Supply Chain planning at Philips Lighting Lumileds
Transcript of Supply Chain planning at Philips Lighting Lumileds
Supply chain planning at Philips Lighting Lumileds
How certain do we like to be?
A design and implementation of a stock control model to balance customer service and stock levels in an
end to end environment to improve product availability.
Author:
R. Hartevelt
Public
Master thesis project – Roy Hartevelt Page 2
How secure do we like to be?
A design and implementation of a stock control model to balance customer service and stock levels in an
end to end environment to improve product availability.
Master thesis project
Bergen op Zoom, April 14 2011
Name Roy Hartevelt
University Delft University of Technology
Faculty Technology, Policy and Management
Program Infrastructure systems & services
Section Transport and logistics
Company Philips
Supervisory committee:
Prof.dr. L. Tavasszy TU Delft
Drs. J.H.R. van Duin TU Delft
Drs. H.G. van der Voort TU Delft
Ir. J-E Talsma Philips
Ir. H. Rulkens Philips
Master thesis project – Roy Hartevelt Page 3
Management summary Improving service level at the lowest possible costs is and will always be one of the key objectives of
Philips Electronics. This research illustrates how a part of the Philips supply chain control is setup
/designed to support this objective. Within Philips, the business unit called Lumileds supplies LED’s to its
customers. One of the components used to manufacture LED’s is made at the component manufacturer
that is subject to this thesis.
The objective of improving service levels at the lowest possible cost can be enabled by supply chain
control. Of course superior service levels can be realized with excessive inventory levels. However that
ignores the objective of lowest costs, because inventory cost money. Therefore an optimal balance
between service level and inventory must be achieved. This balance depends on a number of different
drivers like lead-time, lead-time variability, manufacturing quality and demand pattern. This research
will define a model that generates advice to achieve the desired balance, taking into account all relevant
drivers.
Based on literature research and analysis of the current Supply chain performance between the
component supplier and the manufacturer an advice model is designed. This model generates specific
stock level advices but also generates insights in the Supply chain uncertainties and lead-times. Besides
the stock optimization part of the model, the model functions as a decision supporting tool for Supply
chain balancing and stock level considerations. To gain the maximum output of the model, intensive
cooperation between the component supplier and the manufacturer is a must.
The performed research answers on the main research question:
WHICH SUPPLY CHAIN PLANNING CONTROL IS NEEDED FOR A MOST SUITABLE STOCK SITUATION TO SECURE
THE SAFETY STOCK LEVELS BETWEEN A COMPONENT SUPPLIER AND LED ASSEMBLY & TEST
MANUFACTURER.
The advice models gives a product specific advice based on the different lead-times and the Supply chain
uncertainties including a specific demand pattern. The component supplier’s manufacturing unit
consists of a front-end and a back-end part of the line. The model calculates for the front-end of the
component supplier an advice based on a Kanban replenishment strategy. The back-end of the
component supplier is controlled via replenishment and managed by a Re-Order-Point (ROP) calculation.
The ROP and Kanban boundaries are set monthly, the replenishment orders are calculated weekly. As a
result of above structure the organization is able to measure the different parts of the Supply chain and
the overall Supply chain as well. The measurements about the sub parts of the Supply chain are a result
of the new control model. The models results in transparency and the opportunity to simulate the
impact of variance or lead-time reduction on stock levels. Based on the simulation results the model
demonstrates the improvement potential and proves that an equal service is possible with lower stock
levels. The influence of the component supplier and the LED assembly & test manufacturer can be used
to improve and focus on cycle time reduction and controlling the uncertainties in the Supply chain as
Master thesis project – Roy Hartevelt Page 4
much as possible. Finally, the models are the first step for further improvements between the
component supplier and the manufacturer.
The component supplier and the manufacturer both have a drive to improve their Supply chain controls
and aspire an increased level of cooperation. This is in line with literature; suggesting that Supply chain
transparency, information sharing and an intensive relationship has positive influences on Supply chain
performance. This intensive relationship is decisive for this research but also for future improvement of
Supply chain processes.
During the pilot phase (1 month) the models were used with 2 product families. Based on the
experiences of the field experts modifications were made before we extend the pilot with another
month. During the second pilot 4 extra product families were included. Based on the results out of the
second pilot the models were finalized and handover to operations ready. The last part of the hand-over
to operations was to imbed the models in the standard organization. Term-Of-References (TOR) was
made for creating meeting guidelines and meeting inputs & outputs /results. With the help of the TOR a
structure was set for having efficient and fruitful weekly and monthly alignment meetings. As a direct
effect of the transparent meeting structure thinking in improvement opportunities were embed in the
day to day business easily.
The need to focus more on integral supply chain planning and optimization instead of local optimization
plans has ascended the past few years. There are some appropriate ways to determine an optimal
solution for the optimization challenge. Research has indicated required actions to optimize the LED
supply chain. Without actions there will be too high inventory levels and yield losses with resulting
financial losses. All these different kinds of behavior provoke business losses and will harm continuity. In
case industrial consumers decide not to purchase, major sales loss is incurred.
The advice models (Re-Order-Point, Replenishment and Kanban) are based on specific risks (lead-time
and uncertainties). The models generate advice based on these risks for a desired service level. The
desired service level is expressed in the Z factor. The models give product specific advice for safety
stock, economical stock levels and quantities to produce based on the requested service level. The
outcomes of the model have an average safety stock of 1.85 week at the assembly and test
manufacturer for a service level of 95%. For a service level of 99% a safety stock of 2.6 weeks is required
with a total average stock of 3.1 weeks. The results of the advice model (Figure 1) indicate that a
product safety stock level can optimize to the most suitable stock positions for the LED supply chain at
Philips: meaning high customer service with the right underpinning stock levels in the chain (at the right
balanced cost levels). The benefits of the Supply chain control model are enormous. Without the model
the Lumileds business (at 12 Million pieces sales per week) needs 9 times more stock (see Figure 1).
Master thesis project – Roy Hartevelt Page 5
Figure 1 Results Supply Chain control model
Figure 1 shows the model calculation for the most optimal situation and the situation as it was without
the new control models. It shows the (safety) stock needed to cover the demand during lead-time
(vertical) and the sales per week in Million(s) (horizontal). Due to project effort the lead time of the
component supplier is optimized from 12 to 1.31 weeks and secondly, the supply variance is reduced
from 0.61 to 0.21 (in CV value).
The implemented supply chain control model has initially led to a 720 K Euro stock cost (Safety stock:
420 K Euro, Demand during Lead-time: 300 K Euro) reduction (based on 12 Million pieces sales per
week) with a design and implementation cost of 150K Euro.
-
5,000
10,000
15,000
20,000
25,000
30,000
35,000
1 2 4 6 8 10 12
Nm
br
pie
ces
(* 1
00
0)
Sales per week (M pieces)
Stocklevel effects control model
Dem LT. Old
Safety stock Old
Dem LT. New
Safety stock New
Master thesis project – Roy Hartevelt Page 6
Preface This Master’s thesis is the final project for receiving the Master of Science degree in System Engineering,
Policy Analysis and Management with the specialization Logistics at Delft University of Technology. It has
been carried out at Philips between March and August 2010.
I specially want to thank my supervisors at Philips, Hubert Rulkens and Jan-Edzard Talsma, for their
advice, support and encouragement.
I would also like to thank my supervisors at Delft University of Technology, Ron van Duin and Haiko van
der Voort for their valuable advice.
Finally I would like to direct my warmest thanks to my wife (Annette) and our three little ladies’ (Elmyra,
Elena and Evely) for always supporting me.
Thank you!
Bergen op Zoom April 14, 2011
Roy Hartevelt
Master thesis project – Roy Hartevelt Page 7
Table of contents
MANAGEMENT SUMMARY .................................................................................................................................... 3
PREFACE ................................................................................................................................................................ 6
TABLE OF CONTENTS .............................................................................................................................................. 7
INDEX OF TABLES AND FIGURES ............................................................................................................................. 9
FIGURES ....................................................................................................................................................................... 9
TABLES ....................................................................................................................................................................... 10
1.0 INTRODUCTION .............................................................................................................................................. 11
1.1 BUSINESS CHALLENGE .............................................................................................................................................. 11
1.2 RESEARCH PROJECT ................................................................................................................................................. 11
2 ANALYSIS OF SUPPLY CHAIN CONTROL ............................................................................................................. 19
2.1 INTRODUCTION TO SUPPLY CHAIN CONTROL ................................................................................................................. 19
2.2 WHAT IS SUPPLY CHAIN CONTROL? ............................................................................................................................ 19
2.3 EFFECTS OF SUPPLY CHAIN CONTROL ON THE MANUFACTURER AND THE COMPONENT SUPPLIER .............................................. 21
2.4 ROOT CAUSES AND INFLUENCES ON SUPPLY CHAIN CONTROL ........................................................................................... 21
2.5 SUPPLY CHAIN CONTROL MODELS ............................................................................................................................... 23
2.6 CONCLUSION SUPPLY CHAIN CONTROL ANALYSIS ........................................................................................................... 25
3 ANALYSIS OF THE COMPONENT SUPPLIER AND THE LED ASSEMBLY MANUFACTURER ..................................... 26
3.1 SUPPLY CHAIN MANAGEMENT ................................................................................................................................... 26
3.2 LED ASSEMBLY AND TEST MANUFACTURER .................................................................................................................. 29
3.3 STAKEHOLDER ANALYSIS ........................................................................................................................................... 29
3.4 CONCLUSION ANALYSIS ............................................................................................................................................ 36
4 ANALYSIS OF COMPONENT THROUGH-PUT-TIME ............................................................................................. 39
4.1 INTRODUCTION ...................................................................................................................................................... 39
4.2 FRONT-END ........................................................................................................................................................... 42
4.3 BACK-END ............................................................................................................................................................. 43
4.4 SUPERMARKET STOCK LEVEL CALCULATION ................................................................................................................... 44
5 DESIGN SETUP ................................................................................................................................................... 47
5.1 INTRODUCTION ...................................................................................................................................................... 47
5.2 PERFORMANCE MEASUREMENTS ............................................................................................................................... 47
5.3 UNCERTAINTY ........................................................................................................................................................ 47
5.4 DATA ................................................................................................................................................................... 51
5.5 CONCLUSIONS ........................................................................................................................................................ 52
6 MODEL DESIGN ................................................................................................................................................. 53
6.1 INTRODUCTION MODEL DESIGN ................................................................................................................................. 53
6.2 EVALUATION DESIGN OBJECTIVE................................................................................................................................. 53
6.3 MODEL DESIGN BASED ON THE THEORY ....................................................................................................................... 53
Master thesis project – Roy Hartevelt Page 8
6.4 PRACTICAL DESIGN AND MODEL RESULT ...................................................................................................................... 53
6.5 RESULTS ............................................................................................................................................................... 58
6.6 CONCLUSION MODEL DESIGN .................................................................................................................................... 59
7 SUMMARY DESIGN AND MODELING PHASE ...................................................................................................... 61
8 FROM MODEL TO PRACTICE .............................................................................................................................. 64
8.1 INTRODUCTION ...................................................................................................................................................... 64
8.2 GENERAL USABILITY OF THE MODEL ............................................................................................................................ 64
8.3 USING THE MODEL FOR A PILOT ................................................................................................................................. 65
8.4 SUMMARY FROM MODEL TO PRACTICE ........................................................................................................................ 67
9 CONCLUSION AND RECOMMENDATIONS .......................................................................................................... 68
9.1 INTRODUCTION ...................................................................................................................................................... 68
9.2 CONCLUSIONS OF THE RESEARCH ............................................................................................................................... 68
9.3 RECOMMENDATION FOR FURTHER RESEARCH ............................................................................................................... 72
9.4 REFLECTION ........................................................................................................................................................... 73
REFERENCES ......................................................................................................................................................... 74
