShip , Sizeclass , MarineDiesel Node , TypeShip , Sizeclass<br />

Link , SO 2<br />

=<br />

å<br />

TypeShip , Sizeclass<br />

TypeShip , Sizeclass<br />

å ( )<br />

æ SHIPMOVEMENTS TypeShip , Sizeclass<br />

ç<br />

* FUELCONSUMPTION TypeShip , Sizeclass , Fueltype * SULPHURCONTENT TypeShip , Sizeclass , Fueltype * 20 * LINKLENGTH<br />

ç<br />

FuelType<br />

OPERATINGSPEED TypeShip , Sizeclass * 1,<br />

852<br />

ç<br />

è<br />

( * * * *<br />

)<br />

PORTEMISSIONS<br />

2<br />

= å PORTCALLS FUELCONSUMPTIONPORT SULPHURCONTENT 20 PORTSTAYTIME<br />

Node , SO TypeShip , Sizeclass TypeShip , Sizeclass , FuelType TypeShip , Sizeclass , FuelType Node , TypeShip , Sizeclass<br />

TypeShip , Sizeclass<br />

Link<br />

ö<br />

÷<br />

÷<br />

÷<br />

ø<br />

remark<br />

+ corresponding marginal cost from reference study, for example UNITE<br />

Arial emissions could be considered<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or<br />

agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1 LINK EMISSIONS ­ emissions across link Energy consumption in<br />

tonne­km<br />

128<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Method variable V2 PORT EMISSIONS ­ emissions at ports Energy consumption in<br />

tonne­km<br />

Method variable V3<br />

Ship movements ­ ships moving between a link of ship type and class<br />

*<br />

Method variable V4 Type ship­ specific type of ship default values from COMMUTE<br />

Method variable V5 Size class – specific size of ship default values from COMMUTE<br />

Method variable V6 Fuel type ­ fuel type used default values from COMMUTE<br />

Method variable V7<br />

Method variable V8<br />

Fuel consumption of a ship– regarding type of ship, size class and fuel<br />

type<br />

Capacity (max. number of loading units)*<br />

Method variable V9 Loading factor (number of loading units<br />

carried/capacity) *<br />

Method variable V10 port calls – number of port calls *<br />

default values from COMMUTE<br />

Method variable V11 Emission factor default values from<br />

COMMUTE/MEET<br />

Method variable V12 Sulphur content default values from<br />

COMMUTE/MEET<br />

Remarks<br />

The most important environmental effects in waterways are accident risks. In ports the most important are noise, CO2, NOx, PM small,<br />

VOC.<br />

Document2<br />

27 May 2004<br />

129


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 15<br />

Ref. 1.3.3<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Noise levels generated by road transport + corresponding marginal unit<br />

Noise generate by a link in Watts/ Exposure to over a specific noise level in dB<br />

A general approach has been defined in COMMUTE. The default reference conditions are as follows;<br />

· average vehicle speed, v ­ 75 km/hr<br />

· percentage heavy vehicles, p ­ 0%<br />

· percentage uphill gradient, g ­ 0%<br />

The formula used to derive the combined average vehicle speed and percentage heavy vehicles correction is as follows;<br />

· correction (multiply by): (v+40+500/v)^ 3.3 x (1 + 5 p/v)/7604424<br />

The formula used to derive the percentage uphill gradient correction is as follows;<br />

· correction (multiply by): 10^ (0.03g)<br />

The formula to calculate average sound intensity in watts/m 2 at any given radial distance, r, and ignoring excess attenuation is as follows. For road traffic noise<br />

the standard reference radial distance, d, is 10 m from the side of the carriageway, which is 13.5 m from the centreline of vehicles moving along the road;<br />

· sound intensity = sound power per metre/(p x d)<br />

· sound pressure = square root (sound intensity x 415)<br />

· sound pressure level = 20 x log(sound pressure/0.00002)<br />

· link sound power= number of vehicles*corrected source sound power<br />

· corrected sound power= reference sound power per vehicle per hour*heavy vehicles correction factor* uphill gradient correction factor<br />

130<br />

Document2<br />

27 may 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

+ Corresponding unit cost from UNITE or other reference study?<br />

RECORDIT<br />

for freight transport<br />

Variable definition<br />

Variable computation method<br />

Method variable V1 v ­ vehicle speed *<br />

Method variable V2 p­ percentage of heavy vehicles *<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V3<br />

reference sound power per hour<br />

Method variable V4 number of vehicles per hour *<br />

Method variable V5 percentage of uphill gradient *<br />

Method variable V6 Noise factors for reference traffic conditions Defaults from COMMUTE<br />

Method variable V8 sound intensity Defaults from COMMUTE<br />

Method variable V9 sound pressure Defaults from COMMUTE<br />

Method variable V10 link sound power Defaults from COMMUTE<br />

Document2<br />

27 May 2004<br />

131


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Indicator Compilation Template No. 16<br />

Ref. 2.3.3<br />

Definition<br />

<strong>ETIS</strong> Glossary<br />

Computation Method (Formula)<br />

Noise levels generated by rail transport + corresponding unit cost<br />

Noise generate by a link in Watt/ Exposure to over a specific noise level in dB<br />

COMMUTE<br />

· sound intensity = sound power per metre/(p x d)<br />

· sound pressure = square root (sound intensity x 415)<br />

· sound pressure level = 20 x log(sound pressure/0.00002)<br />

· link sound power= number of vehicles*corrected source sound power<br />

· corrected sound power= reference sound power per vehicle per hour*tread break correction factor* disc brake correction factor<br />

+ Corresponding unit cost from UNITE?<br />

RECORDIT<br />

for freight transport<br />

Variable definition<br />

Variable computation method<br />

Directly from a database or agreed<br />

classifications/nomenclatures<br />

Output of the model<br />

Method variable V1 n. of disc braked passenger trains *<br />

Method variable V2 p­ percentage of freight trains *<br />

Method variable V3<br />

reference sound power per hour<br />

Method variable V4 number of diesel and electric locomotives per hour *<br />

Method variable V5 speed *<br />

132<br />

Document2<br />

27 may 2004


ANNEX C: EXPERIENCE FROM CONCERTED<br />

ACTION ON SHORT SEA SHIPPING<br />

PROJECT


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Introduction<br />

MDS­Transmodal were retained as subcontractors within the Concerted Action on Shortsea<br />

Shipping contract to collect and process statistics that would form part of a detailed regional<br />

traffic matrix for freight flows between EU member states and the rest of the world. Recent<br />

studies have highlighted the importance of accurate OD matrices within transport modelling and<br />

transport policy applications, and the absence of a single reliable source that takes advantage of<br />

the regional statistics that are compiled at national levels.<br />

The over­riding objectives of this work were therefore to:<br />

· Obtain national sources of regional traffic,<br />

· Process them into a common format, and to<br />

· Combine them to create region­region flows.<br />

Scope<br />

There is no ideal set of definitions for a project of this nature, so an attempt has been made to<br />

balance the need for detail with the need for robustness.<br />

The geographical coverage has been designed to address the needs of projects related to short<br />

sea shipping. This includes trade between EU and EFTA member countries, and also flows<br />

between other nearby countries such as Eastern Europe, and the non­European Mediterranean.<br />

Within member states, it has been necessary to design the zone structures so that they follow<br />

existing regional boundaries used by the national governments. In general, it has been possible<br />

to follow the NUTS (Nomenclature of Territorial Units for Statistics) system designed by the<br />

Statistical Office of the European Communities. However, different levels of precision have<br />

been applied in different countries.<br />

This amounts to a "central" matrix of 128 European (EU, Norway and Switzerland) zones,<br />

surrounded by external zones defined either as countries or groups of countries.<br />

As far as commodity detail is concerned, the SITC (Standard International Trade Classification)<br />

system has been used, as this provides a hierarchical framework that works well at the national<br />

level (where there is a detailed record of the commodity split) and regional levels (where the use<br />

of sample data may require a more aggregated approach to commodity detail).<br />

One particular advantage of the SITC system is that at the 2­digit level (65 different commodity<br />

definitions) there is enough detail to be able to identify the main handling characteristics of<br />

specific goods, particularly manufactures, without the need for several hundred definitions.<br />

The alternative NST system is commonly used in Europe, but it has proved possible to convert<br />

from NST to SITC wherever the need has arisen.<br />

Sub­Division of Work<br />

Document2<br />

27 May 2004<br />

135


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

To take advantage of the different areas of specialisation between the team members, the work<br />

has been sub­divided. MDS­Transmodal were responsible for obtaining and processing data<br />

from:<br />

· France<br />

· Spain, and<br />

· The United Kingdom<br />

In addition, MDS­Transmodal was required to produce a database of country­countrycommodity<br />

totals based on the Eurostat Comext (Trade Statistics) database. The database was<br />

enhanced by the estimation of the unitised/non­unitised split for each trade flow. This is<br />

fundamental for understanding the demand for specific types of transport e.g. trailers,<br />

containers, general cargo, liquid bulk, dry bulk etc.<br />

Finally, MDS­Transmodal was supplied with a German regional database, which was combined<br />

with the regional data from the other listed countries to produce a region­region matrix covering<br />

France, Spain, the UK and Germany.<br />

Data Collection<br />

The success of this project depends to large extent on the type of data readily available from the<br />

member states. Although it is technically possible to construct synthetic matrices purely on the<br />

basis of measures of economic activity within specific regions, the probability of error is severe.<br />

It is therefore critical to be able to analyse trip end totals at a regional level. This provides the<br />

necessary input for understanding which industries are located where, and their relative<br />

importance.<br />

For all three countries within the MDS­Transmodal remit, regional data has been successfully<br />

obtained.<br />

· France<br />

The French source is the DNSCE (Customs and Excise) database of external trade. This<br />

conveniently classifies trade movements by country of origin/destination, commodity (NST),<br />

French Department (NUTS3), and volume (weight and value). The data used within this study<br />

is for 1997 full year, and is essentially a complete record of regional imports and exports.<br />

The data was processed to remove superfluous data fields, and to compress the commodity<br />

definition from NST­4 Digit to SITC­2Digit, using a standard correlation table. The country<br />

codes were the EC standard Comext codes.<br />

The main addition was therefore to convert the total tonnes (as given) into unitised and nonunitised<br />

tonnes, using look­up tables detailed for each partner country and commodity, derived<br />

from MDS­Transmodal's trade data archive. The main problem was to deal with country and<br />

136<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

commodity combinations that did not match with any records in the MDS­Transmodal look­up<br />

tables. For these, an iterative process was developed so that mode factors could be obtained<br />

from similar countries or similar commodities or both.<br />

Problems also arise from the inclusion of "Departements d'outre­mer" or "DOM", i.e. the West<br />

Indies as regions of France. Users of the database need to be aware that flows into these<br />

departements do not represent short sea shipping.<br />

The NUTS3 Departements were aggregated to NUTS2 Regions, reducing the total number of<br />

sub­national zones from 100 (96 excluding DOM) to 21 mainland regions <strong>plus</strong> Corse,<br />

