presentation - Eurandom

12
STW-project Effective Process Time:
an overview
Ivo Adan, Onno Boxma, Pascal Etman, Ad Kock,
Erjen Lefeber, Koos Rooda, Marcel van Vuuren
Department of Mechanical Engineering &
Department of Mathematics and Computer Science
Eindhoven University of Technology
Eurandom workshop June 19-20th 2006 Eindhoven
◭◭◭◮◮◮ 1/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Contents
• Introduction
• Effective process time concept
• STW project
• Progress
• Application
◭◭◭◮◮◮ 2/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Introduction
◭◭◭◮◮◮ 3/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Motivation
Performance analysis of manufacturing systems.
◭◭◭◮◮◮ 4/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Motivation
Performance analysis of manufacturing systems
◭◭◭◮◮◮ 5/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Models
of manufacturing systems
Performance measures:
• Throughput δ
• Flow time ϕ
• Work-in-process w
Shopfloor realities:
• Machine downs, repairs, setup, product mix,
operators, batching, product recipes, priority lots, . . .
Modeling approaches:
• Queueing (network) analysis (exact/approximation)
• Discrete-event simulation
◭◭◭◮◮◮ 6/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Models
of manufacturing systems
Queueing (network) analysis (exact/approximation):
• Computationally efficient: ideal for what-if and
optimization studies
• Limited applicability due to restrictive assumptions
regarding shopfloor realities
Discrete-event simulation:
• Shopfloor realities may be modeled in detail
• Computationally intensive
Model parameters:
• Can we obtain all required data to set the model
parameters?
◭◭◭◮◮◮ 7/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12EPT
concept
◭◭◭◮◮◮ 8/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Example:
a production flow line
Station 1
Flowline characterisics:
Station 2
Station 3
• Infinite buffers, single-server stations, single-lot
machines.
Shopfloor realities:
• Process mix, setup-times, recipes, operator
Station 4
availability, machine downs, repair,rework, hot-lots,
sequencing, scheduling, shopfloor control, ...
◭◭◭◮◮◮ 9/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Zooming
in: single-server workstation
arrival
queueing
Reality: Natural processing
processing
departure
◭◭◭◮◮◮ 10/25
M
Random failures (preemptive): e.g. machine downs
Planned process delays (non-preemptive): e.g. setups
Aggregation: Effective process time
Queueing model: CT = c2 a +c 2 e
2
[Hopp, Spearman, 2000]
(EPT parameters te and c 2 e may be estimated
by adding natural processing and preemptive
and nonpreemptive outages)
u
1−u te + te, u = te
ta

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
EPT 1
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
EPT 1 EPT 2
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
EPT 1 EPT 2 EPT 3
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
EPT 1 EPT 2 EPT 3 EPT 4
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Idea:
Measure EPTs from arrival/departures
0 5 10 15 20 25 30
EPT 1 EPT 2 EPT 3 EPT 4 EPT 5
EPTi = di − max(ai, di−1)
Legend
Setup
Processing
Queueing
Waiting for operator
Machine breakdown
◭◭◭◮◮◮ 11/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Generalization
of the idea
Event data
EPT realisations
EPT parameters
Manufacturing
system
EPT−algorithm
Distribution fit
Throughput
Flow time
Aggregate Throughput
model Flow time
Observed
reject accept
Validation
Predicted
Effective process time modeling framework
Bottleneck analysis
System optimisation
◭◭◭◮◮◮ 12/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Research
project
◭◭◭◮◮◮ 13/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Research
objective and approach
Goal:
• To build simple yet accurate models of manufacturing
networks using operational factory data without the
need to characterize all contributing disturbances and
shopfloor realities.
Approach:
• Effective process time paradigm in aggregate
modeling and parameter identification
• Efficient queueing network approximations through
iterative algorithms based on decomposition
◭◭◭◮◮◮ 14/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Challenges
Machine:
• Single-lot, batching, conveyor, assembly, other types,
...
Workstation (multiple parallel servers sharing a buffer):
• Dispatching, recipes, unequal machines, lot
overtaking, ...
Network:
• Blocking (finite buffers), assembly lines, re-entrant
flow lines, general queueing network, production
control, ...
◭◭◭◮◮◮ 15/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Progress
◭◭◭◮◮◮ 16/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Single-lot
machine workstation
EPT parameter estimation
eqm 2 eqm 1
0 1 2 3 4 5 6 7 8 9
lot 1
ept 1
lot 2
Application
6
4
2
0
lot 3
ept 2
ept 3
eqm 2 eqm 1
0 1 2 3 4 5 6 7 8 9
◭◭◭◮◮◮ 17/25
lot 1
ept 1
9.2
A B C D E F G H I J K L M N
lot 2
ept 2
2
1
0
lot 3
ept 3
3.95
[Jacobs, Etman, Van Campen, Rooda, 2001, 2003]
A B C D E F G H I J K L M N