APPENDICES......................................................................................................................................................... 77
APPENDIX A PRODUCT FAMILIES ..................................................................................................................................... 78
APPENDIX B RELATION BETWEEN STOCK LEVEL AND SERVICE LEVEL ......................................................................................... 79
APPENDIX C PHILIPS ANALYSIS ........................................................................................................................................ 80
APPENDIX D DETAILS SUB PROCESSES FRONT-END COMPONENT SUPPLIER ............................................................................... 82
APPENDIX E DETAILS SUB PROCESSES FRONT-END COMPONENT SUPPLIER ................................................................................ 83
APPENDIX F MODEL RESULTS KANBAN FRONT-END COMPONENT SUPPLIER .............................................................................. 84
APPENDIX G RACI MODEL SUPPLY CHAIN CONTROL MODEL .................................................................................................. 85
APPENDIX H IDEF0 LEVEL 2 ........................................................................................................................................... 86
APPENDIX I MONTHLY RE-ORDER-POINT PROCESS ............................................................................................................. 89
APPENDIX J WEEKLY REPLENISHMENT PROCESS .................................................................................................................. 95
APPENDIX K TUTORIAL SUPPLY CHAIN CONTROL MODELS ................................................................................................... 102
APPENDIX L GLOSSARY OF TERMS .................................................................................................................................. 130
Master thesis project – Roy Hartevelt Page 9
Index of tables and Figures
Figures Figure 1 Results Supply Chain control model ............................................................................................... 5
Figure 2 Potential LED market in (billion) US$ (as per 10-2010)................................................................. 11
Figure 3 research scope of the project ....................................................................................................... 13
Figure 4 design methodology (adapted from Herder and Stikkelmans, 2004)........................................... 16
Figure 5 Uncertainty in the supply chain (T. Davis, 1993) .......................................................................... 20
Figure 6 Inventory level with safety stock (S. Lutz e.a., 2003) .................................................................... 20
Figure 7 IDEF0 Supply Chain control model as-is situation ......................................................................... 23
Figure 8 Safety-stock calculation comparison ............................................................................................ 24
Figure 9 value chain LED production .......................................................................................................... 26
Figure 10 supply chain execution run book ................................................................................................ 27
Figure 11 performance current supply chain control model ...................................................................... 28
Figure 12 CV throughput time component supplier ................................................................................... 29
Figure 13 the LED worldwide supply chain organization ............................................................................ 30
Figure 14 the worldwide supply chain organization in functional areas .................................................... 31
Figure 15 stakeholders in the value chain .................................................................................................. 32
Figure 16 stakeholder diagram ................................................................................................................... 33
Figure 17 Simplified Supply chain of Lumileds ............................................................................................ 39
Figure 18 Distribution indentification throughput-time front-end ............................................................ 40
Figure 19 Front-end component supplier ................................................................................................... 40
Figure 20 Back-end component supplier .................................................................................................... 41
Figure 21 Throughput-time back-end component supplier ....................................................................... 41
Figure 22 Entry process steps assembly & test manufacturer ................................................................... 42
Figure 23 Median throughput time front-end ............................................................................................ 42
Figure 24 Median throughput time back-end ............................................................................................. 43
Figure 25 Pareto negative effects on throughput-time component supplier ............................................ 44
Figure 26 Z value development ................................................................................................................... 45
Figure 27 Causal model of uncertainty and customer performance .......................................................... 47
Figure 28 Concept advice model ................................................................................................................. 50
Figure 29 Planning of replenishment orders .............................................................................................. 51
Figure 30 Model design (IDEF0 level 0) ....................................................................................................... 54
Figure 31 Model design (IDEF0 level 1) ....................................................................................................... 55
Figure 32 ROP calculation model ................................................................................................................ 56
Figure 33 Weekly replenishment calculation ............................................................................................. 57
Figure 34 Planning rules and deploy control model ................................................................................... 59
Figure 35 ROP development financially ..................................................................................................... 60
Figure 36 Concept advice model ................................................................................................................. 69
Figure 37 Planning of replenishment orders .............................................................................................. 70
Master thesis project – Roy Hartevelt Page 10
Figure 38 stakeholder diagram ................................................................................................................... 71
Tables Table 1 requirements, criteria, constraints & design option stakeholders ................................................ 36
Table 3 Average stock level advice for data set .......................................................................................... 46
Table 2 Replenishment strategies (P. Suwanruji ao, 2005) ........................................................................ 49
Table 4 Results correlation analysis stock level and CLIP ........................................................................... 79
Master thesis project – Roy Hartevelt Page 11
1.0 Introduction
1.1 Business challenge LED (light emitting diode) is a fast growing product in many consumer and professional applications.
Main reasons for that can be found in the green image of the product, extended use in several
electronic devices and as replacement for current general lighting products. In more detail: the use of
LED is increasing in many (handheld) devices like the mobile phones, displays, remote controllers and TV
screens and brings a boost in industrial need of the new light source. Besides the ‘new’ market for the
LED product, the LED is the replacement for light sources in automotive lighting, aviation lighting and
traffic signals. All of these new demands will materialize in the coming decade.
Financially: the potential market for LEDs has expanded dramatically, from US$7 billion in 2009 to
US$10.7 billion in 2010, which is a growth rate unreachable by any other electronic product. Along with
growing LED brightness and falling prices, the share of LED in general lighting field is expected to be
increased greatly; the general lighting market is of huge potential with the market size reaching US$100
billion. Promisingly, the LED market can reach US$20.4 billion in 20121.
Figure 2 Potential LED market in (billion) US$ (as per 10-2010)
1.2 Research project
1.2.1 Research objective
The objective of this research is shaped by: internal and external business drivers, the business challenge
as described in par 1.1 and a business improvement program within Philips Lighting Lumileds.
Internal drivers are mainly cost driven. Focus areas are yield and utilization improvements, inventory
optimization and cycle time reduction. External drivers have a totally different scope and are shaped by
market circumstances. Market growth caused a ramp up in the industry. Secondly, the technology in the
1 Financial figures according the Global and China LED Industry Report, 2009-2010; ResearchInChina
0
10
20
30
2009 2010 2011 2012
in billion US$
year
Potential LED market
Master thesis project – Roy Hartevelt Page 12
area of color control is moving fast. Product releases due to color control improvements are done at a
frequent and regular base. Finally, introducing the LED in the automotive industry brought besides a
market increase, also new quality standards.
The business challenge is to accelerate the internal and external drivers faster in comparison with
competitors. Philips Lumileds is one of the top three players in the high class technology market in Ultra-
High Brightness LED. For saving the future market position a strong focus is at high quality standards and
cost savings projects as part of the daily operations and the business strategy. As a result of the business
focus the BU management team started up an improvement program. One of the assignments within
the improvement program was to execute an assessment about an introduction of a new product
family. As an outcome of the assessment three main issues were detected in the areas of:
1. Components manufacturing,
2. Logistics control,
3. LED assembly and test.
This research will focus at improvement area 2 – Logistics control at Philips Lumileds. This challenge can
be translated in a problem statement:
Problem statement:
In order to serve the global LED market in a most suitable way Lumileds must continuously improve
their business in a structured way. Lumileds has a clear focus how to deal with product and market
uncertainties and translated this focus in a business wide improvement program. Insights are needed
to balance service- with stock levels and to find an optimum in the utilization (shop floor scheduling)
of the LED assembly and test.
Common to all manufacturing companies, regardless of size, type of product or manufacturing process is
the need to control the flow of materials from suppliers, through manufacturing and distribution to the
customer (G.C. Stevens, 2007). Also in our business configuration the role of supply chain is crucial. This
research will identify which criteria are relevant to determine economical stock values and deliver a
model for stock level calculation for the midterm (monthly) and the short-term (weekly). Those advice
models generate stock advice on the one hand based on customer demand and it’s variance and on the
other hand the supply and it’s variance. For the midterm, those models can help to determine further
tactical decisions. The result of the models (midterm and short-term) can help to take decisions for
these challenging considerations. This research investigates some of these future considerations and
there will be provided recommendations how to use the advice in practice and if the chosen calculation
method will fit in an optimal way for each phase in the product-life-cycle.
The research focus is clearly on improving service level and optimizing stock levels. The model must be
used and understood by several business lines of Philips Lumileds. These business lines will have
different backgrounds, goals and work levels. Attention to this multi actor setting is important and
frequent contact and information sharing will contribute to this. Recognizing the multi actor setting for
Master thesis project – Roy Hartevelt Page 13
stock level adjustments is important. The components supplier and the LED assembly and test
manufacturer both have the intention to optimize stock levels in a structured and mutually agreed way.
The goal of the assembly and test manufacturer is creating a demand/supply advice for the component
manufacturer and gets insights on the supply chain risks from the models, at the other hand the
component supplier owns and controls their own production process and internal optimization. For
model validation the model will be tested by experts in the field from both actors.
Research objective:
Generate, implement and deploy supply chain planning controls for shop floor scheduling and material
replenishment for the LED products between a component supplier and LED assembly and test
manufacturer.
The problem statement and research objective make explicitly that the focus of the research is on stock
level optimization, utilization in relation with service levels.
1.2.2 Research boundaries
Philips practical influence stops at the customer’s factory or DC. The focus of the research is represented
in Figure 3 research scope of the project. Important to mention is that this research will not covers the
integral supply chain of Lumileds. The research areas are the relations and independencies between LED
assembly and test and a key component manufacturer with a focus on the internal drivers as mentioned
in par 1.2.1.
The external factors are regarded as stable for the short term. In the long term the boundaries can easily
change due to the LED market situation as we have today. This research focuses on the sales figures of
the assembly and test (promotions included). The temporary shifts in the demand with the disturbances
for the total chain should be covered in the solution design. When stocks and lead times are reduced but
losses do not outweigh the gains the advice is not successful. A balance must be found between stock
level investment and service level (Jammernegg and Reiner, 2007).
Figure 3 research scope of the project
LED assembly and test
Component manufacturer
Master thesis project – Roy Hartevelt Page 14
Information sharing can significantly improve the performance of the supply chain (Zhao, 2002).
Transparency and information sharing is necessary to gain a complete vision about the risk in the whole
supply chain and fit different parts into one consistent chain. Figure 3 shows the boundaries of the
research but also explicitly mentions the crucial transparency. The model not only gives the component
manufacturer production advice but also a convincing and informative role. Potential users must trust,
understand and be able to use the model, before the advice can be considered successful. Firstly the
interactions between the involved actors will be examined, to create trust and understanding. Frequent
contact with and clarification for the stakeholders is probably an outcome to handle this multi actor
complexity. Beside these frequently updates workshops can be a useful tool to create the required
comprehension and trust.
1.2.3 Research questions
To achieve the research objective stated in section 1.2.1 the following research questions are
formulated:
Research question
WHICH SUPPLY CHAIN PLANNING CONTROL IS NEEDED FOR AN MOST SUITABLE STOCK SITUATION TO SECURE THE
SAFETY STOCK LEVELS BETWEEN A COMPONENT SUPPLIER AND LED ASSEMBLY & TEST MANUFACTURER.
In order to find an answer on the main research question several sub-questions are formulated:
1. Which facts determine the most suitable supply chain planning method?
Method: Literature research of scientific publications (main search criteria: demand planning, supply
planning, Supply chain design, Supply chain planning strategies), internal Philips documents & field
research.
Result: Insights in the factors influencing the current supply chain planning method significantly. The
influence can be used as input to determine key selecting criteria for selecting alternative planning
methods.
2. Which planning model is most suitable for the LED supply chain?
Method: Literature research of scientific publications (main search criteria: Supply chain design, semi
conductor industry, planning strategy), internal Philips documents & field research.
Result: Insights in the factors influencing the supply chain planning. The influence can be used as input
to determine the most optimal supply chain planning model.