Guadeloupe, Martinique, Guyane, and Reunion.<br />

· Spain<br />

The situation regarding Spain is somewhat similar to France. Again, the source was a Customs<br />

& Excise database, containing regional detail for 50 zones in Spain ("Provincias"). This is<br />

equivalent to NUTS3. The database was dispatched on seven magnetic tapes, and amounted to<br />

over 700 megabytes of data.<br />

A database program was developed by MDS­Transmodal to read the files extracted from the<br />

tapes, and to compress the data by lifting out the key fields of data. The commodity<br />

classification system used was the standard 8 digit Harmonised System, as used within Comext<br />

as well as the majority of countries worldwide. Again, correlation tables were used to convert<br />

this to 2 digit SITC.<br />

Country codes were again based on the standard Eurostat practice.<br />

As before, the estimated unitised/non­unitised split was introduced, using the factors already<br />

calculated by MDS­Transmodal at the national level.<br />

The main problem has been the need to compress vast quantities of data to a few megabytes,<br />

and the relative unreliabity of using tapes as a storage media. As with the French database,<br />

offshore regions such as the Canarias, Baleares, Ceuta and Melilla have required special<br />

treatment.<br />

The 52 NUTS3 Provincias (including Ceuta and Melilla) have been aggregated to 15 mainland<br />

NUTS2 Communidades Autonomas <strong>plus</strong> Baleares, Canarias, and Ceuta­y­Melilla.<br />

· The UK<br />

The UK experience is somewhat different to Spain and France. UK trade statistics have never<br />

recorded regional data such as UK country of origin or destination. The main source of regional<br />

data has been the Origin and Destination of International Transport (ODIT) survey, carried out<br />

by the UK's Department of the Environment, Transport and the Regions (DETR). This is based<br />

upon a sampling technique, and is only carried out every five years (1986, 1991, 1996).<br />

Document2<br />

27 May 2004<br />

137


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Ideally, it would have been possible to use just the 1996 survey, which was partly funded by the<br />

STEMM project (DGVII), but experience has shown that this survey contains deficiencies. In<br />

order to complete the freight modelling work within the STEMM project, the survey data was<br />

enhanced by contributing data from other sources such as the 1996 International Road Haulage<br />

Survey (IRHS) and the Railfreight Distribution database. Finally it was grossed up using UK<br />

trade data for 1996, and given greater regional strength by incorporating parts of the 1991<br />

ODIT, which had a stronger <strong>methodology</strong>. (See STEMM Report).<br />

The ODIT source is essentially a record of unitised trade flows only as the consolidation of<br />

bulks in port silos and tanks makes it very difficult to match origins to destinations using a<br />

questionnaire approach. It was not considered sensible to estimate the inland origins and<br />

destinations of such commodities.<br />

Regional data is collected at the county level (NUTS3). Problems have been encountered in<br />

mixing basic data from different years, as a series of changes have occurred in the definition of<br />

county boundaries. However, in aggregating to ten NUTS1 standard regions, (excluding<br />

Northern Ireland) these inconsistencies have been removed.<br />

Building Region­Region Matrices: The Problem<br />

The final stage has been to tackle one of the classic transport problems which is how to convert<br />

data of the form:<br />

Region R1 (belonging to Country C1) to Country C2: T1 tonnes.<br />

Region R2 (belonging to Country C2) to Country C1: T2 tonnes.<br />

Into:<br />

Region R1(in C1) to Region R2 (in C2) : T3 tonnes.<br />

The problem can be observed as an origin­destination matrix where the row and column totals<br />

are known, but the individual cells are unknown.<br />

138<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

For example:<br />

Orig/ Dest Dest1 Dest2 Dest3 Total<br />

Orig1 100<br />

Orig2 200<br />

Orig3 300<br />

Total 150 250 200 600<br />

For this study, the problem can be seen as a series of large matrices, one for each commodity.<br />

The origins and destinations are a mixture of countries and regions. Country to/from country<br />

and country to/from region flows will be known, but subsets of the matrices will appear as<br />

above, e.g. France to/from Spain.<br />

For example, Iron and Steel Trade,<br />

Orig/ Dest Champagne Picardie Hte Normandie Total<br />

Galicia 100<br />

Asturias 200<br />

Cantabria 300<br />

Total 150 250 200 600<br />

There are various iterative procedures for finding "solutions" to these problems, but these are<br />

essentially computer algorithms that have too many degrees of freedom to reach conclusive<br />

results. They fit but they are not necessarily right.<br />

It is possible to "seed" the matrix by filling it with particular values before the iteration solves it.<br />

The seeding biases the results, so if the seeding is performed according to a valid theory of what<br />

the matrix represents, it should improve the result.<br />

One possibility is the so­called "Gravity Model" which takes into account the distance between<br />

any two regions in the matrix. By introducing seeded values inversely proportional to the<br />

distance (interpreted as the attraction between cells) between the regions it ought to be possible<br />

to improve accuracy. Further accuracy might be achieved by extending the analysis to using<br />

Generalised Cost instead of pure distance, and other factors such as language compatibility,<br />

common currency, joint membership of trade bloc, and so on.<br />

Many sophisticated procedures can be hypothesized, but without any base matrices to test the<br />

results against, it is difficult to judge their validity. In these circumstances, if the base matrix<br />

were known, there would be no practical reason for trying to estimate it. This is the problem.<br />

Document2<br />

27 May 2004<br />

139


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Building Region­Region Matrices: A Solution<br />

The solution employed by MDS­Transmodal has been to build a Gravity model for seeding the<br />

unknown cells in the matrix, and to optimise this for each commodity, using an optimisation<br />

algorithm known as an "amoeba". The optimisation was conducted on a country­country matrix<br />

where all the genuine values for the cells were known in advance. The parameters were stored,<br />

and then re­used when the national data was broken down into regions.<br />

Thus, the problem was explored at a national level, and then an optimum solution was<br />

transplanted to a regional level. Naturally, the regional results produced could not be tested, but<br />

at least the matrix could be seeded with values known to produce vastly improved results at a<br />

more aggregated level. Furthermore, the pattern of parameter values for the gravity model used<br />

to implement the seeding, could be used to categorize specific commodities in terms of their<br />

propensity to be widely or narrowly distributed.<br />

Step 1: Selecting the Seeding Model<br />

The selection process was driven mainly by practicality. It was decided that for simplicity, just 3<br />

parameters would be considered:<br />

1. A measure of the cost in terms of distance between the 2 regions<br />

2. The total exports of the exporting region<br />

3. The total imports of the importing region<br />

Volume indicators such as population and GDP tended to be highly correlated with the total<br />

exports/imports.<br />

A simple formula based on physical common sense could be:<br />

T = d<br />

e<br />

n - md<br />

EI<br />

where T= trade, d= measure of cost in terms of distance, E= total exports of exporting region, I=<br />

total imports of importing region. The n and m are variables.<br />

Step 2: Optimising the Parameters<br />

After running this formula, a furnessing algorithm is run to ‘massage’ these ‘seed’ values to the<br />

known totals. The furnessing algorithm finds the existing total of one column and compares this<br />

to the required column total. All values in this column are then scaled up accordingly such that<br />

the new column total equals the required column total. This is done separately for all columns.<br />

This process is then repeated for the rows. If this combined column and row process is repeated<br />

several times, gradually the cell values become stable and converge.<br />

It was noted that the column totals (E) are redundant because the furness scaling cancels them<br />

out immediately but for the formula to have a sensible physical representation, they were still<br />

included.<br />

140<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

The objective of this program is to find the optimum values of the variables so as to create a<br />

formula that will predict the main body of the trade table as accurately as possible.<br />

There has to be a method of evaluating the accuracy of the formula. There is some existing real<br />

trade data where the body of the table is already known. By comparing the calculated cell values<br />

to the known values in this data­set and adding up all the absolute differences between<br />

calculated and known values, an overall measure of the error is obtained.<br />

The variables ‘n’ and ‘m’ have to be varied so as to minimize this total error. This is done using<br />

the “amoeba” function.<br />

This final error can then be compared to the error if there had been no initial seeding (only the<br />

furnessing) and a percent improvement found.<br />

There is data for many different commodities so this process can be run for all of them,<br />

obtaining different variable values depending on the pattern of trade for each particular<br />