Contents
Introduction
EPT concept
Research project
Progress
Application
12Batch
machine workstation
EPT parameter estimation
0 1 2 3 4 5
lot 1
lot 2
batch 1
ept 1
lot 3
Application
t [h]
8.0
6.0
4.0
2.0
0.0
lot 4
batch 2
2
0 1 2 3 4 5
lot 1
batch 1
ept 1
◭◭◭◮◮◮ 18/25
lot 2
lot 3
batch 2
ept 2
[-]
A B C D E F G H I
Equipment family
te
t0 [-]
[Jacobs, Van Bakel, Etman, Rooda, 2006]
0.4
0.3
0.2
0.1
0.0
1.65
2 [-]
ce
2 [-]
c0
A B C D E F G H I
Equipment family

Contents
Introduction
EPT concept
Research project
Progress
Application
12Flow
line subject to blocking
EPT parameter estimation
¡£¢¥¤
¡£¢§¦
¨©
¢
PA0=AA0
ept 0 ept 1
PA1
PD0=AD0=AA1
Queueing network approximation (talk Van Vuuren)
!
5 > 5 > 5 > ! 5 !
◭◭◭◮◮◮ 19/25
PD1
* * * ! !
) > ,
) > ,
) ! > ! , !
[Kock, Wullems, Etman, Adan, Rooda, 2005]
[van Vuuren, Adan, Resing-Sassen, 2005]
AD1
time

Contents
Introduction
EPT concept
Research project
Progress
Application
12Assembly
machine
EPT parameter estimation
lot 0
lot 1
lot 2
lot 3
1SLMB-a
events
ept0 ept1 ept2 ept3
PD0 PD1 PD2
AD0 AD1
AD2
AA2.0
AA3.0
AA4.0
AA1.1
0 2 4 6 8
◭◭◭◮◮◮ 20/25
AA2.1
Queueing approximation (talk Van Vuuren)
AA3.1
time

Contents
Introduction
EPT concept
Research project
Progress
Application
12Assembly
machine
EPT parameter estimation
lot 0
lot 1
lot 2
lot 3
1SLMB-a
events
ept0 ept1 ept2 ept3
PD0 PD1 PD2
AD0 AD1
AD2
AA2.0
AA3.0
AA4.0
AA1.1
AA2.1
0 2 4 6 8
Queueing approximation (talk Van Vuuren)
◭◭◭◮◮◮ 21/25
AA3.1
[Vijfvinkel, 2005]
[van Vuuren, Adan, 2005,2006]
5
time
5
5
555
!
5 "

Contents
Introduction
EPT concept
Research project
Progress
Application
12Lithography
cell
Aggregate modeling and parameter estimation
Head-tail model (talk Van der Eerden)
Arrivals Buffer Head
Tail
Exit
Inside detailed, outside aggregate (talk Kock)
Aggregate modelling: Detailed modelling:
Outside Machine Inside Machine
r s e
S
a
B
b
D
c
L
d
T
[Van der Eerden, Saenger, Walbrick, Niesing, Schuurhuis, 2006]
[Kock, Etman, Rooda, 2006]
◭◭◭◮◮◮ 22/25
E
f