3. Which processes are related to the supply chain planning processes and what are the
interdependencies?
Method: Literature research of scientific publications (main search criteria: planning processes, supply
chain planning, supply and demand match) and field research.
Master thesis project – Roy Hartevelt Page 15
Result: A transparent process description. Mutual understanding and agreement about the business
processes and the independencies between several sub processes is necessary to make the insight of
the model clear and accepted and acceptable to use.
4. What are the relevant actors, what role do they currently have and what role should they have
according the new model?
Method: Field research, interviews with stakeholders, Literature research of scientific publications (main
search criteria: process design, organization setup).
Result: Revaluation of the current used stock control models and advice to adapt or maintain this
procedure, based on the new insights on economical stock calculation for different product families.
5. Which key performance indicators are relevant to evaluate the supply chain plan model and
what are the critical success factors?
Method: Field research, interviews with stakeholders, Literature research of scientific publications (main
search criteria: performance indicators in the Supply chain, performance indicators design
/implementation).
Result: A list with the most relevant supply chain performance indicators and secondly a risk profile in
which all the important factors influencing the supply chain planning performance (be. Stock levels,
lead-time, variance) are mentioned. The risk profile is necessary to evaluate the calculation rules as used
in the model and find general conclusions about the behavior of the model within the different product
life cycle faces.
6. What actions are recommended to convince the actors about the benefits of the new supply
chain planning model and way of working?
Method: Field research, interviews with stakeholders, Literature research of scientific publications (main
search criteria: actor management, change management, soft controls).
Result: An actor analysis in which all actors are described with all relevant arguments (pro /con) about
the new planning model and way of working. The actor profiles are necessary to evaluate the difference
in change acceptance processes and are an input for the advice model.
1.2.4 Methodology of the research
This chapter defines the methodology that will be used to gather an answer on the formulated research
questions. The methodology is a framework for the research and can be used as a guideline in the
research activities. The design method used in this research is based on the complex multi actor and
multi requirements methodology (Herder and Stikkelman, 2004). The conceptual model is based on
IDEF0 diagrams. The IDEF0 diagrams are based on the outcome of several workshops together with the
field specialist. The output of the model construction will be visualized and presented to the
Master thesis project – Roy Hartevelt Page 16
stakeholders. Based on the discussion during the presentation a pilot will be setup. The pilot will provide
insight information to optimize the model in the 2nd design.
Figure 4 design methodology (adapted from Herder and Stikkelmans, 2004).
One of the deliverables of the project is to deliver a tool and guidelines with the goal to optimize the
stock levels and to introduce supply chain controls. The decision support tool will enable management
to make the right decisions before we make the wrong one (R.E. Shannon, 1998).
The goals of the research are determined by the stakeholders (chapter 3). A clear formulation of goals is
essential to create suitable and accepted design and control afterwards if the design fits the desires of
the stakeholders. The goals are translated into objectives and constraints. These objectives and
constraints are matched with the design space. The design space is formed by the boundaries of the
specific situation. The research boundaries in combination with time and data restrictions determine the
design space for this research. With respect to the design space first a conceptual model is designed
(chapter 4). This conceptual model is discussed with experts and stakeholders and when necessary
adapted. After adapting the input of the field experts and stakeholders the conceptual model will be
translated in a real model (chapter 6). This model is tested and discussed again with stakeholders and
experts. After this evaluation the final model is designed and steps to use it in practice are provided
(chapter 8). Finally, results are used to test how good the model fits with the design objective.
1.2.5 Structure of the report
This report is split up in three phases: 1. Analysis, 2. Design and modeling and 3. Evaluation and
Validation. The three phases have an introduction, conclusion are split up in different chapters.
Master thesis project – Roy Hartevelt Page 17
The analysis phase concerns the current state (chapter 2 & 3). In this phase the requirements and
decision space are formulated based on the analysis of the as-is situation, input of field-experts and
strategic knowledge about the direction as set by Lumileds senior management. This analysis will also
draw attention to the multi actor setting of the problem. The tensions between the actors will discussed
and options to handle these differences are formulated.
The second phase, the design and modeling phase, starts with an analysis of an optimal stock level
calculation method for a key component of the LED. This method is based on product life-cycle
characteristics which determine delivery performance (start with chapter 4). The profiles per product
family are input for the next level of detail as used in the short-term planning tool. The objectives and
constraints form the solution space and are together with the risk profile input for the design. Besides
this input a list of criteria will be formulated to evaluate the model. The solution space is an input for the
design of the model. The results of the model and the implications for the component manufacturer and
the LED assembly are described in chapter 5. With the input of the previous chapters the design phase is
descript in chapter 6. How to implement a stock optimization model is answered in chapter 7.
The last phase (phase 3) is the evaluation and validation phase and starts with a validation of the model
with a simulation (chapter 8). Conclusions and recommendations of this research are placed in chapter 9
including the validation of the research questions and other reflections about theory and practice.
Master thesis project – Roy Hartevelt Page 18
Analysis Phase
The analysis phase is the first part of the design process. In this phase the concept and implications of
supply chain control and the structure of and relations between the component manufacturer and the
LED assembly & test are analyzed in detail. Absorbing uncertainties in the supply chain by well
positioned and calculated stock levels at indifferent positions in the supply chain are translated into
design parameters as input for the design and modeling phase. The results of the analysis are translated
in design requirements and objectives and are used for identification of the solution space.
Master thesis project – Roy Hartevelt Page 19
2 Analysis of supply chain control
2.1 Introduction to supply chain control This chapter explains the importance of supply chain planning and control focused on a component
supplier and the LED assembly and test manufacturer. Besides the introduction this chapter is divided
into 5 stages, stage 1 (par. 2.2) introduction of controls in the supply chain and it indicates the
importance of this subject and the related problems and gives an overview of the most important and
interesting academic literature and conclusions. The second stage (par. 2.3) discusses the problems and
losses for manufacturers and component suppliers. As a result of stage 1 and 2 stage 3 brings a
discussion about root causes and effects of limited supply chain control. The fore last paragraph
introduces the different ways of supply chain control models. The last paragraph of this chapter starts a
discussion on the design phase for the setup of solution model to identify the optimal solution.
The analysis phase is the first step in the model description as provided in Figure 4. The analysis provides
a basic building block for understanding the business issue. Secondly, the insights gained in the analysis
supply an input for the design phase. The analysis follows the structure of the TIP approach
(Koppenjan&Groenewegen, 2005). TIP categorizes the system aspects into three sections.
T Technology; analyzes the content of the issue
I Institutions; analyzes the important stakeholder requirements
P Process; refers to the interplay of stakeholders and their interests
2.2 What is supply chain control? Doing a good job in supply chain starts with living with uncertainties. In Supply chain control means
coping with several uncertainties of all activities in the scope of business, like variance in customer
demand, supplier performance and manufacturing and last but not least it means to follow the demand
pattern of our customers (T. Davis, 1993). At the end of the day management is most interested in
customer satisfaction and low inventory levels.
Uncertainties in the supply chain are daily business; beating the uncertainties will never result in
satisfaction or in a different way an uphill battle will never end in a great victory. Living with uncertainty
in Supply chain means integrating variances of each part of the process into a stock calculation model.
Figure 5 shows the supply chain uncertainty as published by T. Davis in 1993. Unfortunately, not all
uncertainties can be eliminated. However, other initiatives can redesign the Supply chain to reduce the
impact of uncertainties (T. Davis, 1993).
How to live with uncertainty? This question can be split in a longterm and a shortterm answer. For the
longterm a supply chain modeling methodology can adress the network, interdependencies and other
more structure related issues. For the shortterm, we will use models with scheduling algorithms to fine-
tune day to day performance.
Master thesis project – Roy Hartevelt Page 20
Figure 5 Uncertainty in the supply chain (T. Davis, 1993)
The real challenge is to manage the uncertainties in the supply chain in relation with customer service
level without incurring an undue burden of cost. Balancing the activities in the chain is the main goal to
achieve the necessary balance between cost and customer service.
Inventory has a stabilizing effect in supply chains (M.P. Baganha a.o., 1996) as a buffer to absorb
demand variability. The stabilizing function of stock will result in a lower variability in production than
the variability in market demand.
Figure 6 Inventory level with safety stock (S. Lutz e.a., 2003)
Figure 6 shows the development of inventory over time for a process with input and output quantities.
In case no uncertainty is involved the mean inventory can be lowered with the safety stock. With the
safety stock (SS) the described process can guarantee the deliveries. The mean inventory level is the
sum of the safety stock and half the quantity of input (lot size) into stock necessary to cover the usage
during lead-time (S. Lutz e.a., 2003).
The next step into supply chain control after understanding the supply chain uncertainties; translate all
known uncertainties into an inventory stocking policy. The constantly changes, so the uncertainties are
constantly changing too. Thus the inventory stocking policy is a dynamic process. Suppliers delivery
reliability can become better or worse with the time. On the other hand, demand for some products
becomes more predictable as products mature; demand for other products becomes more
unpredictable.
Master thesis project – Roy Hartevelt Page 21
In creating a dynamic stock policy we commonly use generic settings for A/B/C stock-keeping units. The
classification of items by transaction volume does not necessarily reflect the combination of all
uncertainties of the total supply chain (H.L.L. Lee e.a., 1992). In the default setup of the model the
possibility should be there to differentiate the service levels based on a potential ABC classification.
Currently no classification is made due the fact that all products are rather new and product launches
are planned twice a year. For managing the long term and the day to day business, more complex
calculation models are necessary for creating the most optimal and suitable supply chain for the LED
business. Consequences of calculating stock levels and customer deliveries with wrong uncertainties
results in overstock on some items but under-stock on others. Miscalculating the lead times for material
movement along the supply chain could result in investing in the wrong resources for performance
improvement.
2.3 Effects of supply chain control on the manufacturer and the component
supplier The value stream network is relatively small compared with other networks in the high tech industries,
but nevertheless the limited level of alignment between the key component supplier and the assembly
and test manufacturer can be marked as critical. The complexity of coordinating the supply chain is
exploding for the assembly and test manufacturer due to the introduction of new products and
demanding customers in a booming market. The ever-shortening product life-cycles combined with long
lead-times are challenging for the planning department at the manufacturer besides the supply side
with their struggle to follow the demand of their customers. A certain portion of the total stock is
reserved for covering the demand during lead-time. Therefore decreasing lead-time is the right lever to
cut inventories. The second portion of the inventory is reserved for the safety stock; the safety stock is a
function of the service level, the demand uncertainty, the replenishment lead-time and the lead-time
uncertainty. For a ‘fixed’ service level there are three levers that affect the safety stock; demand
uncertainty, replenishment lead-time and lead-time uncertainty. This thesis focuses on modeling the
right calculation method between the key component supplier and the assembly and test manufacturer.
As a result of the fragmentation of the supply chain in independent entities that tends to optimize
locally instead of coordinating and optimizing the entire supply chain. Competitive advantage of a
multinational is lost or gained by how well a company manages a dynamic web of relationships that run
throughout its chain of suppliers, distributors and alliance partners (de Kok e.a., 2005).
