1

Analysis of Production Systems (4DC10)

Introduction

Michel Reniers

Department of Mechanical Engineering

3
Who am I?

Associate professor in Control Systems Technology section

• Coordinator electives package High-Tech Systems Design with Model-based

Systems Engineering (4TC00)
• Analysis of Production Systems (4DC10 & 4CC40)
• Control of Manufacturing Systems (4CB40)
• Supervisory Control of Cyber-Physical Systems (4SC080)
• BFPs and MFPs

Research interest:
• Theory and applications of Discrete Event Systems
• Model-based engineering of supervisory controllers
• Simulation-based performance analysis of manufacturing systems
4
Motivation

Importance of Dutch manufacturing industry:

• constitutes about 17% of the Dutch gross national product
• employs about 15% of the Dutch work force

Because of network-centric communication, digitization, robotization and intelligent

manufacturing technologies, the manufacturing industry is now rapidly changing. The
future of Dutch manufacturing industry depends on how well manufacturing
managers and engineers are able to exploit newest developments in ICT and
manufacturing technology:
• Smart Industry or Fourth industrial revolution
• Internet of Things: Products, machines, factories, warehouses and customers can
exchange information
• Production becomes personalized!
5
Objectives

Challenge is to optimally design and control the manufacturing networks of the future!

Develop understanding of the behavior and basic principles of manufacturing

operations

Understanding is critical to manufacturing managers and engineers trying to:

• develop and control new systems
• improve existing systems

6
Topics

Probability models Elementary concepts in Probability Theory, including:

• probability models
• conditional probabilities
• random variables
• expectation and variance

Manufacturing models Manufacturing Science explaining:

• basic models and principles in manufacturing operations
• focus on material flow and key performance indicators

Simulation models Vehicle is the python simulation tool

The art of modeling is in the selection of the right model for a given situation
7
Organization – Classes

Lectures Tuesday morning, hour 1-2 (8.45-10.30)

Guided Self Study Tuesday morning, hour 3-4 (10.45-12.30)

Online support Canvas discussions & online meetings (on request)

Team

• Martijn Goorden • Ilse Mateman

• Michel Reniers • Willemijn Markus
• Niek Gorissen
• Ignacio Becerra Osorio

8
Tests

Group assignment: weight 30%

• Available!
• Due: 31 January 2025
• Group size: 3 (register groups in Canvas ASAP)
• Topic: Python simulation

Written examination: weight 70%

• Date: 24 January 2025, 9.00-12:00
• Topic: Probability theory & Manufacturing models

9
Course material

• Lecture presentations available in Canvas

• Lecture notes 4DC10 available in Canvas

• Each group gets its own GitLab repository, which includes:
• Instruction on installation and use of tool

• Tutorial in Jupyter notebook

• Assignment in Jupyter notebook

• Models for the examples in the lecture notes and slides

Additional background material

• Probability Theory:
Henk Tijms, Understanding Probability, 3rd ed, Cambridge, 2012.
David Morin, Probability - For the Enthusiastic Beginner, 2016.

• Manufacturing Science:
Wallace J. Hopp, Mark L. Spearman, Factory Physics, McGraw-Hill, 2008.
Rajan Suri, Quick Response Manufacturing, 1998.
10
Getting started with the Python simulation tool

Simulation is done using a simulation library for Python

• You will be programming in Python. The tutorial contains some instructions on
how to use Python.
• We expect you to have some experience with programming (such as matlab).
• However, the tutorial also contains some links to information on some of the
basics of Python.
All Python course material is in Jupyter Notebook.
• Jupyter notebook is an interactive environment which contains both text and
programming code.
• This way, you can learn about the code, and immediately run the code to find out
how it works.
• The assignment is also in such a Jupyter Notebook.
11
KIVA robots: Underrider AGV system
Kiva robot Operating in warehouse

Benefits are high throughput capability, flexibility and scalability, but how to design a
KIVA system?
• Length-to-width ratio of storage area?
• Storage strategy of the racks?
• How many pick locations? And where?
• How many robots? Routing of robots?
12
Storage and retrieval cycle of robots
Layout with 12 aisles, 14 cross-aisles, 5 pick stations Storage and retrieval cycle of robots
N circulating robots

Storage area

Buffer
Pick station

• Average storage and retrieval time is 15 minutes Fig. 11. Variants with different length-to-width ratios.

• Average pick time at pick station is 3 minutes

of the robots and the workstation from the analytical method is analytical method and the simulation. However, the differences are

•
nearly the same as for the simulation. The mean length of the ex- relatively small for the average order cycle time. The estimates of

Question: What is minimal number of robots to reach throughput capacity of 15

ternal order queue does differ between the analytical method and the analytical method typically stay below 10 percent of the esti-
the simulation and this affects the estimates of the order cycle mates of the simulation, except for high arrival rate when R = 2. As
time. The average order cycle time depends on the mean length is evident from the tables, using zones lowers the robot utilization
racks per hour per picker?
of the external order queue and therefore also differs between the and order cycle time.

• First order intuition: more robots, higher throughput

• What about second order intuition: How much do you gain by adding additional
13 robots?
Python simulation model of storage and retrieval cycle of robots

def model():
N = 1; la = 4.0; mu = 20.0

env = Environment()
a = Channel(env)
b = Channel(env)
c = Channel(env)
G = Generator(env, a, N)
Ss = [Storage(env, a, b, la) for j in range(N)]
B = Buffer(env, b, c)
P = Pick(env, c, a, mu, 10000)

env.run()

14
Simulation output

20
15
Throughput (racks/hour)

10
5
0

2 4 6 8 10

nr of robots

• Answer: At least 6 robots are required per pick station

15
 Queueing formula for throughput

1 µ N

!
Throughput = µ 1 − N! λ
1 µ i
PN

i=0 i! λ
with
• number of robots N
• average pick time 1
µ

• average storage and retrieval time of robot 1

λ

Note: This is one of Erlang’s famous formulas, the founding

father of Queueing Theory

16