commodity.<br />

The optimized values for ‘m’ and ‘n’ for each SITC commodity are:<br />

Results<br />

SITC no. and name n m Error After<br />

Seeding<br />

Error Before<br />

Seeding %better<br />

00 Animals 4.72864 0.030949 663,155 850,538 22<br />

01 Meat ­0.429564 0.001106 2,287,460 3,214,260 29<br />

02 Dairy ­0.55379 0.003387 4,077,450 8,069,130 49<br />

03 Fish ­1.04392 0.000825 1,381,200 2,204,350 37<br />

04 Cereals ­2.86474 ­0.000378 2,567,290 4,326,630 41<br />

05 Fruit/Veg ­0.738579 0.000339 8,730,710 12,457,600 30<br />

06 Sugar ­0.637274 0.002179 2,196,140 3,458,890 37<br />

07 Coffee/Tea ­1.6522 0.000228 829,043 1,126,480 26<br />

08 Animal Feed ­2.0288 0.000598 2,582,060 4,520,360 43<br />

09 Misc Edibles ­1.07391 0.000701 1,057,500 1,742,980 39<br />

11 Beverage ­0.110612 0.055329 10,267,100 20,839,500 51<br />

12 Tobacco ­1.637 ­0.001496 274,518 291,650 6<br />

21 Hides Raw ­1.6443 0.000575 365,381 450,471 19<br />

22 Oil Seed ­1.54893 0.005966 16,746 28,214 41<br />

23 Cr Rubber ­0.364645 0.001242 517,973 660,432 22<br />

24 Wood ­4.25486 0.001628 332,351 1,076,000 69<br />

25 Pulp ­0.337764 0.005893 785,465 2,029,800 61<br />

26 Textiles ­1.3155 7.51E­05 716,393 1,049,480 32<br />

27 Crude Ferts ­4.05698 ­0.001776 6,818,050 10,439,800 35<br />

28 Ore/Scrap ­0.752897 0.001408 1,170,690 1,813,910 35<br />

29 Oth Crude Mats ­0.878426 0.001608 1,899,030 2,592,650 27<br />

32 Coal/Coke ­2.8805 0.000712 783,348 1,871,700 58<br />

Document2<br />

27 May 2004<br />

141


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

SITC no. and name n m Error After<br />

Seeding<br />

Error Before<br />

Seeding %better<br />

33 Petroleum ­4.18336 ­0.000854 335,128 551,877 39<br />

41 Animal Fats ­1.6663 0.001616 245,791 389,645 37<br />

42 Veg Fats ­0.180219 0.004108 948,911 1,621,800 41<br />

43 Other Oils ­1.33137 0.000556 378,736 506,854 25<br />

51 Organic Chem ­0.176975 0.000949 3,615,750 4,391,430 18<br />

52 Inorganic Chem ­0.545743 0.001132 8,905,440 12,919,100 31<br />

53 Dyes ­1.3526 0.000245 1,692,210 2,630,380 36<br />

54 Pharmaceuticals ­0.04056 0.000915 332,371 438,903 24<br />

55 Essential Oils ­1.18802 0.000254 1,714,550 2,565,630 33<br />

56 Other Fertilisers ­0.298037 0.001676 3,261,300 4,948,300 34<br />

57 Prim Plastics ­0.370048 0.000889 5,892,760 9,592,560 39<br />

58 Other Plastics ­1.01699 0.000331 1,333,080 1,886,520 29<br />

59 Chemical Materials ­1.78155 ­0.000152 4,747,050 7,263,560 35<br />

61 Leather ­2.85753 ­0.001231 82,734 108,033 23<br />

62 Rubber Manuf. ­0.759114 0.000127 977,820 1,174,680 17<br />

63 Wood Manuf. ­2.09861 ­6.79E­05 2,266,900 4,268,710 47<br />

64 Paper ­1.15295 0.000397 5,180,710 7,184,460 28<br />

65 Textiles ­0.749137 0.000149 1,657,590 2,222,400 25<br />

66 Non Metallic Manuf ­2.57571 ­0.000317 11,136,100 22,532,500 51<br />

67 Steel ­1.14903 0.000178 3,901,870 5,928,690 34<br />

68 Non Ferrous Metals ­1.15747 0.000115 3,705,220 5,077,670 27<br />

69 Other Metal Manuf. ­2.44048 ­0.000498 2,992,420 5,806,260 48<br />

71 Power Machinery 0.531799 0.001082 1,149,270 1,236,270 7<br />

72 Specialise Machinery ­1.11267 ­4.78E­06 866,170 1,284,450 33<br />

73 Metal Machinery ­1.22889 ­0.000244 197,304 236,321 17<br />

74 General Industrial Mc ­0.668388 0.000239 1,867,980 2,319,570 19<br />

75 Office Machinery ­0.064217 0.000876 330,987 391,512 15<br />

76 Telecom ­0.566591 0.000382 532,228 614,457 13<br />

77 Electrical Mc ­0.72881 0.000168 2,315,210 2,639,640 12<br />

78 Road Vehicles 0.089321 0.000762 4,156,750 4,607,230 10<br />

79 Oth Transport Equip ­1.17526 0.000567 98,373 130,445 25<br />

81 Prefabs ­2.88769 ­0.00085 772,798 1,202,500 36<br />

82 Furniture ­1.89748 ­0.000158 1,397,840 2,228,640 37<br />

83 Travel Goods ­1.5887 ­0.000142 60,105 69,651 14<br />

84 Clothes ­1.3858 ­0.00024 431,867 521,253 17<br />

85 Footwear ­0.456656 0.000104 184,464 196,684 6<br />

87 Scientific Machinery ­0.274025 0.000514 161,173 181,466 11<br />

88 Photographic Mc. ­0.421964 0.000564 136,003 165,466 18<br />

89 Miscellaneous Manuf ­1.31818 0.00017 2,686,140 4,202,180 36<br />

90 Others 27.7121 0.021775 52,801 68,881 23<br />

Total ­1.22196 9.08E­05 101,163,000 179,847,000 44<br />

142<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Step 3: Appraising the Results<br />

This data is represented below in graphical form: m versus n.<br />

The more negative n is, the quicker the function goes to zero because of the d to the power of n<br />

term. The more positive m is the quicker the function goes to zero because of the exponential<br />

term. The difference is in the shape of the curve:<br />

· An exponential curve behaves sensibly near zero ­ starting at 1 and gradually decreasing.<br />

Then at high distance it becomes very small.<br />

· A power­n (n negative) curve starts at infinity (at distance = zero) and quickly comes<br />

down but is less active than the exponential term at higher distances.<br />

If n is positive, the seeding graph of trade vs distance initially goes up from zero (for small d ­<br />

while the exponential term is negligible). As the exponential term catches up and becomes<br />

dominant at larger d, trade falls back down again to approach zero at high distances.<br />

A large positive n keeps the exponential term negligible until higher distance so the graph<br />

continues to rise for longer. A large positive m allows the exponential term to quickly dominate,<br />

keeping the peak close to zero distance.<br />

The variable m should not be negative. Where it is, n should always be negative and the<br />

exponential term does not become dominant until beyond the fitted points. Although this may fit<br />

the data for the distances fitted, it is unrealistic for larger distances. Negative m should only<br />

result where the data is very sparse.<br />

Most commodities fit into the n negative, m positive position in mn space. The actual position<br />

depends on the shape of the trade vs distance curve. A highly negative n shows that the graph<br />

quickly descends within relatively small distances without giving much information as to trade<br />

at large distances. A large m shows that there is very little trade at large distances without<br />

giving much information about the small distance trade.<br />

So on the graphs below; negative n is a measure of the speed of descent of the curve at short<br />

distances. Positive m is a measure of the speed of descent of the curve at larger distances.<br />

The apparently anomalous value for ‘90 Others’ is not necessarily wrong ­ it is just based on<br />

very sparse data and is the best estimate.<br />

11 Beverage has a high value of m. As it’s n is not very negative, this shows that at small<br />

distances, there is little dependence of trade on distance but at high distances, the trade quickly<br />

decreases.<br />

00 Animals again suffers from sparse data. This point suggests that the trade has a non­zerodistance<br />

peak before descending rapidly at high distance.<br />

Document2<br />

27 May 2004<br />

143


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Step 4: Application<br />

New software was created to apply the results of this Gravity model to the actual trade<br />

databases, at the regional level. The software used the basic country­country­commodity<br />

database which originated from Eurostat.<br />

Where regions for neither country were known, the Eurostat record was passed straight into the<br />

output database.<br />

Where regions for one country were known (France, Spain, UK and Germany) the national<br />

totals were split according to the proportions recorded in the regional databases, and grossed up<br />

to the Eurostat figure for consistency.<br />

Where regions were known in both <strong>report</strong>ing and partner countries. The matrix was seeded with<br />

values from the optimisation process and the furnessing technique was applied.<br />

144<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Plotted Values for ‘n’ and ‘m’<br />