2.4 Root causes and influences on supply chain control Managing uncertainty in the LED supply chain has currently a main focus on the lead-time and the lead-
time variance of the key component supplier. Current implemented models are calculating the demand
versus the supply. Based on calculations and experience, a production order is placed. No uncertainty is
taken into account in the current planning models. The current supply chain control based on demand is
controlled by the S&OP (Sales and Operations Planning) of the assembly and test manufacturer. In figure
7 the current supply chain control model is visualized with the IDEF0 (Integration DEFinition for
Function) method. This model illustrates that the control parameters are limited (capacity constraints
Master thesis project – Roy Hartevelt Page 22
and carrier lead-time). This translates directly into limited possibilities to control the supply chain by
managing the uncertainty and stock levels. Between the planning setup of the key component supplier
and the assembly and test manufacturer there is a missing link between the short term and medium
term planning. The medium term planning cycle based on the sales and operations planning (S&OP) of
the assembly and test manufacturer is the starting point for determining the needed production
capacity. The short term planning cycle has a decentralized focus and is disconnected from the midterm
planning cycle. The key component supplier has independent weekly planning cycles calculating material
requirements based on component consumption. These disconnected processes cause long information
lead times and distortion. The long information lead time is caused by the several independent planning
cycles with a monthly update via the S&OP cycle. As an average half of the monthly period is a waste of
information lead time which can result in additional obsolescence risks. The focus of the logistics
manager of the assembly and test manufacturer is at reduction of replenishment lead-time from the key
component supplier and the variability of lead-time.
The IDEF0 (Structured and Design Technique, D.T. Ross, 1981) of the current supply chain control is
displayed in Figure 7. In step A1 the master production schedule (MPS) is setup based on the sales and
operations planning (S&OP) and the distribution of the components. Based on the MPS of the LED
assembly & test the component supplier will generate a material requirement plan (MRP) (A2). The
capacity constraints of the components production is the control element. The capacity constraints
control element takes into account the available production capacity and the factory supplies for the
long term (be. Investments for capacity increase). As input for the MRP the factory yield of the back-end
processes are taken into account. The work-in-process (WIP) is taken as a point of departure for the new
plan cycle.
Master thesis project – Roy Hartevelt Page 23
Figure 7 IDEF0 Supply Chain control model as-is situation
The next step after the MPS and MRP is to align the factory plan (A3) with the demand of the assembly
and test manufacturer, a weekly schedule of a day to day plan. Al the steps after activity A4 are
transport and order-receiving at the LED assembly & test factory.
2.5 Supply chain control models Control of the supply chain means understanding and managing the uncertainty in the scope of the
chain. In a simplistic way the cycle stock (demand during lead-time) will be increased with a certain
percentage or some extra days of inventory. For calculating the safety stock levels as part of a
replenishment strategy model several variants are available to use, the question is which calculation
method of the safety stock is the most suitable for the LED situation and what are the differences and
explain them. Based on publications of S. Chopra eo, 2004, P. Suwanrauji eo, 2005 and S. Chandandeep
2010 three safety stock calculation methods have been selected:
SAFETY STOCK CALCULATION 1 (CHARLES ATKINSON, 2005);
[1]
SS1 = safety stock calculation according Charles Atkinson
Z = service level
LT = Average lead-time
δdem2 =Standard deviation of demand^2
TITLE:NODE: NO.: 0As is A0 SADT Lumileds
S&OP
A1
Prepare MPS
Png
A2
Generate MRP
Mhz
A3
Align Factory
planning back-
end Mhz
A6
Stock update
Penang
Platelet stock Png
A5
Prepare In-
transit overview
Yield back-end Mhz
WIP Mhz
Platelet distibution
MES
Capacity
constraints
Front-end Mhz
Carrier
Lead-time
Stock overview
Png
A4
Generate report
WIP back-end
MhzStock pre grinded wafers
Capacity
constraints
JD Edwards stock data
MPS –
Long term
MRP -
Short term
Confirmed
Planning
WIP
report
Goods
in-transit
Master thesis project – Roy Hartevelt Page 24
dem2 = Average demand ^2
δLT2 = Standard Deviation of Lead Time ^2
SAFETY STOCK CALCULATION 2 (KENT LINFORT, 2006);
[2]
SS2 = Safety stock calculation according Kent Linfort
σFE2 = Forecast error
LTI = Lead-time interval
FI = Forecast interval
σLT2 = Lead-time error
D2 = Average demand during lead-time^2
Z = Desired service level
OCI = Order cycle interval
SAFETY STOCK CALCULATION 3 (DAVE PIASECKI, 2009) ;
[3]
SS3 = Safety stock calculation according Dave Piasecki
δ = standard deviation demand
Z = Service factor
LT = Lead-time factor
OC = Order cycle factor
Dem = Forecast-to-mean-demand factor
Figure 8 Safety-stock calculation comparison
With the help of a small example the three calculation methods (SS1, SS2 and SS3) are compared. The
result of the comparison is visualized in Figure 8, which shows a safety stock level at three different
service levels. The Lumileds Supply Chain needs Safety-stock for absorbing the uncertainties at the
demand and supply. In making the right choice related to the safety-stock calculation it is most
-
50
100
150
200
250
300
350
Z=1.64 Z=1.88 Z=2.33
SS in
Pcs
Safety-stock calculation comparison
SS1
SS2
SS3
Master thesis project – Roy Hartevelt Page 25
important that the calculation method is able to deal with the different types of Supply Chain
uncertainties. The limited possibilities in SS3 to integrate the different uncertainty factors drop out this
calculation method. Calculation method SS2 needs a lot of not standard available information and
secondly the calculated safety-stock level is very close but higher than the SS1 calculation method.
Based on the characteristics (fast growing and many different uncertainties) of the LED business of today
the safety stock calculation of C. Atkinson (SS1) has the best fit and will be used in further stock level
calculations.
2.6 Conclusion supply chain control analysis The need to focus on supply chain control drastically increase due to the expanded product portfolio of
the LED products. The market is growing very fast and the competitors are aggressively competing for
market share. LED industrial customers require high service standards: Out of stock situations are not
tolerated. The direct impact of supply chain controls in out of stock improvement actions can be
significant. This project will deliver a model design and implementation and will open new ways to
improve or to minimize out of stock situation caused by supply chain controls (sub research question 1
and 2).
There are multiple root causes for out of stock situations in the supply chain of Lumileds. The most
important and relevant root causes for this research are: underestimation of uncertainty (production
lead-time and production quantity related), long order lead-time, demand underestimation, new
product introductions and data inaccuracy. All these root causes haves an effect on the supply chain
control model and will be incorporated in the design and modeling phase (sub research question 3)
Measuring stock levels can be done in several ways, the question is which one is the most suitable for
their business situation and how the calculation method can be integrated into the overall control
model. Finally a control mechanism is needed to control the supply chain between the component
supplier and the assembly and test manufacturer to serve the fast growing LED market in a for Philips
optimal way (main research question).
Master thesis project – Roy Hartevelt Page 26
3 Analysis of the component supplier and the LED assembly
manufacturer The highlights of the Philips Company, within this research (Philips, Lighting and Lumileds) are
positioned in appendix C. The analysis on supply chain control and organization starts in paragraph 3.1.
In this chapter also the underlying reasons for this research are made explicit. After a brief introduction
and description the processes within and between the component supplier and the assembly and test
manufacturer are discussed (section 3.2). The multiple stakeholders all have other interests, goals and
influences in relation to supply chain planning optimization. Mapping this information (section 3.2) can
help to improve the process. Section 3.3 provides round up and summaries around these aspects.
Section 3.4 finally gives a wrap up of the whole analysis phase.
3.1 Supply chain management In a fast growing business environment supply chains are facing some typical pitfalls and opportunities
(H.L. Lee 1992). Pitfalls like not having the right metrics, a misunderstanding about what is customer
service and even more basic understanding who is the customer and finally limited and poor supply
chain coordination are the backbone of the analysis.
The value chain of LED is build up out of 3 blocks, see Figure 9. The first block “component production”
covers the component manufacturing in Maarheeze (NL) and Singapore (SI). Transport is executed by an
express courier. The LED assembly & test is located in Penang (MA). The LED assembly & test is driven by
sales orders. Activities done in Singapore are out scope of this research.
Figure 9 value chain LED production
3.2.1 Current supply chain processes
In Figure 10 shows the current run-book. Key elements in this run-book are the coupling between the
monthly update from sales and operations planning (S&OP) with the material requirement planning of
the LED assembly & test factory. The MRP is translated towards the component supplier. The back-end
factory planning is the fundament for next steps in the component factory.
The monthly S&OP update and the weekly alignment via the MRP with the component supplier have at
least a delay of one week in information retrieval and transmission. The monthly and weekly alignment
Component production
TransportLED
assembly & test
Master thesis project – Roy Hartevelt Page 27
sessions are also discourage short production planning cycles leading to gross forecast errors and
inventory and backorder accumulation (L.L. Hau, 1992).
Besides sharing of information, coordinating of information is an important element in optimizing
Supply chains (W. Jammernegg, 2006). In the supply chain of LED the ownership of the component
stocks is at the LED assembly and test manufacturer. In running business the planner of the component
supplier is managing the stock levels and takes a decision if applicable (experienced based).
Figure 10 supply chain execution run book
As a result of shared responsibilities there are no suitable performance measures for the complete
supply chain. A reason for not having those kinds of metrics is that they are not directly linked to
customer satisfaction. As shown in Figure 11 the planned versus the real delivered quantities are
measured besides the confirmed line item performance (CLIP=yellow line) and the confirmed volume
performance (CVP=grey line). Not taken into account are the total order cycle time, average backorder
levels, average lateness or earliness and back order profile (backorders that are one week late, two
weeks later). The main focus at the component supplier was to produce the forecasted quantities during
the selected period. Most of the time the volume was there, a minor element is that the mix of the
products to be delivered was not in line with expectation. The plan performance based on volume (CVP
=grey line) was 90% during the selected period; the performance based on the mix (products delivered
versus ordered) was around 40% (CLIP=yellow line) during the selected period. Main reason for having a
focus on total quantity instead of delivering the right mix is difficult to answer. After several interviews
with field experts the overall conclusion was that the day target asked for a minimum number of moves
at a workstation. After beating the target of the required number of moves management was satisfied,
Master thesis project – Roy Hartevelt Page 28
without taken any notice about the right moves. In this kind of situations the workforce will do were is
asking for; making moves!
Figure 11 performance current supply chain control model
Two focus areas in a supply chain area are the reduction of the replenishment lead-time from suppliers
and the variability of this lead-time (S. Chopra, 2004). In supply chains with large variability in lead-time,
reducing variance have a greater impact on reducing safety stock levels than cycle time reduction. In our
research we recognize that cycle time and the variance are both increasing. In phase 2 of this thesis we
focus on the three levers that affect the safety stock (demand uncertainty, replenishment lead-time and
lead-time variance).
Figure 12 shows de coefficient of variance (CV) of the throughput time of the component supplier per
period (CV calculation by S. Chopra et al, 2004). A period is defined as a week in 2010. The CV value is
increased by almost 100% in a period of 10 periods. The increase of the CV is caused by a growing
(standard deviation) δ and μ (mean) and on the other hand due to diffused production priorities
because of several new products in production. Secondly, balancing capacity between the component
supplier and the assembly and test manufacturer is done during the monthly Sales & Operations Plan
meeting. Deviation from plan due to higher sales volume as forecasted will directly effects the CV in a
negative way (increase).
0%
20%
40%
60%
80%
100%
120%
140%
160%
180%
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
3,500,000
4,000,000
4,500,000
5,000,000
wk1
00
1
wk1
00
2
wk1
00
3
wk1
00
4
wk1
00
5
wk1
00
6
wk1
00
7
wk1
00
8
wk1
00
9
wk1
01
0
wk1
01
1
wk1
01
2
wk1
01
3
wk1
01
4
wk1
01
5
wk1
01
6
wk1
01
7
wk1
01
8
wk1
01
9
wk1
02
0
wk1
02
1
wk1
02
2
wk1
02
3
wk1
02
4
wk1
02
5
wk1
02
6
"Co
nfi
rme
d L
ine
Ite
m P
erf
orm
ance
-C
LIP
& C
on
firm
ed
vo
lum
e
pe
rfo
rmac
e C
VP
(%
)
Pla
nn
ed
/ R
eq
uir
ed
vo
lum
e (
qty
)
Planned qty Achieved qty
CVP (%) actual CVP (%) target
CLIP (%) actual CLIP (%) target
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 2 4 6 8 10 12
Co
eff
icie
nt
of
Var
iati
on
Period
CV
Linear (CV)
Master thesis project – Roy Hartevelt Page 29
Figure 12 CV throughput time component supplier
CV = δ/μ [4]
CV = coefficient of variance
δ = standard deviation
μ = mean
Managing the customer satisfaction with a better supply chain performance starts by managing the
variances at the upstream part of the chain. A part of the solution model in phase 2 will take care about
the upstream part of the chain – the component supplier.