Contents
Introduction
EPT concept
Research project
Progress
Application
12Application
◭◭◭◮◮◮ 23/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Application
Industry:
• Automotive industry (talk Nijsse)
• Semiconductor industry (talk Van der Eerden)
(talk Van Campen)
• Consultancy (talk Resing/v. Doremalen)
• ...
Research:
• Control of manufacturing networks (talk Lefeber)
• In work by our students and colleagues
Related research and work by others:
• See e.g. talks on Tuesday ...
◭◭◭◮◮◮ 24/25

Contents
Introduction
EPT concept
Research project
Progress
Application
12Bibliography
J.H. Jacobs, L.F.P. Etman, E.J.J. van Campen and J.E. Rooda (2001): Quantifying Operational Time Variability: the Missing
Parameter for Cycle Time Reduction. In: 2001 IEEE/SEMI Advanced semiconductor manufacturing conference, April, Munich,
pp. 1-10.
J.H. Jacobs, L.F.P. Etman, E.J.J. van Campen and J.E. Rooda (2003): Characterization of operational time variability using
effective process time. IEEE Transactions on Semiconductor Manufacturing, vol. 16 (3), pp. 511-520.
M. van Vuuren, I.J.B.F Adan, S.A.E. Resing-Sassen (2003): Performance Analysis of Multi-Server Tandem Queues with Finite
Buffers. In: Fourth International Conference on “Analysis of Manufacturing Systems”. pp. 11-20.
A.A.A. Kock, F.J.J. Wullems, L.F.P. Etman, I.J.B.F. Adan and J.E. Rooda (2005): Performance Evaluation and Lumped Parameter
Modelling of Single Server Flowlines subject to Blocking: an Effective Process Time Approach. In: Fifth International
Conference on “Analysis of Manufacturing Systems-Production Management”. pp. 137-144.
(http://www.icsd.aegean.gr/aic2005/program.htm)
M. van Vuuren and I.J.B.F Adan (2005): Performance Analysis of Tandem Queues with Small Buffers. In: Proc. Fifth
International Conference on “Analysis of Manufacturing Systems-Production Management”. pp. 127-135.
(http://www.icsd.aegean.gr/aic2005/program.htm)
M. van Vuuren and I.J.B.F. Adan (2005): Performance analysis of assembly systems. In: Proc. Fifth International Conference
on Matrix Analytic Methods in Stochastic Models, June, Pisa.
M. van Vuuren, I.J.B.F Adan, and S.A.E. Resing-Sassen (2005): Performance Analysis of Multi-Server Tandem Queues with
Finite Buffers and Blocking. OR Spectrum, vol. 27, pp. 315-338.
M. van Vuuren and I.J.B.F Adan (2005): Approximation the ΣGI/ G/s Queue by using Aggregation and Matrix Analytic
Methods. Stochastic Models, vol. 21, pp. 767-784.
J.H. Jacobs, P.P. van Bakel, L.F.P. Etman and J.E. Rooda (2006): Quantifying variability of batching equipment using effective
process times. IEEE Transactions on Semiconductor Manufacturing, vol. 19, pp. 269-275.
A.A.A. Kock, L.F.P. Etman and J.E. Rooda (2006): Lumped Parameter Modelling of the Litho Cell In: Proc. INCOM 2006, May,
St. Etienne.
M. van Vuuren and I.J.B.F. Adan (2006): Performance analysis of assembly systems. In: Proc. Markov Anniversary Meeting,
June, Charleston.
M. van Vuuren and I.J.B.F Adan (2006): Approximating Multiple Arrival Streams by using Aggregation. Stochastic Models, to
appear.
◭◭◭◮◮◮ 25/25