0.06<br />

Commodities<br />

MT­11 Beverage<br />

0.05<br />

0.04<br />

0.03<br />

MT­00 Animals<br />

m<br />

MT­90 Others<br />

0.02<br />

0.01<br />

0<br />

MT­22 Oil MT­25 Seed Pulp<br />

MT­42 Veg Fats<br />

MT­02 Dairy<br />

MT­06 Sugar<br />

MT­24 Wood MT­41 MT­29 MT­28<br />

MT­56 Anm Oth<br />

MT­23<br />

MT­03<br />

MT­01<br />

Ore/Scra Fats Oth C Ma Fert<br />

MT­52 Cr<br />

Fish<br />

Meat Rubbe<br />

MT­32 MT­08 MT­09<br />

MT­07<br />

MT­21 Coal/Cok MT­51<br />

MT­05<br />

Anm<br />

MT­57 MT­54 MT­71 Inorg<br />

Misc Org<br />

Feed<br />

Prim Pharmac Power<br />

Edb Chem<br />

Pla MC<br />

MT­43<br />

Coff/Tea<br />

Hides Fruit/Veg Oils Ra<br />

MT­26 MT­55<br />

MT­79 MT­58 MT­53 Textiles Ess Oth Dyes Oils Plas<br />

MT­04<br />

MT­63 MT­59<br />

MT­68 MT­62<br />

MT­64 MT­67<br />

Cereals<br />

Wood MT­65<br />

Chem<br />

Non Rubb<br />

Paper<br />

Man Steel<br />

Mat<br />

Ferr Man<br />

MT­66 MT­69 MT­73<br />

MT­72<br />

N MT­74 MT­75<br />

Met M M Manf Metal<br />

Spec<br />

M<br />

Gen Off<br />

MC<br />

MC Indu<br />

MC<br />

MT­76 MT­88 MT­87<br />

MT­77 MT­78 Oth Telecom Photo Scient Tr<br />

Elec Road E MC M<br />

MC<br />

Veh<br />

MT­82 MT­83<br />

MT­89 MT­85 MT­Total Furnitur Travel<br />

Misc Footwear Man<br />

MT­84 Clothes G<br />

MT­33 MT­81 Petroleu Prefabs<br />

MT­61<br />

MT­27 Cr Ferts<br />

MT­12<br />

Leather<br />

Tobacco<br />

­0.01<br />

­5 0 5 10 15 20 25 30<br />

n<br />

Document2<br />

27 May 2004<br />

145


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

0.006<br />

Commodities<br />

MT­22 Oil Seed<br />

MT­25 Pulp<br />

0.005<br />

0.004<br />

MT­42 Veg Fats<br />

MT­02 Dairy<br />

0.003<br />

m<br />

0.002<br />

MT­06 Sugar<br />

MT­24 Wood<br />

MT­41 Anm Fats<br />

MT­56 Oth Fert<br />

MT­29 Oth C Ma<br />

MT­28 Ore/Scra<br />

0.001<br />

0<br />

MT­32 Coal/Cok<br />

MT­03 Fish<br />

MT­23 Cr Rubbe<br />

MT­52 MT­01 Inorg Meat Ch<br />

MT­51<br />

MT­54<br />

Org Chem<br />

MT­57 Prim Pla Pharmac<br />

MT­75 Off MC<br />

MT­64 Paper MT­76 Telecom<br />

MT­58 Oth MT­05 Plas Fruit/Veg<br />

MT­07 Coff/Tea MT­53 MT­55 Dyes Ess Oils MT­74 Gen Indu<br />

MT­89 MT­67 Misc Man Steel MT­77<br />

MT­62 MT­65 Elec<br />

Rubb Textile MC<br />

MT­68 Man<br />

MT­26 MT­Total Non Ferr<br />

Textiles<br />

MT­85 Footwear<br />

MT­72 Spec MC<br />

MT­63 Wood Man<br />

MT­82 MT­59 Furnitur MT­83 Chem Travel Mat G<br />

MT­84 MT­73 Clothes Metal MC<br />

MT­66 N M M M<br />

MT­04 Cereals<br />

MT­69 Met Manf<br />

MT­78 Road Veh<br />

MT­09 Misc Edb<br />

MT­08 Anm MT­21 Feed Hides MT­43 MT­79 Ra Oth Oth Oils Tr E MT­88 Photo MC<br />

MT­87 Scient M<br />

MT­71 Power MC<br />

­0.001<br />

MT­33 Petroleu<br />

MT­81 Prefabs<br />

MT­61 Leather<br />

MT­12 Tobacco<br />

MT­27 Cr Ferts<br />

­0.002<br />

­4.5 ­4 ­3.5 ­3 ­2.5 ­2 ­1.5 ­1 ­0.5 0 0.5 1<br />

n<br />

146<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

0.0012<br />

Commodities<br />

0.0011<br />

MT­52 Inorg Ch<br />

MT­01 Meat<br />

0.001<br />

MT­51 Org Chem<br />

0.0009<br />

MT­57 Prim Pla<br />

MT­54 Pharmac<br />

MT­75 Off MC<br />

MT­03 Fish<br />

0.0008<br />

0.0007<br />

MT­09 Misc Edb<br />

0.0006<br />

MT­21 Hides Ra<br />

MT­79 Oth Tr E<br />

MT­43 Oth Oils<br />

MT­88 Photo MC<br />

0.0005<br />

MT­87 Scient M<br />

m<br />

0.0004<br />

MT­64 Paper<br />

MT­76 Telecom<br />

MT­58 Oth Plas<br />

MT­05 Fruit/Veg<br />

0.0003<br />

MT­07 Coff/Tea<br />

MT­55 Ess Oils<br />

MT­53 Dyes<br />

MT­74 Gen Indu<br />

0.0002<br />

0.0001<br />

MT­67 Steel<br />

MT­89 Misc Man<br />

MT­68 Non Ferr<br />

MT­Total<br />

MT­26 Textiles<br />

MT­77 Elec MC<br />

MT­65 Textile<br />

MT­62 Rubb Man<br />

MT­85 Footwear<br />

0<br />

MT­72 Spec MC<br />

­1E­04<br />

MT­83 Travel G<br />

MT­82 Furnitur<br />

MT­59 Chem Mat<br />

­0.0002<br />

MT­84 Clothes MT­73 Metal MC<br />

­0.0003<br />

­2 ­1.8 ­1.6 ­1.4 ­1.2 ­1 ­0.8 ­0.6 ­0.4 ­0.2 0<br />

n<br />

Document2<br />

27 May 2004<br />

147


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Sample Output <strong>Database</strong><br />

Type 1: No Regional Data<br />

Direction Rep RepName RepReg Par ParName ParReg Sitc2 UTonnes<br />

Imp 39 SWITZERLAND DKReg 804 NEW ZEALAND DKReg 84 6<br />

Imp 39 SWITZERLAND DKReg 804 NEW ZEALAND DKReg 85 1<br />

Imp 39 SWITZERLAND DKReg 804 NEW ZEALAND DKReg 87 5<br />

Imp 39 SWITZERLAND DKReg 804 NEW ZEALAND DKReg 89 33<br />

Imp 39 SWITZERLAND DKReg 807 TUVALU DKReg 3 1<br />

Imp 39 SWITZERLAND DKReg 810 US OCEANIA DKReg 3 1<br />

Imp 39 SWITZERLAND DKReg 810 US OCEANIA DKReg 5 6<br />

Imp 39 SWITZERLAND DKReg 813 PITCAIRN DKReg 5 11<br />

Imp 39 SWITZERLAND DKReg 813 PITCAIRN DKReg 58 9<br />

Imp 39 SWITZERLAND DKReg 815 FIJI DKReg 82 4<br />

Type 2: Regional Data at One End<br />

Direction Rep RepName RepReg Par ParName ParReg Sitc2 UTonnes<br />

Exp 1 FRANCE R210 2 BELGIUM/LUX DKReg 0 446<br />

Exp 1 FRANCE R221 2 BELGIUM/LUX DKReg 0 3735<br />

Exp 1 FRANCE R222 2 BELGIUM/LUX DKReg 0 7428<br />

Exp 1 FRANCE R223 2 BELGIUM/LUX DKReg 0 904<br />

Exp 1 FRANCE R224 2 BELGIUM/LUX DKReg 0 332<br />

Exp 1 FRANCE R225 2 BELGIUM/LUX DKReg 0 99<br />

Exp 1 FRANCE R226 2 BELGIUM/LUX DKReg 0 628<br />

Exp 1 FRANCE R230 2 BELGIUM/LUX DKReg 0 39441<br />

Exp 1 FRANCE R241 2 BELGIUM/LUX DKReg 0 2058<br />

Exp 1 FRANCE R242 2 BELGIUM/LUX DKReg 0 48<br />

Type 3: Regional Data at Both Ends<br />

Direction Rep RepName RepReg Par ParName ParReg Sitc2 UTonnes<br />

Exp 1 FRANCE R241 4 GERMANY R17 4 3558<br />

Exp 1 FRANCE R241 4 GERMANY R15 4 61619<br />

Exp 1 FRANCE R241 4 GERMANY R16 4 2736<br />

Exp 1 FRANCE R241 4 GERMANY R1D 4 70<br />

Exp 1 FRANCE R241 4 GERMANY R1E 4 115<br />

Exp 1 FRANCE R241 4 GERMANY R18 4 5043<br />

Exp 1 FRANCE R242 4 GERMANY R11 4 3318<br />

Exp 1 FRANCE R242 4 GERMANY R19 4 14768<br />

148<br />

Document2<br />

27 May 2004


ANNEX D: THEORETICAL ANALYSIS OF<br />

TRANSPORT CHAIN STRUCTURES<br />

IN DATABASES; CONSTRAINTS AND<br />

POSSIBILITIES


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Introduction<br />

As already mentioned one of the major gaps in freight data availability is the freight OD­matrix.<br />

Within several projects (for instance INFOSTAT, CONERTO, MESUDEMO, INFREDAT) it<br />

has been identified that it is important to consider a transport chain structure in the freight<br />

database including transhipment of transport flows between modes and other characteristics.<br />

Before describing how the freight OD transport chain database should be constructed, the<br />

difficulties being encountered first are being discussed in a broader context in this chapter. This<br />

is a general theoretical analysis of transport chain data without looking at availability of sources.<br />

From the analysis in this chapter questions evolve that need to be answered within <strong>ETIS</strong> in the<br />

light of (inter­modal) freight transport. First some basic concepts are being discussed after<br />

which the characteristics of transport chain are described. Eventually some words are being said<br />

on the level of detail and concluding remarks are given.<br />

Transport chains<br />

The transport Chain Concept<br />

In most databases a strict distinction is made between trade and transport databases. In trade<br />

databases the commodity flows are described from origin to destination and in transport<br />

databases the vehicle/vessel flows are described from origin to destination. <strong>Database</strong>s<br />

containing both definitions are very rare. Some sources of this type exist for a limited<br />

geographical area like for instance the trade data of the UK and France.<br />

Using the demand­oriented philosophy we can estimate the future transport flows by estimating<br />

the future trade flows between regions/countries. An important requirement is that the economic<br />

trade relation of the commodity is maintained in the database describing the current transport<br />

flows. Since transport from the producing region/country to the consuming region/country often<br />

takes place with several modes it also has large advantages to have these possible mode changes<br />

in the database. Furthermore if a change of modes (or vehicles) occurs it is also interesting to<br />

have this place of transhipment described. A record structure that includes all these variables<br />

can be called a ‘transport chain structure’; the good is followed from the place of production via<br />

transhipment locations to the place of consumption where several modes can be used.<br />

This transport chain concept has the following characteristics:<br />

1. It is an element of response to the complex situation of transport<br />

2. It has been proven in the past to be a practicable concept<br />

3. It reconciles transport/economical approach<br />

4. It opens ways for more insights in the transport organisation<br />

5. It eliminates double counting<br />

6. It is compatible with current statistics<br />

<strong>Database</strong>s that do exist with transport chain information are mostly constructed from several<br />

sources where on some places estimates have to be made and inconsistencies can exist between<br />

databases of different countries. Also the level of detail of the transport chain is limited in<br />

Document2<br />

27 May 2004<br />

151


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

constructed databases; because of the estimates that have to be made the reliability of the data<br />

decreases with the increase of the level of detail.<br />

It therefore can be concluded that a reliable transport chain database with a high level of detail<br />

can only become available if these data are collected by forms like has been attempted in the<br />

MYSTIC project and is being tested in some national pilot exercises (France, Netherlands,<br />

Sweden) of which no final results have been obtained yet. In any other case the level of detail<br />

has to be reduced to come to a suitable reliability level.<br />

Restrictions on the Ideal <strong>Database</strong><br />

The issue to be discussed is what the ideal transport chain information to be collected must<br />

consist. In this chapter the problems to collect the data will be ignored but the focus will only be<br />

on possibilities and problems of construction and handling (usage) of such a database. An ideal<br />

situation of the database to be constructed will be shown. The ideal transport database has the<br />

following general restrictions:<br />

· Technical restrictions<br />

· Understandability<br />

· Policy requirements<br />

There still are some technical restrictions. If the database becomes large it becomes more<br />

complex and time consuming to perform manipulations and extractions. If the record structure<br />

becomes very complex it will take even more effort and time to develop the necessary software.<br />

Besides this technical side of having a large and complex database we also have the side of<br />

interpretation of the results from this database. If one wants to know the movements between an<br />

origin and destination it will be very important to know whether one wants to know the<br />

movements of the commodities or the vehicles. The difficulty lies in the fact that now also the<br />

movements between transhipment points should be taken into account, making it more complex<br />

to make the right restrictions on the database. When having made an extraction from the<br />

database for a certain origin and destination, what can be seen? Are these the end points of the<br />

vehicle/vessel trip? Does it mean that the commodities transported have actually been switched<br />

between modes? Are these endpoints also the origin and destination of the commodities or for<br />

which part of the total flow does this hold? Much information should be provided to be able to<br />

make the right interpretation of the figures obtained from the database.<br />

Also aggregated information, which one is used to, like for instance a modal split figure,<br />

becomes more complex. How do we define the modal­split for a country knowing the transport<br />

chain structure of the database? To be able to provide a modal­split one needs to make<br />

assumptions. These assumptions have to be provided together with the figures to be able to<br />

make the right interpretations.<br />

152<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

In case different methods are applied based on the different available data sources that can vary<br />

significantly among the different countries also some diffusion of interpretations can occur. On<br />

some elements of the database and for some geographical areas more estimation will be<br />

necessary than for others. It is important that these differences can easily be traced back from<br />

the results shown from the database to be able to make the right interpretation.<br />

These situations where assumptions have to be made to be able to generate the database and<br />

information from this database, and where data can not be shown without the necessary<br />

information on these assumptions, we could capture under the ‘understandability’ restriction.<br />