3.2 LED assembly and test manufacturer
3.2.1 General introduction LED assembly and test manufacturer
The LED assembly and test manufacturer of Philips Lighting is located in Penang (Malaysia). The two
main suppliers of key elements are a Philips component supplier in Singapore and Maarheeze. The
components from Singapore are the most expensive ones and for that reason marked as leading in the
assembly factory. The leading position of Singapore brings Maarheeze into the position of a follower.
Key elements of a follower are that flexibility and production capacity should be no issue. However,
currently Maarheeze is not able to fill the pipeline with key components due to capacity issues. As a
result of this situation the assembly and test manufacturer is not able to follow the demand of our
customers.
This research will have a focus on the improvement possibilities between the component supplier and
the assembly and test manufacturer. The leading position of the components from Singapore is taken
into account in the analysis done in the next phase.
3.2.2 Ramp up in a fast growing market
Due to the recovery of the global economic market the LEDs is widespread use as the backlighting units
not only of large-sized LCD panels used in TV and computer screens but also on smaller LCDs in a broad
range of devices including notebooks, cell phones, portable navigation devices, keypads and many other
applications. For 2010 till 2015 the global LED market expects a double digits growth
(www.semiconductor-today.com), this double digit growth emphasized the need for controlling the
supply chain.
As a result of the fast growing LED demand a shortage is expected on the market by 2011. As a reaction
on the market shortage vertical integration at our competitors is going on for maximum safeguarding
their needed capacity.
3.3 Stakeholder analysis The stakeholder analysis has three main parts. Firstly, the supply chain organization (section 3.3.1.) of
the LED business unit and the position of the strategic plans. Secondly, the interests and objectives of
Master thesis project – Roy Hartevelt Page 30
the several stakeholders in the value chain (section 3.3.2/3). And finally, organizational acceptance and
the role of the project principal are discussed (section 3.3.4/5).
3.3.1 Supply chain organization
The world wide supply chain manager (WW SCM) is part of the management team of the business unit
(BU). In this position the WW SCM has a powerful business position to drive the BU into the direction of
the DIMEs project goals. The director customer service, director APAC (Asia Pacific) supply and the world
wide program manager are reporting to the WW SCM, see also Figure 13. SCM fulfills an important role
in the BU, in this setting the BU management prevent themselves from product and process design
without any supply chain consideration (H.L. Lee, 1992). This setup results that all of the anticipated
savings done by product design, manufacturing and assembly are not be lost by increased distribution
and inventory costs. Potentially costs increases are involved due to SCM intervention in the design,
manufacturing and assembly processes at the short term. At the long run the company benefits from an
optimal flexible supply chain concept for meeting demand. Flexibility is especially important in new
product introduction and ramp up, where demand is highly variable as well as unpredictable. The colors
in chart Figure 13 will be used in Figure 14, based on that the functional areas in the supply chain will be
discussed.
Figure 13 the LED worldwide supply chain organization
Every organizational layer in the three tier management organization of SCM has own tasks and
responsibilities to perform most optimal in design, manufacturing, assembly and sales processes.
Demand and supply processes are visualized in Figure 14. Customer service is responsible for the
customer interfaces, order handling, allocation processes during scarcity and customer specific demand
solutions, all demand related. Order scheduling activities at supply fulfillment close the loop between
customer demand and order fulfillment.
WW SCM
Director
Asia Pacific
Supply Chain
Planning
Manager
Director
Customer
Services
Logistics
Manager
Director
WW
Programs
Supply
Chain Exc.
Manager
Master thesis project – Roy Hartevelt Page 31
Figure 14 the worldwide supply chain organization in functional areas
The APAC supply manager has overall responsibilities in demand and supply. The customer complexity is
translated via a Sales and operations planning (S&OP) process into a supply plan. Based on the S&OP (an
outlook of 1.5 year ahead) the midterm and short term planning activities can be aligned for meeting
the demand as efficient as possible. S&OP is also used for longer term investment decisions. The
logistics and planning managers are running the day to day business and balancing demand and supply
variance versus customer lead-times and stock management.
An important role in the APAC supply organization is the supply chain excellence department. This
department initiates and executes improvement projects in the total LED supply chain. Key supply chain
topics which correspond to the strategies for improvement in supply chain design, integral supply chain
performance measurements and integration of control and planning support systems are covered and
managed by the supply chain excellence manager.
3.3.2 Stakeholder setting
Figure 15 shows that in the component production many stakeholders are available. The many
stakeholders at the component production will compete for required capacity in achieving individual
goals. As mentioned in the introduction of this master thesis, the LED business is growing fast. In a fast
growing market development and engineering is an enabler for further business growth. Besides
manufacturing capacity, engineering and development activities require factory capacity. An optimal
supply chain planning model will bring stability and clearness for involved actors. Based on customer
forecasted demand all capacity left can be used by engineering /development.
DEMAND
PLANNING FULFILLMENT
Asia Pacific Supply Management
PlanningOrder Confirmation/scheduling
Key Customer Interface
Scenario Planning
Inventory Monitoring
Weekly Build Plans
MRP
Supply Chain ExcellenceS&OP Planning
Scenario Planning
Inventory Analysis
Customer ServicesCustomer interface
Order Handling
Customer Allocation
Consignment/VMI Inventories
LogisticsDie Bank/RDC/FGI handling
Kitting
Shipments
Impex
Program ManagementApple Demand / Supply Planning
SUPPLY
Supply Chain ExcellenceS&OP Process Improvement
Reporting requirements
System requirements
Order
Scheduling
Binning
Officer
Forecast /
Demand
collection
Master thesis project – Roy Hartevelt Page 32
Figure 15 stakeholders in the value chain
Stakeholders in transport are service oriented with limited or no influence on product capacity.
Important for supply to know are the lead-times and variance for calculation purposes.
In the LED assembly & test stage many stakeholders are present. A supply chain planning model will
bring more clearness about product availability, throughput time and lead-time variance. On the other
hand if there is no capacity and/or flexibility in the chain the variance and lead-time should be covered
by stock. The balance between flexibility and delivery reliability is one of the key items.
3.3.3 Methodology of stakeholder analysis
The implementation of an optimal solution depends on the stakeholders in relation with the problem
situation. For that reason it is important to be aware of the perceptions, personal goals and interests of
the stakeholders in an early stage. Besides the identification of the stakeholders, the stakeholders can
bring relevant information about our demand. Also the powers and dependence of the different actors
related to the discussed subjects are analyzed. The result of this analyze contributes to finding an
accepted and most suitable solution for both, the component supplier and the assembly and test
manufacturer.
3.3.4 The stakeholders’ action field
Based on the business goals and the functions involved it can be concluded that commitment exist for
the implementation of a supply chain planning model. Main barriers are uncertainties about the effects
of the supply chain optimization model on their daily job and the interaction between the several
involved stakeholders.
Component production
TransportLED
assembly & test
Component suppliers
Production planner
Production
manager
Engineering
manager
Development
manager
Product manager
Shipping:
Parcel /Express carrier
Logistics:
Logistics department
component supplier
Component planner
Production planner
Supply manager
Planning manager
Supply chain excellence
manager
Assembly & test
Production manager
Master thesis project – Roy Hartevelt Page 33
An overview of the actors and their position is presented in Figure 16 . The figure divides the
stakeholders in four groups. The first distinction is based on the dedication of the stakeholder to the
implementation of a supply chain optimization model. Those stakeholders that are dedicated to build a
model are positioned on the positive vertical axis, those who oppose it are on the negative side of the
vertical axis. The second distinction is based on the critically of the stakeholders in the process. If the
participation of stakeholders is critical, read if their participation is required for setting up a supply chain
optimization model, they are positioned accordingly on the horizontal axis.
It is important to note that this figure gives a static overview. Stakeholders can switch positions and
change from non-dedicated actors when perceptions change or compensation is offered.
Figure 16 stakeholder diagram
3.3.5 Importance of stakeholders
As indicated in Figure 16, 11 stakeholders are directly involved in supply chain optimization. To avoid
problems due to stakeholders during the research it is obvious to manage the stakeholders closely by
organizing workshops and bi-lateral meetings during the full process. During the planned workshops the
planners learn the structure of the models and have the opportunity to ask questions and come up with
possible additional requests.
Some opposition for the planning model can be expected from the logistic department at the
component supplier (LCS). The planner of the LCS is currently taking ownership about the planning and
ordering process between the two locations. During the research the current jobs of both, the LCS
planner and the planner at the manufacturer side will be challenged. The planners probably will show
their use and try to counter against the supply chain optimum. Besides the overall optimum there is a
CriticalNon-Critical
Dedicated
Non-Dedicated
Production planner
Maarheeze
Production
manager
Maarheeze
Engineering
manager
Maarheeze
Development
manager
Maarheeze
Product manager
MaarheezeSupply manager
PenangProduction planner
Penang
Component planner
Penang
Production
manager Penang
Supply chain
excellence
manager
Planning manager
Penang
Master thesis project – Roy Hartevelt Page 34
continuous drive to go for local optimum solutions. The success of the supply chain optimum solution
depends on in which degree these people can be convinced about the robustness, flexibility and trust
about the overall optimum. The main difference between the overall optimum and a local optimum
solution is that the overall optimum is not touchable for the people working at the supply side. At the
other side, the people working at the supply side are responsible for the overall performance. For this
reason development of supply chain performance indicators are in scope of this thesis.
Effective structure requires an internal consistency among the design parameters and contingency
factors (H. Mintzberg, 1983). During the preparation phase of this assignment as assigned by the Supply
Chain manager at head-office the Supply chain organization was looked-over. The Lighting organization
has created a new business unit called Lumileds. Lumileds is smaller and less old than the traditional
Lighting organization. Formalization of systems, processes, tasks profiles and specialized job functions
were less developed in Lumileds compare to Lighting. The strategic drive into more Supply chain
controls in the Lumileds business is from that perspective not that strange. Within the background of a
less formalized organization the system model requirements were drawn up.
The actor contributes to the list of requirements, poses constraints, recognizes performance criteria and
identifies areas of interest for research options; these are presented in Table 1.
The actors as described in table 1 are all part of the Lumileds organization but part of different local
orientated organizations. All main stakeholders have a functional relationship with the Lumileds global
supply chain manager, the principal of the assignment to improve the supply chain controls. Within the
vision of the global supply chain manager supply chain controls are an enabler to jump to a next level of
maturity, meaning: reliability improvement, throughput-time reduction, stock reduction and an
improved customer service level.
The production planner of Maarheeze (the component supplier) is the current process owner and takes
care about the current stock positions at the manufacturer. The production planner Maarheeze has a
key role in the in the current way of working within Lumileds. According the production planner of
Maarheeze the expected improvements can be managed by the currently available planning tools.
The component planner Penang is responsible for managing the component supplier and the
components goods flow at the manufacturer. The manufacturer has direct contact with the final
customer in the total chain, for that customer satisfaction is one of the key drivers of the overall
organization.
The supply chain excellence manager is responsible for supply chain improvement projects within
Lumileds, for that reason the SC excellence manager is an ambassador and a sponsor for our
improvement project.
Finally, the material requirement manager has the responsibility for the overall incoming goods flow.
The component planner Penang is reporting to the requirement manager. The material requirement
Master thesis project – Roy Hartevelt Page 35
manager has a need for supply chain performance reporting and model decision parameters to set
which influences performance indicators.
The main stakeholders as mentioned in table 1 have different interests in the results of the supposed
improved situation. At the one hand the production planner of the component manufacturer has his
own way of working and has limited interests to change the current situation. At the other hand the
manufacturer is facing an unreliable supplier as key component supplier. In the mindset of the Lumileds
an improved set of supply chain controls are necessary to manage the fast growing market with all the
expected product proliferations. The global supply chain manager asked the supply chain improvement
manager and the material manager to take care about the success of the supply chain control
improvement project. The planners at the manufacturer and at the component supplier are both key
knowledge owners necessary to design a most suitable supply chain control model.
Master thesis project – Roy Hartevelt Page 36
Table 1 requirements, criteria, constraints & design option stakeholders
Main
stakeholder
Requirements Criteria Constraints Design options
Production
planner
Maarheeze
The interrelation
between
departments in the
value chain should
be efficient
Minimal
transaction time
Efficiency in
process design
Ownership of
optimization model
and decision making
process.
Vertical supply
chain integration
Component
planner Penang
The risks should be
covered by stock
policies.
Capacity
constraints to be
covered in the
S&OP
No below customer
service level
Options for lead-
time, variance
and service level
regulation.
Supply chain
excellence
manager
The model should be
stable and
predictable
To be ready within
4 months
- Options for lead-
time, variance
and service level
regulation.
Design of
performance
measurements
Material
manager Penang
The model should be
able to simulate
scenarios’
- - Design of report
generation
To resolve these conflicts effectively and turn the supply chain into a weapon for gaining competitive
advantage requires the development of an integrated supply chain driven by the needs of the business
(G.C. Stevens, 2007).
3.4 Conclusion analysis The need to focus more on supply chain planning and optimization instead of local optimization plans
has ascended the past few years. The LED market is growing enormously and actions to save the
company market share are necessary to survive. Industrial customers can switch, can go for a more
traditional solution or something don’t know yet. All these different kinds of behavior provoke business
losses and will harm continuity. In case industrial consumers decide not to purchase, this costs Philips
billons of turnover per month. However there are some appropriate ways to determine an optimal
solution for the optimization challenge. Research has indicated that are required actions to optimize the
LED supply chain. Without actions there will be too high inventory levels, utilization- and yield losses
Master thesis project – Roy Hartevelt Page 37
with resulting financial losses. Current losses are caused by a lack of communication between the
component supplier and the manufacturer. Figure 11 shows a delivery mix performance (=CLIP;
customer line item performance) a score of 40% (average) with exceptions to below 20%. This mix
performance results in an additional uncertainty in the Lumileds supply chain of 5%.
Sub research question 3: which processes are related to the supply chain planning processes and what
are the independencies, is answered within chapter 3. As discussed in chapter three one of the enablers
of an optimal supply chain are communication and clear roles and responsibilities related to the supply
chain planning processes. The conclusions made after analyzing the lumileds supply chain is that the
current task and responsibilities are not sound and clear communicated throughout the Lumileds supply
chain including the key component suppliers.
Master thesis project – Roy Hartevelt Page 38
Design and modeling phase
The analysis phase illustrates the need for Philips Lighting Lumileds to focus on growth. To make this
possible, the controls in the supply chain must be optimized. Information from the assembly and test
manufacturer and the component supplier can be used to identify the supply chain uncertainties and
can be turned into safety stock level, scheduling and replenishment advice. In the design phase, the
process and model to draw up a supply chain control advice will be designed. The basic building blocks
for the model are the involved product families and their uncertainty. Finally, an improved planning
method for the component supplier will be introduced.
Master thesis project – Roy Hartevelt Page 39
4 Analysis of component through-put-time
4.1 Introduction Figure 17 is a simplification of the complex supply chain of the component supplier and the assembly
and test manufacturer. The 13 process steps in the production of the component are divided into two
stages; a front- and a back-end. The customer decoupling point is arranged via a supermarket between
the front- and back-end.
Figure 17 Simplified Supply chain of Lumileds
Managing stock levels asked for visibility in throughput time and for the possible uncertainties in the
chain. Section 5.2 and 5.3 elaborate about the difference between the front- and back-end processes at
the component supplier. The differences in throughput time per product family are analyzed based on
literature (normality testing) with Minitab and insights of professionals in the field (section 5.4).
For a better understanding of the component supply chain and the complexity at the component
supplier see below figures (Figure 19, Figure 20, Figure 21 and Figure 22). In the first picture the front-
end of the component supplier is visualized. Starting with the raw materials, 5 process steps end in a
supermarket. The semi finished product is stored in a supermarket to shorten the customer reaction
time as much as possible. Figure 22b shows the total throughput-time of the Front-end of the
component supplier. Figure 18, shows the distribution of the throughput-time of the front-end
processes. In this picture the distribution of the front-end is challenged between a normal distribution, a
3 parameter Weibull distribution, a gamma distribution and a 3-parameter gamma distribution. The
probability plot as shown in figure 18 measures how well the data follow a particular distribution. The
better the distribution fits the data, the smaller this AD (Anderson-Darling) statistic will be. When trying
to determine which distribution the data follows multiple Anderson-Darling statistics can be applied to
compare the distributions. The distribution with the smallest Anderson-Darling statistic has the closest
fit to the data. For our data set of the front-end this will mean that a 3-parameter gamma distribution
fits best. However, the data is selected in a period without any improvement action implemented like
fifo at the work centers and buffers in between. Due to the limited process controls production batches
are held up with or without any reason resulting in longer and unreliable throughput-times.
Master thesis project – Roy Hartevelt Page 40
Figure 18 Probability plot front-end component supplier
Figure 19 Front-end component supplier
The strict fifo lanes between the process steps in the front-end are buffers for compensating the
different processing times. The fifo lanes and process steps are supervised by a MES (manufacturing
execution system), based on system setup production batches are managed through the line.
The back-end of the component supplier has 7 or 9 process steps. Product group A has 7 process steps
and product group B has 9 steps. For product group B, one process step is executed at a different
location (outsourced to a subcontractor).
7550250
99.99
99
90
50
10
1
0.01
T hroughput-time in days
Pe
rce
nt
100.0010.001.000.100.01
99.99
90
50
10
1
0.01
T hroughput-time - T hreshold
Pe
rce
nt
100101
99.99
99
90
50
10
1
0.01
T hroughput-time in days
Pe
rce
nt
100.010.01.00.1
99.99
99
90
50
10
1
0.01
T hroughput-time - T hreshold
Pe
rce
nt
Gamma
A D = 13.636
P-V alue < 0.005
3-Parameter Gamma
A D = 3.869
P-V alue = *
Goodness of F it Test
Normal
A D = 44.149
P-V alue < 0.005
3-Parameter Weibull
A D = 8.791
P-V alue < 0.005
Probability Plot for front-end component supplier
Normal - 95% C I 3-Parameter Weibull - 95% C I
Gamma - 95% C I 3-Parameter Gamma - 95% C I
Granulation Pre-grindingSinteringBBOPressing
FIFO
Max X
FIFO
Max X
FIFO
Max X
FIFO
Max X
Raws
Front-end
907560453015
400
300
200
100
0
Days
Fre
qu
en
cy
Throughput-time front-end component supplier
Master thesis project – Roy Hartevelt Page 41
Figure 20 Back-end component supplier
The various throughput-times for product group A and B are shown in Figure 21.
Figure 21 Throughput-time back-end component supplier
The level of uncertainty is visualized in the standard deviation in Figure 21, for A and B types products is
such unreliable that extra stock at the assembly and test manufacturer are necessary. Reducing the
variation in the back-end of the component supplier needs several actions to succeed. Important is that
the model should be able to plan with all uncertainties as currently available. Optimization of the
uncertainties at the component supplier is not a primary task of the solution design but the model
should also be useful to show the effects via the stock levels due to the high level of variance in the
chain.
Dichroic
Aachen
Coating
Measurem
ent platelet
100%
Visual
inspection
100%
SeparationSeptaping
Waf
measuringGrindingGrind-taping
FIFO
Max X
FIFO
Max X
FIFO
Max X
FIFO
Max X
FIFO
Max X
FIFO
Max X
FIFO
Max X
WW
Hikari
Back-end
75604530150
300
250
200
150
100
50
0
Throughput-time in days
Fre
qu
en
cy
Throughput-time back-end component supplier A-type
6048362412
14
12
10
8
6
4
2
0
Throughput-time in days
Fre
qu
en
cy
Throughput-time back-end component supplier B-type
Master thesis project – Roy Hartevelt Page 42
Figure 22 Entry process steps assembly & test manufacturer
Figure 22 shows first steps of the assembly and test at the manufacturer in the Asia-pacific region.
Interesting is the second incoming goods flow from another key-component supplier located in
Singapore. Demand planning information of the second key-supplier will be used as input for the weekly
replenishment orders at the component supplier.
4.2 Front-end In Figure 23 the cycle time of the front-end is shown. At the primary X-axis the median cycle time in days
is mentioned. At the secondary X-axis the WIP is indicated as the total number of different batches in
the line during the time frame as mentioned at the Y-axis. The vertical bars in the graph representing the
variance in the selected period. Remarkable is the difference in spread over the selected period. The
selected period is chosen based on the following pillars: 1. Ramp-up of the based products was started
at the beginning of quarter 2 of 2010, the period before can be identified as a development phase. 2.
Medium quarter 3 2010 the first effects of the model implementation are visible in the weekly
performance measurement. Based on the analysis as presented in Figure 23 the conclusion is that the
variance of the cycle time in the front-end process is growing. At the other hand (and in line with the
first conclusion) the work in process (WIP) is increased too. Due to limited process control the cycle time
of the production batches will be influenced. The median of the cycle-time is relative stable with a
growing variance and WIP. Due to limited process control the cycle time of the production batches can
be influenced in a positive or negative way. More details about the sub processes of the front-end part
of the component manufacturer are described in appendix D.
Figure 23 Median throughput time front-end
Set matching
(combine tile &
platelet)PnP sorting
Platelets in
DieBank
I
Platelets in
DieBank
I
LLM 2702 RAW
I2702 DB
Singapore
Saber Fab
Tile Altilon & Flash
Tile WW & Amber FIFO
Max X
I
2702 DB
Flash
Visual
inspection
100%
Expand to
metal foton ring UV Cure Quality control
100%
Lumi visual
inspection
100%
0
10
20
30
40
50
60
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
14 16 18 20 22 24 26 28 30 32
WIP
Med
ian
Cycle
Tim
e (d
ays)
Work Week 2010
Cycle-time front-end component supplier
CT (days) WIP
Master thesis project – Roy Hartevelt Page 43
A storage location is located between the front- and the back- end. Based on stock keeping rules as
described in section 5.5 an optimal service level towards the back-end is managed.
4.3 Back-end The variance of the back-end processes is representing by the vertical error bars in Figure 24. The back-
end process has a direct link with the assembly and test manufacturer. Based on the production process
of the manufacturer the back-end of the component supplier is managed. Due to uncertainties in the
production process of the component supplier extra stock is necessary at the manufacturer to cover the
risks of the uncertainties. Comparable with the front-end no strict hand over rules from one process
step to the other are used, the limited level of controls ends in a higher level of uncertainties.
Figure 24 Median throughput time back-end
More details about the sub processes of the back-end part of the component manufacturer are
described in appendix E.
Variance in the back-end of the component supplier is caused by several topics. Most of the struggles
responsible for the current variance in lead-time are related to the strong ramp-up at the production in
2010. Based on a defect analysis (same period as Figure 23) a Pareto analysis was prepared, see Figure
25.