This restrictions requires that the database and the surrounding system is structured in such a<br />

way that information needed for the interpretation of the information extracted can be easily be<br />

made available.<br />

In the first place the transport chain freight database should be a simplification of reality in<br />

order to be able to understand how and where transport takes place. This means that many<br />

variables that could be observed have to be left out. Since each analysis of a problem requires<br />

another level of detail of the database. The policy requirement restriction which are in this case<br />

the TEN policies, determine which data should be included and which not. This is described in<br />

section 4.3.<br />

In the next paragraphs realistic possibilities for the format of the transport chain database will be<br />

described.<br />

The ideal Transport Chain<br />

To get a better view of which variables should be included in the transport chain database, an<br />

example is presented in figure D.1.<br />

Document2<br />

27 May 2004<br />

153


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Figure D.1:<br />

Transportation of a computer<br />

manual<br />

cables<br />

mouse<br />

VAL<br />

A B C D E F G H I<br />

Ro­Ro<br />

container<br />

In this figure we follow a computer from its place of production in point A to the consumer. The<br />

computer is transported by truck to point B where it is stuffed in a container together with other<br />

commodities. This container is transported by truck to point C where the truck goes on a ship to<br />

be transported to point D (Ro ­Ro). In point D the truck drives of the ship and goes to point E<br />

where the container is unloaded from the truck and stripped. The computer now is loaded on<br />

another truck to be transported to point F. In point F Value Added Logistics (VAL) takes place;<br />

the computer is put in another box together with additional features like cables, mouse and users<br />

manual.<br />

The computer is now transported to point G where it is transhipped on a train to be transported<br />

to point H. Finally it is transhipped again on a truck to be transported to point I.<br />

From a transport point of view the following problems relevant for the <strong>ETIS</strong> reference database<br />

development project can be identified in this fictional example:<br />

154<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

1. Who do we define to be the consumer of the computer?; which type of transhipment to<br />

include?<br />

2. How to include simultaneous mode use in the database?<br />

3. How to include loading units (containers, swap bodies) in the database?<br />

4. Should there be a restriction on the number of links in a transport chain?<br />

Something that does not occur in the example is the case in which a container is empty. The<br />

following question can therefore be added:<br />

5. How to include empty loading units flows in the database?<br />

These problems will be discussed in the following paragraphs.<br />

A way to a “pragmatic” approach of transport chains<br />

Who do we define to be the Consumer of the Computer?; which Type of Transhipment to<br />

include?<br />

The first question to be answered is whom we will define to be the customer. Another and more<br />

general way of looking at this problem is which type of transhipment we should include. If it<br />

has been decided which transhipment locations to include this implicitly means that also the<br />

production and consumption locations are defined.<br />

Several types of transhipment can be identified:<br />

1. VAL (Value Added Logistics)<br />

2. Distribution<br />

3. Wholesale<br />

4. Change of mode<br />

5. Entrepot<br />

In our example VAL and change of mode can be identified.<br />

In the case of VAL a change of the characteristics of the commodities take place. The elements<br />

of this commodity that have to be considered the basic products find the end location at the<br />

VAL location while in fact they are transported in combination with the other basic products to<br />

another final location of consumption. In the computer example one of the basic products can<br />

even be given the same name as the final product.<br />

Distribution is another logistic phenomenon resulting in registration problems. Is the<br />

distribution centre the end point or just a transhipment location? Furthermore the transport from<br />

the distribution centre to the customers might be done with routing where different deliveries<br />

are combined into one trip, resulting in double countings in the database.<br />

For wholesale there is still another nuance. Here the commodity changes of owner, but is<br />

eventually transported to another location where it is consumed.<br />

Document2<br />

27 May 2004<br />

155


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Solutions to these problems are outside the scope of this project and will therefore not be<br />

considered.<br />

The focus will be mainly on change of mode and where possible on entrepot. Transhipment as a<br />

result of change of mode is the best known principle. It is also the easiest to include in transport<br />

chain databases. Entrepot can also be seen as a transhipment action. The difference is the<br />

duration of the transhipment. It would be an added value to include entrepot in a transport chain<br />

structure to be able to preserve the relation between the real origin and the destination.<br />

Including change of mode and entrepot will give a much clearer view on the routes that the<br />

products follow.<br />

How to include Simultaneous Mode Use in the <strong>Database</strong>?<br />

In figure 5.1 it can be observed that a truck is transported from point C to point D by ship. The<br />

problem that arises in this case is that in fact two modes are being used: road and sea. According<br />

to the classical approach three possibilities exist:<br />

1. Register as road from B to E<br />

2. Register as road from B to C and as sea from C to D and in another record registration as sea<br />

from C to D and as road from D to E<br />

3. Independent registration of flow from B to C as road, C to D as sea and D to E as road<br />

If 1. is used we loose the information that part of the trip was over sea. If 2. is used we loose the<br />

information that the cargo was not transhipped but that the truck was also on the ship. In 3 like<br />

in number 1, the information what type of transhipment takes place is lost. So none of the<br />

possibilities can cope with this Ro ­ Ro situation and the chain structure. These problems can be<br />

solved by using the transport chain concept. Another problem that is solved by using transport<br />

chains is that no double countings can occur when analysing the data.<br />

The same problems of course occur in case a train is transported by ship or a truck is transported<br />

by train. All these possible combinations of modes transported by other modes as ‘simultaneous<br />

mode use’ are defined.<br />

The registration of simultaneous mode use can be solved in several ways:<br />

1. Introduce another mode name for the combination of modes<br />

2. Introduce another variable that indicates whether simultaneous mode use has taken place.<br />

Possibility 1. has as disadvantage that in case we want to know what is actively transported by<br />

ship we have to aggregate several modes.<br />

With possibility 2. this problem is solved. Besides with possibility 2. a strict distinction can be<br />

made between the active mode and the passive mode that is transported.<br />

156<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

For instance the variables ‘active mode’ and ‘passive mode’ can be used. Both can have all<br />

distinct modes as value.<br />

Table D.1<br />

Example of Simultaneous Road use Registration<br />

Active mode Passive mode Meaning<br />

Sea<br />

Sea<br />

Rail<br />

Rail<br />

Road<br />

Rail<br />

Road<br />

­<br />

Commodity on a truck transported by a ship<br />

Commodity on a train transported by a ship<br />

Commodity on a truck transported by a train<br />

Commodity transported by train<br />

The concept of transport chain requires that definitions are clear and well used.<br />

It can be concluded that simultaneous mode use can be best implemented in a database by using<br />

a variable for the active mode and one for the passive mode in a transport chain structure.<br />

How to include loading units in the <strong>Database</strong>?<br />

In figure 5.1 it can be observed that between point B and E the computer is transported in a<br />

container. This container in its turn is transported first by truck then by Ro­ Ro on a ship and by<br />

truck again.<br />

Comparing this problem with the simultaneous mode use problem in the previous chapter it<br />

looks about the same. Also similar solutions are possible. It is difficult however to say whether<br />

a container must be defined as a mode, a commodity or something else.<br />

Since commodities are transported in a container it has some characteristics of a mode.<br />

However a container can not transport a commodity by itself. It has to be transported by a mode<br />

like a commodity. A container can also be seen as a package of products which makes it also a<br />

commodity.<br />

The most convenient solution is to define a distinct variable to indicate whether transport by<br />

container has taken place like in table 5.2.<br />

Document2<br />

27 May 2004<br />

157


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Table D.2<br />

Example of Container Registration<br />

Mode Container Meaning<br />

Sea<br />

Road<br />

Yes<br />

No<br />

Commodity transported over sea in a container<br />

Commodity transported by road not in a container<br />

A possible extension to this method is to include also a type of container (large, other) on the<br />

place where otherwise a ‘yes’ would occur. Still a further extension would be to include in stead<br />

of a variable for container only, a variable for classification of loading units types.<br />

Should there be a Restriction on the Number of Links in a Transport Chain?<br />

Combining all the solutions in the previous paragraphs for the problems encountered in the<br />

computer example we see that for registering one flow, many variables are needed. If we take a<br />

look at the transportation of the computer from point A to point F we see that six locations have<br />

to be registered. One origin, four transhipment locations and a destination. Also five modes (one<br />

for each link) have to be registered together with the passive modes and container indications.<br />

This means 21 (= 3 x 5+ 6) variables are necessary for this transport chain. Furthermore<br />

variables are needed to describe the commodity, the weight and the value. Many transport flows<br />

will be less complex but there will also be more complex transport flows meaning that even<br />

more variables are needed.<br />

A partial solution to reduce the size of the database is to use a dynamic data structure in which<br />

for every transhipment place a variable is ‘made’ in the software. This means that sophisticated<br />

software has to be developed. The flow structure however still will be complex and difficult to<br />

comprehend.<br />

Another more pragmatic solution is to restrict the number of links. This also has disadvantages<br />

however, since it is important in case of inter­modal transport to describe all transhipment that<br />

takes place. Furthermore reducing the number of transhipment points in the database also has<br />

some negative impacts on the reliability of the information to be extracted. To get more insight<br />

in this a look will be taken at the example in figure 5.4<br />

158<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Figure D.4:<br />

Flow with two Transhipments<br />

A B C D<br />

road sea road<br />

In this figure a transport flow can be seen from point A to point D with transhipment in point B<br />

and point C. In point B the commodities are transhipped from a truck to a ship and in point C<br />

from this ship to another truck. In case a record structure is considered where at least two<br />

transhipment points are possible, all information can be preserved. In case a restriction on the<br />

record structure is made, which makes it possible to include only one transhipment point part of<br />

the flow information has to be left out; not only one transhipment point has to be left out but<br />

also one mode. In the following table all possible registrations in this case are listed.<br />

Table D.3<br />

Possible Registrations of the Example in Figure D.4 with one transhipment<br />

variable<br />

Record Origin mode 1 transhipment mode 2 Destination<br />

1<br />

2<br />

3<br />

4<br />

A<br />

A<br />

A<br />

A<br />

Road<br />

Road<br />

Road<br />

Sea<br />

B<br />

B<br />

C<br />

C<br />

sea<br />

road<br />

road<br />

road<br />

D<br />

D<br />

D<br />

D<br />

The four possibilities are the result of two choices:<br />

1. which transhipment point to chose?<br />

2. which mode to chose?<br />

Without additional information on the transhipment points no good answer can be given on<br />

question 1. It can be recommended however to make one decision rule for the whole database.<br />

For example priority can be given to certain transhipment points. This means that in case this<br />

transhipment point is in the chain it will be chosen with higher priority than others. As a result,<br />

when analysing the total transhipment in for instance a port, the figures of the port with the<br />

highest priority will be the most complete. This also means that data for the other ports are not<br />

complete. Using other decision rules implicates that none of the transhipment points will have<br />

complete data.<br />

Several options can be chosen for the modes to be included.<br />

Document2<br />

27 May 2004<br />

159


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

1. modes at origin and destination<br />

2. modes before and after transhipment<br />

The records 2 and 3 result from the first decision rule and 1 and 4 from the second. Using the<br />

second decision rule has as advantage that it is made visible what type of transhipment has<br />

taken place. Again, due to the restriction of transhipment point loss of information occurs.<br />

What we learn from the problems described in this paragraph is that:<br />

· Including many variables results in a large database<br />

· restricting the number of transhipment points implies a reduction of information (smaller<br />

database)<br />

· restricting the number of transhipment points implies difficult interpretation of<br />

transhipment information<br />

· restricting the number of transhipment points implies that clear decision rules have to be<br />

chosen to avoid confusing results<br />

How to include empty loading unit flows in the database?<br />

To be able to analyse loading units movements it is also important to have information about the<br />

empty loading units transported. How should this type of information be included in the<br />

database? The question that arises is whether the empty loading units should be considered in<br />

the same way as the loaded loading units or as a commodity that is transported.<br />