0
10
20
30
40
50
60
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
14 16 18 20 22 24 26 28 30 32
WIP
Med
ian
Cycle
Tim
e (d
ays)
Work Week 2010
Cycle-time back-end component supplier
CT (days) WIP
Master thesis project – Roy Hartevelt Page 44
Figure 25 Pareto negative effects on throughput-time component supplier
Figure 25 shows the main sources of the variance as present in the factory of the component supplier.
Important in the Pareto analysis is that five out of nine key issues are topics related to staffing and/or
the organization. Related to Mintzberg the organization is changed from an adhocracy towards a
Machine Bureaucracy. Process alignment is moved from mutual understanding towards a Standard
Operating Procedure (SOP) in which the work processes are described including controls, escalation
paths, communication flow and performance measurements.
The Pareto as presented in Figure 25 is a result of a measurement as held in week 10 till week 40 2010.
The negative effects of the defects were not registered only the root-cause of the incidents, for that
reason no detailed overview of the total delay can be presented.
4.4 Supermarket stock level calculation In the previous sections the process of the component supplier is discussed. Optimizing the supply chain
with a focus on short lead-times to the LED assembly and test manufacturer are the most important
design objectives. Meaning, the supply from the component supplier to the manufacturer should be
boring predictable. A common way to determine the number of Kanban tickets is described with the
formula (Marmostein and Zinn, 1990):
[5]
KB = number of kanban tickets
D = forecast demand (weekly)
LT = production lead-time back-end component supplier (in weeks)
Z = service level
STD = standard deviation of demand * coefficient of variance
Q = batch size
Sta
ffin
g c
ap
acity is
sue
Cap
acity a
lignm
ent
sub
co
ntr
acte
rs
Eq
uip
em
ent d
ow
n
Qualit
y is
sues
Main
tenance -
pla
nned
Main
tenance -
unp
lanned
Prio
ritiza
tio
n o
f activitie
s a
t shif
t le
vel
Fix
ed
bre
ak-t
imes
Deta
iled
activity s
ched
ulin
g
mis
sin
g
Sup
ply
co
nsum
ab
les
Share
d r
eso
urc
es w
ith o
ther
pro
ductio
n f
acili
ty
Sup
po
rtin
g m
ate
rials
are
m
issin
g
Wro
ng
measure
ment
0%
20%
40%
60%
80%
100%
0
5
10
15
20
25
30
35
40
45
Cum
ula
tive %
De
fec
ts
Causes
Pareto analysis defects production component supplier
Vital Few Useful Many Cumulative% Cut Off % [42]
Master thesis project – Roy Hartevelt Page 45
The number of kanban tickets per product type is determined by the demand during lead-time plus the
service level multiplied with the standard deviation of the demand. The service level is a dimension free
number and is represented by the value Z and can be calculated in excel (NORMSINV), see Figure 26.
Figure 26 Z value development
This formula to calculate the number of Kanban tickets in the front-end of the component supplier and
especially the Z-factor calculation presumes that the lead time and demand have a normal distribution.
The demand pattern as discussed in section 4.1 resulted in a 3-parameter Gamma distribution as best fit
but due to the limited controls as currently available resulted in many outliners in the data set. The data
as used in the analysis will not reflect the goods flow after implementing the controls as discussed;
therefore a normal demand distribution is assumed.
The model generates advice for the safety stock and the total number of Kanban tickets (= safety stock +
WIP) for a desired service degree to the back-end based on the demand (is S&OP) of the back-end. The
lead-time and the coefficient of variation are determined for the different product groups. The demand
pattern is unique for all products. The model combines the lead time and the demand pattern and uses
a Z value as service level indication, in this model 95%, 97.5% and 99%.
The results of the model are represented in Appendix F. These outcomes presume that it’s possible for
the front-end to serve the back-end based on daily scheduling /replenishment and a monthly review
process to align the demand uncertainties as used in the calculation (N-2) and the variance as it was in
practice during the last period (N-1). The average results for the data set are presented in Table 2.
0
0.5
1
1.5
2
2.5
3
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 0.995
Serv
ice
fac
tor
valu
e
Service factor (Z)
Z value development
Master thesis project – Roy Hartevelt Page 46
Table 2 Average stock level advice for data set
Dividing the safety stock production batches by the average production capacity of the back-end per day
gives the safety stock level in days. The Work-In-Process (WIP) based on the advice model is 106 Kanban
tickets. This part of the calculation is stable for all service levels due the fact that the WIP covers the
expected demand; the safety stock covers the uncertainties and the demand during lead-time. The total
stock (total number of Kanban tickets) is the sum of the safety stock and the WIP.
95% 97.50% 99%
Safety stock (tickets) 52 58 66
Safety stock (days) 5.2 5.8 6.6
WIP (tickets) 106 106 106
WIP (days) 10.6 10.6 10.6
Total stock (tickets) 158 164 172
Total stock (days) 15.8 16.4 17.2
Service level
Master thesis project – Roy Hartevelt Page 47
5 Design setup
5.1 Introduction This chapter presents the design setup for the model which optimizes product availability based on a
product family specific variance factor with attention to stock, service level and lead time. In the current
situation all SKU’s are treated similarly. Therefore an improvement opportunity is to adapt stock levels
per product family to fit better with the specific SKU family identities. The design of the model is
explained hereafter, starting with the supply chain performance indicators (section 4.2). Section 4.3
discusses the end-to-end risks in the supply chain and how they are related to the model. Section 4.4
elaborates on the input data for the design. Finally, section 4.5 discusses design setup conclusions.
5.2 Performance measurements Performance management (as discussed by W. Jammernegg ao in: Performance improvement of supply
chain processes by coordinated inventory and capacity management) illustrates how successful the
uncertainties are managed. All activities that are done, should improve service to our customers.
Increased performance of the component supplier has a positive effect on service and will therefore
potentially result in more sales and profit. Improved product availability through better balanced (to
cover the uncertainties) stock positions lead to improved end-to-end delivery reliability, expressed in
CLIP (Customer Line Item Performance) and CVP (Confirmed Volume Performance). The appendix on
page 79 shows the relation between stock levels on the one hand and CLIP and CVP on the other hand.
5.3 Uncertainty The relation between uncertainty within the supply chain, service and stock levels is represented in a
causal model (Figure 27).
Figure 27 Causal model of uncertainty and customer performance
CLIP
Stock levels at
manufacturer
+
Uncertainties
product
families-
Stock levels at
component
supplier
+
Responsivene
ss on demand
+
Service level
Demand
during lead-
time
Demand
variance
+
+
+
S&OP
reliability
+
Customer
demand
+
-
Production
lead-time
Service level
Supply
variance
+
+
+
Master thesis project – Roy Hartevelt Page 48
The component production and the related uncertainty are positioned upstream in the supply chain.
When these uncertainties increase, the delivery performance potentially decreases. As discussed, stock
levels influence the delivery performance to the assembly and test manufacturer. The stock levels both
at the component supplier and the assembly and test manufacturer are affected by desired service level,
lead time and (supply or demand) variance.
5.3.1 Replenishment strategy
A proper selection of a replenishment strategy is one of the keys to achieving low inventory while
maintaining high customer delivery performance (P. Suwanruji eo, Evaluating the effects of capacity
constraints and demand patterns on supply replenishment strategies, 2005). In Table 3 a comparison is
made between three replenishment strategies; using DRP /MRP (Distribution /Material Requirement
Planning), ROP (Re-Order-Point) and KBN (Kanban) were compared.
Using DRP/MRP strategy requires full visibility of order and inventory information across all locations in
the Supply Chain with periodic review. ROP is a reactive strategy with replenishment decisions based on
continuous review. KBN is a reactive replenishment strategy; inventory within a replenishment loop is
controlled by a fix number of cards. Kanban cards in a replenishment loop were allocated to inventory at
a location, inventory in transit to a location and unfilled orders placed to an upstream location.
The three replenishment strategies are compared under two levels of Manufacturing Constraints (MC)
No time-delay capacity constraint and Time-delay capacity constraint. No time-delay capacity constraint
ignores time delay operations by setting all setup and part processing times equal to zero
(manufacturing resources have no capacity constraint). The time-delay capacity constraint, production
resources at manufacturing are able to process one order at the time. A capacity constraint exists and
batches waiting to be processed must join the queue.
The second level of challenge is found in the Demand pattern (DP): level demand and seasonal demand.
During a level demand pattern the expected period demand was assumed to be stable through time for
each item. For the seasonal pattern, the demand pattern is assumed to follow a sinusoidal pattern with
a cycle length equal to one year.
Master thesis project – Roy Hartevelt Page 49
Manufacturing
constraint (MC)
Demand pattern
(DP)
Replenishment strategy (strgy)
DRP/MRP (strgy 1) ROP (strgy 2) KBN (strgy 3)
No capacity
constraint
Level demand
Responds slow to
changes introduced
by demand
uncertainty.
Responds
directly to
changes by
demand &
supply
uncertainties.
Results in
steadiest stream
of arrivals which
in turn results in
shorter average
queues.
Seasonal demand
Responds on
demand variation
including a season
correction based
on history and
forecast input.
Responds on
demand
variation
forecasted for
the next period.
Does not use
forecasting
information and
cannot utilize
backorders
information.
Capacity
constraint
Level demand
Effective at co-
coordinating
material flow and
limited waiting
time for the same
level of component
inventory
Score between
KBN and MRP –
managing the
material flow
and anticipates
on the demand
uncertainties.
Beneficial when
demand
uncertainty is of a
random nature,
results in
relatively low
inventory in front
of capacitated
resources
Seasonal demand
Responds on
demand variation
including a season
correction based
on history and
forecast input,
capacity limitations
are integrated.
Backorder
information is
taken into
account in the
next calculation
but frequency is
lower than at
the MRP
strategy.
No backorder
information is
maintained.
Table 3 Replenishment strategies (P. Suwanruji ao, 2005)
The green framework indicates the generic best suitable replenishment strategy per MC & DP
combination.
Master thesis project – Roy Hartevelt Page 50
Key characteristics of the Lumileds Supply Chain are directly linked to extreme business growth and the
many uncertainties in the current Supply Chain. Seasonal effects are not visible at the current state,
meaning no bullwhip effects of seasonal demand in the Supply Chain. This implies that for Philips
Lumileds ROP or Kanban replenishment strategy is most suitable for the Lumileds Supply Chain
(depending on the existence of capacity constraint).
To generate a stock advice, the advice model is divided in three sub models: the Kanban component
supplier, the economical stock value calculation and the replenishment model (Figure 28).
Figure 28 Concept advice model
5.3.2 Replenishment strategy component supplier
The upstream part of the chain is managed by kanban principle. Based on the main drivers as
mentioned in the box in Figure 17, the number of Kanban tickets is determined. Per product family the
value of the main drivers can be different. The performance of the Kanban concept is managed by daily,
weekly and monthly processes. Each period has a different focus; the daily process has a focus on
starting the right number of Kanban tickets per product. The weekly process has a focus on balancing
throughput time in relation with customer demand. The last process (the monthly) focuses on updating
supply variance, lead time, batch sizes and service levels. The uncertainty to manage is covered by the
supply variance, in the variance we have two parts lead-time and batch size. The different products are
divided in groups (product family, see appendix A) based on their uncertainty profile.
5.3.3 Re-order-point calculation
Calculating the re-order-point (ROP) is a monthly process. The S&OP update brings new input about
customer’s behavior. The ROP level consists out of two parts; the first part covers the demand during
lead time the second part covers the safety stock (see also par 2.5 Supply chain control models). The
uncertainty to manage is formed by the demand pattern as represented by the real shipments. The
demand and supply uncertainties are input to determine the necessary safety stock in the advice model.