Since in practice the transport of an empty loading unit is handled in the same way as the<br />

transport of a commodity it might be more convenient to do the same in the database; in practice<br />

one has to pay for this transport and a weight is registered for the loading unit. If the empty<br />

loading unit is treated as a commodity, a new code could be introduced for it. If this is done it<br />

can be treated in the same way as other commodities. A disadvantage of this method is that<br />

empty and loaded loading units must be traced in the database in different ways. It is also<br />

possible to introduce a variable indicating the number of containers transported. In this way the<br />

number of loaded and empty containers can be traced by combining it with the weight<br />

information (no weight means empty containers).<br />

In case it is considered in the same way as a loaded container, a special code can be defined for<br />

the loading unit variable for ‘empty container’. In case we use only one weight variable for the<br />

whole transport chain we register the weight of the transported commodities. As a result the<br />

weight of the transported ‘empty loading unit’ will be zero, which is a problem for both<br />

approaches. This means that if we want to know the amount of loading units transported we<br />

need to introduce another variable that registers for instance the weight of the loading units<br />

transported or the total weight of the commodities <strong>plus</strong> the weight of the container.<br />

160<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Another option is to include another variable with the number of empty loading units. However<br />

since the characteristics of this variable is so different it might be an option to register the empty<br />

loading units separate from the rest of the transport chain database.<br />

The theoretical ideal record structure<br />

In this section the ideal record structure of the transport chain freight database is described. The<br />

elements described before are included here:<br />

· origin<br />

· active mode 1<br />

· passive mode 1<br />

· loading unit 1<br />

· transhipment point1<br />

· active mode 2<br />

· passive mode 2<br />

· loading unit 2<br />

· transhipment point2<br />

. . .<br />

. . .<br />

. . .<br />

. . .<br />

· active mode n<br />

· passive mode n<br />

· loading unit n<br />

· transhipment point n<br />

· destination<br />

· commodity<br />

· number of loading units<br />

· net weight<br />

· value<br />

It must be observed that this ideal record structure can handle all existing transport flows. So not<br />

only inter­modal but also uni­modal transport.<br />

Document2<br />

27 May 2004<br />

161


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Table D.4:<br />

Examples for the use of the chain record structure<br />

Variable name 1) Direct road 2) Ro­Ro, Truck on<br />

ferry<br />

· origin<br />

· active mode 1<br />

· passive mode 1<br />

· loading unit 1<br />

· transhipment point1<br />

· active mode 2<br />

· passive mode 2<br />

· loading unit 2<br />

· transhipment point2<br />

. . .<br />

. . .<br />

. . .<br />

. . .<br />

· active mode n<br />

· passive mode n<br />

· loading unit n<br />

· transhipment point n<br />

· destination<br />

· commodity<br />

· number of loading units<br />

· net weight<br />

· value<br />

n.r.=not relevant<br />

Cataluna<br />

Road<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

­<br />

Bayern<br />

Agriculture<br />

n.r.<br />

n.r.<br />

n.r.<br />

Bremen<br />

Road<br />

­<br />

­<br />

Hoek van Holland<br />

Sea<br />

Road<br />

­<br />

Harwich<br />

Road<br />

­<br />

­<br />

­<br />

West Midlands<br />

Manufactured prod.<br />

n.r.<br />

n.r.<br />

n.r.<br />

3) Sea–rail­road,<br />

container<br />

USA<br />

Sea<br />

­<br />

Container<br />

Rotterdam<br />

Rail<br />

­<br />

Container<br />

Milano<br />

Road<br />

­<br />

­<br />

­<br />

Lombardia<br />

Manufactured goods<br />

n.r.<br />

n.r.<br />

n.r.<br />

In table D.4 some examples are shown on how this structure can be used. In the first example<br />

one can see how direct road transport can be included. In this example Agricultural products are<br />

transported to Bayern by truck without any transhipment or use of loading unit.<br />

In the second example one can see how Ro­Ro can be included. In this example manufactured<br />

products are transported from the region Bremen (D) to West Midlands (UK). The part between<br />

Hoek van Holland and Harwich is by ferry; the truck drives on the ship in Hoek van Holland<br />

and off again in Harwich and can be considered a passive mode on this link while it is active on<br />

the other links.<br />

In the final example we see transport from the USA to Lombardia (It) of manufactured<br />

products. The first part of the trip goes by ship to Rotterdam. The commodities are transported<br />

in a container, which goes from the ship on the train in Rotterdam. This train drives to Milano<br />

162<br />

Document2<br />

27 May 2004


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

where the container is unloaded and stripped. Eventually the commodities are loaded on the<br />

truck to be brought to the final destination in Lombardia not far from the terminal.<br />

Conclusions<br />

Introducing the transport chain principle for data collection on a European scale can solve part<br />

of the existing information problems. Not all problems are easy to solve due to the complexity<br />

of the possible solutions and the fact that a lot of effort is needed to perform data handling.<br />

Experience with transport chain databases show that it is not easy to use transport chain<br />

information to its full extent and to make the right interpretations. Even if only one<br />

transhipment location is included new insights can be obtained resulting in a better<br />

understanding but also resulting in spending more effort to find the right meaning of the data.<br />

Expert database users or sophisticated data extraction software can help overcoming this<br />

difficulty.<br />

Despite all these difficulties transport chains must be seen as a powerful tool with which much<br />

more information will be available than now is the case. The fact that besides the economic<br />

relation also (a part of) the transport relation is preserved, makes it possible to obtain better<br />

understanding of the economic mechanisms behind transport. This in its turn will lead to better<br />

understanding of the implications of policy decisions.<br />

Document2<br />

27 May 2004<br />

163


ANNEX E: NEWCRONOS TABLES (RELEVANT<br />

FOR <strong>ETIS</strong>)


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

aviation<br />

Inland<br />

waterways<br />

(inlandwww)<br />

Collection definition Collection heading table name table contents description<br />

F.V1. Transport<br />

measurement ­ passengers<br />

F.V2. Transport<br />

measurement ­ goods<br />

avmepass<br />

avmegood<br />

avpat40w<br />

avpat40e<br />

avpaieu<br />

avpaiwr<br />

avparkt<br />

avparksc<br />

avparkns<br />

avpat_ft<br />

avpatw<br />

avpat_f<br />

avpat_fr<br />

avgorkeu<br />

avgorkw<br />

1. Top 40 routes worldwide for each of the ICP airports in the Trans­European Airport<br />

Network<br />

2. Top 40 routes within the EU for each of the ICP airports in the Trans­European Airport<br />

Network<br />

3. International passengers at each of the ICP airports to all EU countries<br />

4. International passengers at each of the ICP airports to world regions<br />

5. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination passengers ­ total passengers carried within EU and Switzerland<br />

6. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination passengers ­ scheduled passengers carried to/from rest of world<br />

7. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination passengers ­ non­scheduled passengers carried to/from rest of world<br />

8. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination passengers ­ total passengers carried to/from rest of world<br />

9. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination passengers ­ total passengers carried worldwide<br />

10. EU countries and Switzerland : to/from EEA and Switzerland ­ origin/destination<br />

international passenger traffic<br />

11. EU countries and Switzerland : to/from world regions origin/destination international<br />

passenger traffic<br />

1. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination freight ­ total freight loaded/unloaded within EU and Switzerland<br />

2. Ranking of top 50 airports in EU and Switzerland in terms of international<br />

origin/destination freight­ total freight loaded/unloaded worldwide<br />

C.I. Transport infrastructure iwinfras iwintwvc 01. Navigable inland waterways by carrying capacity of vessels<br />

C.II. Transport equipment<br />

iwequipt<br />

iweqnvc<br />

iweqpowe<br />

iweqpowa<br />

iweqycon<br />

iweqycoc<br />

iweqycoa<br />

01. Number of self­propelled vessels, of dumb and pushed vessels by load capacity<br />

02. Power of self­propelled vessels by load capacity<br />

03. Power of self­propelled vessels, tugs and pushers by age<br />

04. Number of vessels by age<br />

05. Load capacity of self­propelled vessels and dumb and pushed vessels<br />

06. Load capacity of self­propelled vessels and dumb and pushed vessels by age<br />

Document2<br />

27 May 20043<br />

167


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

maritime<br />

Collection definition Collection heading table name table contents description<br />

C.III. Enterprises<br />

C.V3. Transport<br />

measurement ­ goods<br />

(Council Directive<br />

80/1119/EEC)<br />

iweconom<br />

iwmegood<br />

E. I. Passenger transport mamepa<br />

E. II. Freight mamego<br />

iwecentv<br />

iwecentc<br />

iwecnemp<br />

iwecexpe<br />

01. Number of inland waterway transport enterprises by number of vessels<br />

02. Carrying capacity of the enterprises vessels by number of vessels in enterprises<br />

03. Employment in inland waterways transport enterprises by number of vessels in<br />

enterprises<br />

05. Investment and maintenance expediture in vessels and infrastructure<br />

iwgoildg 01. International annual transport by link with loading country and by group of goods<br />

(1000 T, Mio Tkm)<br />

iwgoiulg 02. International annual transport by link with unloading country and by group of goods<br />

(1000 T, Mio Tkm)<br />

iwgoildv 03. International annual transport by link with loading country and vessel nationality (1000<br />

T, Mio Tkm)<br />

iwgoiulv 04. International annual transport by link with unloading country and vessel nationality<br />

(1000 T, Mio Tkm)<br />

iwgoild 05. International monthly transport by link with loading country (1000 T)<br />

iwgoiul 06. International monthly transport by link with unloading country (1000 T)<br />

iwgondcg<br />

iwgonvn<br />

07. National annual transport by distance class and group of goods (1000 T, Mio Tkm)<br />

08. National annual transport by vessel nationality (1000 T, Mio Tkm)<br />

iwgon 09. National monthly transport (1000 T)<br />

iwgorldg<br />

iwgorulg<br />

iwgotg<br />

iwgotvn<br />

iwgoalkv<br />

mamepaaa<br />

mamepaqm<br />

mamepaqc<br />

mamegoaa<br />

10. National annual transport by loading regions and by group of goods (T)<br />

11. National annual transport by unloading regions and by group of goods (T)<br />

12. Annual transit transport by country and group of goods (1000 T, Mio Tkm)<br />

13. Annual transit transport by country and vessel nationality (1000 T, Mio Tkm)<br />

14. Total quarterly transport by kind of vessel (1000 T, Mio Tkm)<br />

E.I.1. Passenger transport ­ Total annual figures for all ports (main and small from 2000) of<br />

<strong>report</strong>ing country<br />

E.I.2. Passenger transport ­ Quarterly figures for main ports of <strong>report</strong>ing countries (ports<br />

<strong>report</strong>ing over 200.000 passengers per year)<br />

E.I.3. Passenger transport ­ Quarterly figures for main ports of each <strong>report</strong>ing country<br />

E.II.1. Freight­ Total annual seaborne transport for all ports (main and small from 2000) of<br />

<strong>report</strong>ing countries<br />

Document2<br />

Document2<br />

27 May 2004<br />

168


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

pipeline<br />

rail<br />

Collection definition Collection heading table name table contents description<br />

mamegoam<br />

mamegoqm<br />

mamegoac<br />

E.II.2. Freight ­ Annual seaborne transport for main ports of <strong>report</strong>ing countries (ports<br />

handling over 1 mio tonnes per year)<br />

E.II.3. Freight ­ Quarterly seaborne transport for main ports of <strong>report</strong>ing countries (ports<br />

handling over 1 mio tonnes per year)<br />

E.II.4. Freight ­ Annual seaborne transport for main ports for each <strong>report</strong>ing country<br />

mamegoqc E.II.5. Freight ­ Quarterly seaborne transport for main ports for each <strong>report</strong>ing country<br />

(ports handling over 1 mio tonnes per year)<br />

E. III. Vessel traffic mameve mameveqm E.III.1. Vessel traffic<br />

D.I. Transport Infrastructure<br />

D.III. Enterprises, economic<br />

performances and<br />

employment<br />

D.V. Transport measurement<br />

­ goods<br />

A.I. Infrastructure<br />

A.II. Transport equipment<br />

plinfras<br />

pleconom<br />

plmealtp<br />

plinlgpl<br />

plincppl<br />

plecnent<br />

plecnemp<br />

plecexpe<br />

02. Transport within the national territory by type of transport operations and products<br />

(Mio Tkm)<br />

01. Lenght of pipelines operated<br />

02. Carrying capacity of pipelines operated<br />

01. Oil pipeline enterprises<br />

02. Employment in oil pipeline enterprises<br />

03. Investment and maintenance in oil pipeline infrastructure<br />

plmegood plmealpd 01. Transport within the national territory by type of transport operations and products<br />

(1000 t)<br />

rainlgt<br />

01. Length of tracks<br />

rainfras<br />

raequipt<br />

rainlgnt<br />

rainlggt<br />

rainlgpg<br />

rainlgtc<br />

raeqnlsp<br />

raeqnrsp<br />

raeqnltp<br />

raeqnrtp<br />

raeqpvtv<br />

raeqpacv<br />

raeqpvtc<br />

02. Length of lines, by number of tracks<br />

03. Length of lines, by track gauge<br />

04. Length of lines, by nature of transport<br />

05. Length of electrified lines, by type of current<br />

01. Number of locomotives, by source of power<br />

02. Number of railcars, by source of power<br />

03. Number of locomotives, by tractive power<br />

04. Number of railcars, by tractive power<br />

05. Passenger railway vehicles, by type of vehicle<br />

06. Passenger railway vehicles, by category of vehicle<br />

07. Capacity of passenger railway vehicles, by type of vehicle<br />

Document2<br />

27 May 20043<br />

169


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

Collection definition Collection heading table name table contents description<br />

A.III. Enterprises, economic<br />

performances and<br />

employment<br />

A.IV. Traffic<br />

A.V1. Transport<br />

measurement ­ passengers<br />

A.V3. Transport<br />

measurement ­ goods<br />

(detailed data from EC<br />

Directive and regulation from<br />

1982 onwards<br />

raeconom<br />

ratrafic<br />

ramepass<br />

ramegood<br />

raeqpacc<br />

raeqnvan<br />

raeqnwan<br />

raeqnwal<br />

raecnent<br />

raecnemp<br />

raecnems<br />

raecexpe<br />

ratrtvsp<br />

ratrvsp<br />

ratrhvsp<br />

ratrhvtv<br />

ratrsekm<br />

ratrsetk<br />

rapancin<br />

rapancit<br />

rapancim<br />

rapancic<br />

rapaiar<br />

rapaidp<br />

rapaal<br />

rapaap<br />

08. Capacity of passenger railway vehicles, by category of seats or berths<br />

09. Number of vans<br />

10. Number of wagons, by status of enterprise<br />

11. Load capacity of wagons, by status of enterprise<br />

01. Principal railway enterprises<br />

02. Employment in principal railway enterprises, by type of activity<br />

03. Employment in principal railway enterprises by sex<br />

04. Nature of expenditure in principal railway enterprises, by type of expenditure<br />

01. Train­movements, by type of vehicle and source of power<br />

02. Tractive vehicle movements, by type of vehicle and source of power<br />

03. Hauled vehicle movements, by source of power<br />

04. Hauled vehicle­kilometres, by type of hauled vehicle<br />

05. Hauled vehicles movements ­ seat kilometres offered<br />

06. Hauled vehicles movements ­ Tkm offered<br />

01A. Passenger transport by national/international transport (1000 passengers)<br />

01B. Passenger transport by class (1000 passengers)<br />

02A. Passenger transport by national/international transport (Mio pkm)<br />

02B. Passenger transport by class (Mio pkm)<br />

03. International passenger transport by country of arrival in the <strong>report</strong>ing country and by<br />

class<br />

04. International passenger transport by country of departure from <strong>report</strong>ing country and<br />

by class<br />

05A. Accompagnied passenger car transport, by type of transport (passenger cars)<br />

05B. Accompagnied passenger car transport, by type of transport (number of passengers)<br />

ragoildg 01.International annual transport by link with loading country and by group of goods (1000<br />

T, Mio Tkm)<br />

ragoiulg 02. International annual transport by link with unloading country and by group of goods<br />

(1000 T, Mio Tkm)<br />

ragoild 03. International monthly transport by link with loading country (1000 T)<br />

Document2<br />

Document2<br />

27 May 2004<br />

170


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

road<br />

Collection definition Collection heading table name table contents description<br />

A.VII. Accidents<br />

B.I. Infrastructure<br />

B.II1. Transport equipement<br />

­ Stock of vehicles<br />

raaccidt<br />

roinfras<br />

roeqstok<br />

ragoiul 04. International monthly transport by link with unloading country (1000 T)<br />

ragondcg<br />

05. National annual transport by distance class and group of goods (1000 T, Mio Tkm)<br />

ragon 06. National monthly transport (1000 T)<br />

ragorldg 07. National annual transport by loading regions and by group of goods (1000 T)<br />

ragorulg 08. National annual transport by unloading regions and by group of goods (1000 T)<br />

ragotg<br />

ragotlk<br />

ragoalrr<br />

raacoa<br />

raacoat<br />

raacota<br />

raacot<br />

roinlgmw<br />

roinlger<br />

roinlgor<br />

rostmopd<br />

rostmotr<br />

rostpacr<br />

rostpaca<br />

rostpaco<br />

rostbus<br />

rostbuss<br />

rostbusa<br />

rostbuso<br />

rosttram<br />

rostlowc<br />

rostlown<br />

09. Transit annual transport by group of goods (1000 T, Mio Tkm)<br />

10. Transit annual transport by link (1000 T, Mio Tkm)<br />

11. National, international and transit annual transport by container and road/rail (number,<br />

1000 T)<br />

01. Number of victims by type of injury<br />

02. Number of victims by origin of accident<br />

03. Accidents by type of accident<br />

04. Dangerous goods transport accidents<br />

1. Length of motorways<br />

2. Length of e­roads<br />

3. Length of other roads by category of roads<br />

1. Mopeds<br />

2. Motorcycles, by power of vehicles<br />

3A. Passenger cars, by motor energy<br />

3B. Passenger cars, by age<br />

3C. Passenger cars, by alternative motor energy and by power of vehicles<br />

4A. Motor coaches, buses and trolley buses, by motor energy<br />

4B. Motor coaches, buses and trolley buses, by number of seats<br />

4C. Motor coaches, buses and trolley buses, by age class<br />

4D. Motor coaches, buses and trolley buses, by alternative motor energy<br />

5. Trams<br />

6A. Lorries, by load capacity (1000t)<br />

6B. Lorries, by load capacity (number)<br />

Document2<br />

27 May 20043<br />

171


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

Collection definition Collection heading table name table contents description<br />

B.II2. Transport equipement<br />

­ New registration of vehicles<br />

roeqregi<br />

rostloec<br />

rostloeo<br />

rostvehv<br />

rostvehc<br />

rostveha<br />

roststct<br />

roststcn<br />

rosttct<br />

rosttcn<br />

roremotr<br />

rorevehe<br />

roreveho<br />

rorebus<br />

roreloc<br />

rorevkt<br />

rorevkn<br />

rorestct<br />

rorestcn<br />

roretct<br />

roretcn<br />

rorellc<br />

rorelme<br />

6C. Lorries, by type of motor energy and load capacity<br />

6D. Lorries, by type of alternative motor energy and load capacity<br />

7A. Load capacity of lorries, semi­trailers and trailers, by kind of transport (1000t)<br />

7B. Lorries, road tractors, semi­trailers and trailers, by kind of transport (number)<br />

7C. Lorries and road tractors, by age (number)<br />

8A. Semi­trailers, by load capacity (1000t)<br />

8B. Semi­trailers, by load capacity (number)<br />

9A. Trailers, by load capacity (1000t)<br />

9B. Trailers, by load capacity (number)<br />

1. New registrations of motorcycles<br />

2A. New registrations of passenger cars, motor coaches, buses and trolley buses, by type<br />

of vehicle and motor energy<br />

2B. New registrations of passenger cars, motor coaches, buses and trolley buses, by type<br />

of vehicle and alternative motor energy<br />

3. New registrations of motor coaches, buses and trolley buses, by seat capacity<br />

4. New registrations of lorries, by load capacity (1000T)<br />

5A. New registrations of lorries, semi­trailers and trailers, by kind of transport and load<br />

capacity (1000T)<br />

5B. New registrations of lorries, road tractors, semi­trailers and trailers, by kind of transport<br />

(number)<br />

6A. New registrations of semi­trailers, by load capacity (1000T)<br />

6B. New registrations of semi­trailers, by load capacity (number)<br />

7A. New registrations of trailers, by load capacity (1000t)<br />

7B. New registrations of trailers, by load capacity (number)<br />

8A. New registrations of lorries, by load capacity (number)<br />

8B. New registrations of lorries, by motor energy and load capacity (+/­ 1500kg) (number)<br />

rorelmo 8C. New registrations of lorries, by alternative motor energy and load capacity (+/­ 1500<br />

kg) (number)<br />

B.III. Enterprises, economic roeconom roecente 1A. Goods transport enterprises, by number of employees<br />

Document2<br />

Document2<br />

27 May 2004<br />

172


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

Collection definition Collection heading table name table contents description<br />

performances and<br />

employment<br />

roecentv<br />

roecnemp<br />

roecnems<br />

roecexpe<br />

1B. Goods transport enterprises, by number of vehicles<br />

2A. Employment in goods transport enterprises<br />

2B. Employment in goods transport enterprises, by sex<br />

3. Investment and maintenance expenditure, by nature of expenditure<br />

B.IV. Traffic rotrafic rotrmveh 1. Motor vehicle movements on national territory, by vehicles registration<br />

B.V1. Transport<br />

measurement ­ passengers<br />

B.V3. Transport<br />

measurement ­ goods<br />

romepass ropantv 1. Passenger transport on national territory, by type of vehicles registered in the <strong>report</strong>ing<br />

country<br />

rogoildt<br />

01. International annual transport by link with loading country, by group of goods and type<br />

of carriage (1000 T)<br />

rogoildm 02. International annual transport by link with loading country by type of carriage (Mio<br />

Tkm)<br />

rogoiult<br />

03. International annual transport by link with unloading country, by group of goods and<br />

type of carriage (1000 T)<br />

rogoiulm 04. International annual transport by link with unloading country by type of carriage (Mio<br />

Tkm)<br />

rogoild 05. International quarterly transport by link with loading country and type of carriage (1000<br />

T, Mio Tkm)<br />

romegood<br />

rogoiul<br />

06. International quarterly transport by link with unloading country and type of carriage<br />

(1000 T, Mio Tkm)<br />

rogong<br />

07. National annual transport by group of goods and type of carriage (1000 T, Mio Tkm)<br />

rogondct 08. National annual transport by distance class, type of carriage and group of goods (1000<br />

T)<br />

rogondcm 09. National annual transport by distance class and type of carriage (Mio Tkm)<br />

rogorldg 10. National annual transport by regions of loading and by group of goods (1000 T)<br />

rogorulg 11. National annual transport by regions of unloading and by group of goods (1000 T)<br />

rogon<br />

12. National quarterly transport by type of carriage (1000 T, Mio Tkm)<br />

rogogbg2 13. Cross trade transport annual by link, group of goods and type of carriage (1000 T)<br />

rogogbg1<br />

14. Cross trade transport quarterly by type of carriage (1000T, Mio Tkm)<br />

B.V4. Transport romecabo rocahaul Cabotage by hauliers from each <strong>report</strong>ing country<br />

Document2<br />

27 May 20043<br />

173


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL – FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

tranlink<br />

Collection definition Collection heading table name table contents description<br />

measurement ­ goods ­<br />

cabotage<br />

B.VII. Accidents<br />

regitran Regional transport<br />

statistics<br />

roaccidt<br />

regitran<br />

rocacoun<br />

roacka<br />

roacmt<br />

roacas<br />

roacta<br />

reavgocc<br />

reavgf98<br />

reavgu98<br />

reavpacc<br />

reavpf98<br />

reavpu98<br />

remagocc<br />

remagf98<br />

remagu98<br />

remapacc<br />

remapf98<br />

remapu98<br />

reroacci<br />

reroaccc<br />

rerotruc<br />

reroequi<br />

reroeqcc<br />

reinlinf<br />

reinlicc<br />

Cabotage transport by country in which cabotage takes place<br />

1. Injury accidents during the year, by kind of area<br />

2. Casualties in injury accidents (killed and injured), by means of transport<br />

3. Casualties in injury accidents (killed and injured), by age and sex<br />

4. Casualties in injury accidents (killed and injured), by means of transport and by age<br />

(detailled up to 20 years)<br />

Air transport of freight ­ Candidate Countries<br />

Air transport of freight from 1998 onwards (new <strong>methodology</strong>)<br />

Air transport of freight until 1998 (old <strong>methodology</strong>)<br />

Air transport of passengers ­ Candidate Countries<br />

Air transport of passengers from 1998 onwards (new <strong>methodology</strong>)<br />

Air transport of passengers until 1998 (old <strong>methodology</strong>)<br />

Maritime transport of freight ­ Candidate Countries<br />

Maritime transport of freight from 1998 onwards (new <strong>methodology</strong>)<br />

Maritime transport of freight until 1998 (old <strong>methodology</strong>)<br />

Maritime transport of passengers ­ Candidate Countries<br />

Maritime transport of passengers from 1998 onwards (new <strong>methodology</strong>)<br />

Maritime transport of passengers until 1998 (old <strong>methodology</strong>)<br />

Road safety<br />

Road safety ­ Candidate Countries<br />

Transport data for medtrans med_avia Aviation statistics<br />

Road transport of goods ­ Journeys made by vehicles<br />

Road transport, stock of vehicles by category<br />

Road transport, stock of vehicles by category ­ Candidate Countries<br />

Road, rail and navigable inland waterway networks<br />

Road, rail and waterway networks ­ Candidate Countries<br />

Document2<br />

Document2<br />

27 May 2004<br />

174


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Domain<br />

heading<br />

tranover<br />

Collection definition Collection heading table name table contents description<br />

Mediterranean countries<br />

Ad hoc tables used in<br />

Eurostat Yearbook<br />

Transport Euro­indicators<br />

(EUROZONE and EU15)<br />

tran_eyb<br />

transif<br />

treurind<br />

med_mari<br />

med_pipe<br />

med_rail<br />

med_roac<br />

med_road<br />

5a1fr<br />

5a1fv<br />

pass_tr<br />

good_tr<br />

5h6st<br />

5a1fo<br />

5a1fq<br />

5a1fk<br />

5h5st<br />

5h4st<br />

5a1fu<br />

5h7st<br />

5a1fh<br />

infra<br />

length<br />

5s1tr<br />

5a1fp<br />

sif79909<br />

Maritime statistics<br />

Pipelines<br />

Rail statistics<br />

Road accidents<br />

Road infrastructure and equipment<br />

Air transport of goods. 1 000 t<br />

Air transport of passengers. Millions<br />

Car, bus and rail transport of passengers (Mio PKM)<br />

Goods transport by road, by rail, by inland waterways and by oil pipelines (Mio TKM)<br />

Goods transport by road. Million tonne­km<br />

Inland goods transport. 1 000 million tonne­km. EEA and Switzerland<br />

Million tonne­km Sea transport of goods. Million t<br />

Passenger cars per 1 000 inhabitants<br />

Passenger cars per 1 000 inhabitants<br />

Passenger cars. 1 000s<br />

Passenger transport. 1 000 million passenger­km. EEA and Switzerland<br />

Persons killed in road accidents<br />

Total inland transport per mode: EEA and Switzerland<br />

Total length of motorways and railways lines in km<br />

Total length of motorways and railways lines in km (EU15 and Candidate countries)<br />

Transport growth. EU­15(I90=100)<br />

Worldwide commercial space launches<br />

Statistics in focus 09/1999 : Long distance passenger travel<br />

sif70202 Trends in road freight transport ­ 1990­1999<br />

treugoqt 1. International and national transport of goods ­ Quarterly (1000 T)<br />

treugoat 2. International and national transport of goods ­ Annual (1000 T)<br />

treupaan 3. International and national transport of passengers by plane and boat ­ Annual (1000)<br />

Document2<br />

27 May 20043<br />

175


ANNEX F:<br />

OVERVIEW OF DATA SOURCES<br />

DATA COLLECTION UP TO TEST<br />

OF METHODOLOGY


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER<br />

MANUAL – FREIGHT TRANSPORT DEMAND<br />

Transnational datasources<br />

Source Description Status<br />

Eurostat COMEXT trade by mode (quarterly data 1997 – 2002)<br />

NewCronos<br />

International road O/D matrix on regional level (estimate)<br />

Partly received<br />

Received<br />

Received<br />

UN UN trade by commodity on country level (value and weight (partly)) Received<br />

UN­ECE UNECE common database; quarterly trade data Eastern Europe In discussion<br />

WTO Trade on country level In discussion<br />

UIRR Intermodal statistics on country level Received<br />

ICF Intermodal statistics on country level Received<br />

UIC Rail transport statistics (intermodal) In discussion<br />

Lloyds Data about movements of ships between ports (costs 28.000) In negotiation<br />

Piers Detailed maritime (container) data with route information In negotiation<br />

CAFT Cross Alpine Freight Transport Received<br />

MDS Data about bulk and unitised transport on country level Received<br />

ISL Data about containerised transport flows in Europe Not available<br />

National datasources Western Europe<br />

Source Description Status<br />

SITRAM (FR) National and int. regional transport data (incl. transhipment) In discussion<br />

NBB (B) International transport flows via Belgium ports Received<br />

B­RAIL (B) National and int. Regional transport data – rail Received<br />

NIS (B) National and int. Regional transport data – road and inl.ww Received<br />

Statec (L) ­ No contact<br />

CBS (NL) National and int. Regional transport data by mode and by comm. Received<br />

SBA (DE) National and int. Regional transport data – excl. Road Received<br />

KBA (DE) National and int. Regional transport data – Road transport Ordered<br />

SLA Hamburg (DE) International transport flows via Hamburg Received<br />

SLA Bremen (DE) International transport flows via Bremen Received<br />

ISTAT (IT)<br />

Regional trade data by transport mode and commodity<br />

Maritime data by port<br />

Received<br />

Received<br />

DFT (UK) Domestic road transport / rail data / maritime data Received<br />

MDS (UK) Regional data UK by port, by mode and by commodity Received<br />

CSO (IE) Regional road transport, aggregate maritime data Received<br />

DST (DK) Regional road transport, aggregate data other modes Received<br />

Greece Statistics Aggregate transport data Received<br />

INE (PT) Regional road transport, aggregate data other modes Received<br />

MoT (ES) Regional information road transport Received<br />

AEAT (ES) Regional trade by mode, by comm. And by country Received<br />

SSB (NO) Trade by mode of transport, domestic road (regional) Received<br />

SCB (SE) Regional road transport –nat.and int.; aggr data other modes Received<br />

SIKA (SE) Rail data Not available<br />

Statistics Finland Domestic and international road transport / rail / detailed maritime Partly received<br />

BFS (CH) National and international regional road transport data (1998)<br />

Aggregate data other modes<br />

Received<br />

In discussion<br />

Satistics Austria National and int. regional transport by mode and comm. Received<br />

National datasources applicant countries<br />

Document2<br />

27 May 20043<br />

179


<strong>D5</strong> <strong>Annex</strong> <strong>WP</strong> 3: DATABASE METHODOLOGY AND DATABASE USER MANUAL –<br />

FREIGHT TRANSPORT DEMAND<br />

Source Description Status<br />

PMI/MoI (PL) National and int. Transport by mode and by comm. Received<br />

Viapont (CZ) Aggregate data by mode and comm. For int. transport Received<br />

MoTC (CZ) ­ In discussion<br />

Viapont (SK) Aggregate tables road, rail, inland waterways Received<br />

Slovak statistics Trade and transport data Received<br />

KTI (HU) International regional transport by commodity for road and rail Received<br />

KSH (HU) ­ In discussion<br />

TRI (EE) Aggregate data Received<br />

Estonia statistics ­ In discussion<br />

CSB (LV) ­ In discussion<br />

Mot / STD (LT) ­ In discussion<br />

Slovenia statistics Aggregate data by mode and by commodity In discussion<br />

Malta statistics ­ No reaction<br />

Cyprus statistics ­ No reaction<br />

180<br />

Document2<br />

27 May 2004

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!