Re-order-point
calculation
Replenishment
orders
Stock
levels
Manuf
acturer+/- +/-
S&OP
Supply variance
Demand variance
Moving average
Lead-time
Service level
Product life-cycle
phase
Stock take
WIP
In-Transit
ROP
Kanban
component
supplier
S&OP
Supply variance
Lead-time
Batch size
Service level
+/-
+/-
Master thesis project – Roy Hartevelt Page 51
5.3.4 Replenishment orders
Based on the boundaries as set in the ROP, the replenishment orders from the manufacturer to the
component supplier2 are calculated. The products on stock at the manufacturer are in-transit and in the
work-in-process are taken into account and are a minus on the calculated demand. The calculated
demand is based on a process step at component supplier1 of the assembly and test manufacturer. The
result of this action is a decrease of one week of stock to cover (see also Figure 29).
Week 1 Week 2 Week 3 Week 4
Assembly & test
manufacturer
Component supplier 1
Component supplier 2
Assembly & test
PP
PP
Production &
transport
Production & transport
PP = production plan
Difference in production lead
time to cover with stock
Figure 29 Planning of replenishment orders
5.4 Data The data used for this research is taken from the MES (Manufacturing Excellence System) of the
component supplier, the ERP system of the assembly and test manufacturer and the sales and
operations planning of the business unit Lumileds. The selected data represents the situation before the
results of the proposed solution were visible in the supply chain performance indicators of Lumileds.
4.4.1 MES
Data from the MES database is filtered on product family. The lead-time per process step is calculated
by subtracting all delivered production batches from the database. In the standard MES report the
throughput time per process element is measured in hours, the totals are expressed in days of
throughput time. All delivered batches (orders) in the selected period are taken into account no
exceptions. The data integrity for all figures is ensured by using production (MES) data integrated in the
ERP of Philips Electronics N.V.
5.4.2 ERP
Required data for calculating the ROP are the shipped products out of the assembly and test
manufacturer. Based on the shipped out data the demand variation is calculated. In the weekly
replenishment calculation a stock take out of the ERP system is copied into the model. The integrity of
the data is safeguarded by Finance and accounting of the assembly and test manufacturer; all system
mutations have financial consequences and for that controlled by F&A and periodically checked by an
independent accountant.
Master thesis project – Roy Hartevelt Page 52
5.4.3 S&OP
Customer demand is consolidated in an S&OP for the Lumileds business. A monthly update provides the
business new insight in the customer’s forecast with an outlook for 15 months. The supply chain
department of the business unit Lumileds is responsible for the content and reliability to translate sales
data into a demand plan. Based on the demand plan the supply plan is generated, a part of the overall
supply plan is for the component supplier. This component supply plan is the S&OP input for the
component supplier. The component supplier will use the S&OP as input for the day to day business and
on the other hand for mid- and long term capacity increase/decrease business decisions.
5.5 Conclusions Improving service through the Lumileds supply chain is the key objective of this research. Optimizing the
stock levels has a direct and positive effect on the customer service (CLIP). When the stock levels are
better adapted to the end-to-end uncertainties in the supply chain, the CLIP will improve because more
risks or uncertainties are covered.
The supply chain is represented as an uncertainty model. There are three types of uncertainties
influencing stock safety levels: at the component supplier lead-time and batch size variances and at the
assembly and test manufacturer the variance in customer demand. All those uncertainties including the
lead-time are influencing required stock levels at the component supplier and at the assembly and test
manufacturer. A lead-time means a higher stock level to handle the demand during lead-time to
guarantee delivery performance. Reducing lead time will directly effect in lower required stock levels.
Chapter four (partly) answered sub question 2 (which planning model is most suitable for the supply
chain).
Master thesis project – Roy Hartevelt Page 53
6 Model design
6.1 Introduction model design In the previous chapters the design objectives and boundaries for the model are discussed in depth. In
the next section the structure of the model, based on literature, is explained and there is some attention
to the relation between the data and the choices of the model structure. The chapter ends with a
description of an excel tool which will be used to determine the optimal Re-Order-Point and safety stock
levels. This tool is developed together with field experts to ensure the practical workability of the tool
and the acceptation of the end users.
6.2 Evaluation design objective First point of attention is to challenge the accuracy of the specified requirements, the quality of the
model and the acceptance during the hand-over to operations at the end of the project depends greatly
on the investment during this activity (Lindland et al., 1994). The first step in the design phase is the
setup of a conceptual model. This is taken in section 4.3. The system and its environment are identified
and the mutual relations are determined (Figure 27). The relations are translated into a conceptual
model, wherein the three sub models are represented (Figure 28). In the analysis phase the design
objectives are identified.
1. Design a model to optimize the stock positions throughout the Lumileds Supply Chain, based on
product specific Supply chain uncertainties.
2. Design a model to improve the service level of the Lumileds Supply Chain.
The advice model has a focus on these two objectives and the success depends strongly on the degree
of fulfillment of these two objectives.
6.3 Model design based on the theory As described in Figure 28 the overall solution is designed in 3 sub models; one model for the Re-Order-
Point calculation, a model for replenishment determination and a Kanban model for balancing the
production of the component supplier (chapter 5).
6.4 Practical design and model result Based on the inputs of chapter 4 (design setup) a model is designed as a pilot. This first design was built
based on the requirements gathered during interviews and workshops. After evaluation a next version
of the models were created. Finally, with input of field experts the models and the results out of the
pilot were created. Figure 30 shows the first level of the IDEF0 model of the designed situation. Twelve
input data flows, six control flows, one supporting mechanism and one output dataflow are the number
of instruments used in the models. In the next paragraphs the design of the models is discussed based
on the IDEF0 model.
Master thesis project – Roy Hartevelt Page 54
Figure 30 Model design (IDEF0 level 0)
Figure 31 shows the supply chain control model of the component supplier. The model generates advice
for several topics needed at the different stages in the chain. The final result of the integrated models is
a consolidated stock data overview based on re-order-point levels. The re-order-point calculation is a
monthly process based on the boundaries as set during the monthly cycle. The replenishment orders will
be generated in the weekly cycle. The Kanban calculation of the front-end of the component supplier
takes the demand forecast out of the re-order-point calculation.
In the next sections the model is designed and built to satisfy the design objectives within the design
boundaries.
Supply chain planning model
A0
Pe
riod
MA
T
Pro
d. L
ife C
ycle
Se
rvic
e le
ve
l
Ca
p. c
on
stra
ints
Ca
rrier le
ad-tim
e
S&OP
Shipments
Lead-time platelets
Prod. Fam. table
SM comp supl.
GR prod order
GGI starts
Yield
Stock overview
In-transit
Platelet distribution
WIP comp. Supl.
ME
S
Consolidated stock
overview
Master thesis project – Roy Hartevelt Page 55
Figure 31 Model design (IDEF0 level 1)
6.4.1 Monthly ROP calculation
The monthly re-order-point calculation (A1 in Figure 31) generates the new or updated Re-order-point
values. The inputs for calculating the re-order-point levels are the platelet distribution, the S&OP, the
actual shipments of the last 60 days, the lead-time of the back-end of the component supplier including
the carrier lead-time and the product family table. In the product family table the decision is taken if a
product is able to move over from a manual to an automated stock calculation method. The coefficient
of variation (CV) is the decision making qualifier. If the CV value is above 1; manual intervention is
necessary in the economical stock level calculation. If the value is below 1; no manual intervention is
required, the model will take care about the uncertainties in the supply chain. There are several ways to
determine the ROP level (as partly discussed in chapter 2.5). The most common, simple way to
determine the re-order-point level is described with the formula (Charles Atkinsin, 2005):
[6]
ROP = Re-Order-Point
DEM = Demand (weekly)
LT1 = Median lead-time back-end component supplier including transport time
Z = Service level
LT2 = Median LT1 – lead time GGI/Saber
Master thesis project – Roy Hartevelt Page 56
Dvar = Demand variation
Svar = Supply variation
The ROP is determined by two parts: the first part is for covering the demand during lead-time, the
second part is for covering the uncertainties in the chain (safety stock). The safety stock is determined
by the service factor Z. This number represent S the required service level and can be calculated in excel
(NORMSINV). Figure 32 shows the results of the Re-Order-Point calculation model.
The ROP (Re-Order-Point) is the sum of Demand during lead-time (Dem LT) and the Safety Stock. The
Demand during Lead-time is a calculation of the Weekly demand and LT1 (lead-time back-end
component supplier + transport lead-time + replenishment lead-time). LT2 is used in the safety stock
calculation and is used together with LT1, LT2 stands to the difference between the lead-time demand
and the GGI starts (gold to gold interconnect process: start production process marked as leading within
the SC of Lumileds). The supply variance as used in the safety stock calculation is based on the supply
standard deviation related to throughput-time and the supply standard deviation related to quantities.
The demand variance is calculated with data based on products shipped, with the assumption that the
demand variance of the past is comparable with the demand variance of the future. The Z value is
directly related to the service level as indicated in the last column. The yellow fields are manual input
possibilities and will be updated by the production planners of the component supplier and the
manufacturer. The product family as mentioned in figure 32 is the total product portfolio of the lumileds
component supplier. Figure 32 reflects the ROP calculation for January 2011.
Figure 32 ROP calculation model
6.4.2 Weekly replenishment
The weekly replenishment calculation (A2 in Figure 31) is calculated based on the boundaries as set in
A1 in Figure 31 (the ROP). The input data flows are needed to calculate the needs for the coming period.
A common and simple way to determine the replenishment levels is described with the formula:
ROP calculation sheet(MAT)
Product Line Bin ROP Dem LT. Safety stock Weekly dem. LT 1 LT 2 Suppl var. Dem var. Z Safety[wks] Avg stock [wks]Supply Std dev BESupply std dev # transit fileLT BE LT GGI Service l.Product family A
330 479,536 307,717 171,819 137,199 2.24 1.00 0.48 0.59 1.64 1.25 1.75 0.46 0.15 1.24 1.71 95%
440 418,344 280,277 138,067 137,199 2.04 1.00 0.22 0.59 1.64 1.01 1.51 0.17 0.14 1.04 1.71 95%
Product family B
222 - - - - 2.43 1.00 0.74 0.48 1.64 0.72 0.18 1.43 1.71 95%
333 4,679,922 3,083,530 1,596,392 1,438,981 2.14 1.00 0.30 0.48 2.05 1.11 1.61 0.24 0.18 1.14 1.71 98%
444 - - - - 1.00 1.00 0.11 0.48 1.64 - 0.11 - 1.71 95%
555 10,988,393 7,338,801 3,649,592 3,357,621 2.19 1.00 0.31 0.48 2.05 1.09 1.59 0.21 0.22 1.19 1.71 98%
Product family C
241 39,462 23,018 16,444 6,975 3.30 1.59 0.52 0.98 1.64 2.36 2.86 0.46 0.24 2.30 1.71 95%
242 39,462 23,018 16,444 6,975 3.30 1.59 0.52 0.98 1.64 2.36 2.86 0.46 0.24 2.30 1.71 95%
Product family D
111 42,374 25,958 16,416 8,259 3.14 1.43 0.15 1.00 1.64 1.99 2.49 0.12 0.09 2.14 1.71 95%
222 2,491,901 1,275,450 1,216,451 368,932 3.46 1.74 0.40 1.45 1.64 3.30 3.80 0.34 0.22 2.46 1.71 95%
333 - - - - 3.61 1.90 1.20 1.45 1.64 1.20 0.10 2.61 1.71 95%
444 104,250 63,865 40,385 20,601 3.10 1.39 0.23 1.00 1.64 1.96 2.46 0.13 0.19 2.10 1.71 95%
555 1,614,924 1,096,981 517,943 298,789 3.67 1.96 0.23 0.74 1.64 1.73 2.23 0.09 0.21 2.67 1.71 95%
666 - - - - 3.63 1.91 0.41 0.74 1.64 0.40 0.10 2.63 1.71 95%
777 226,717 168,586 58,131 64,841 2.60 1.00 0.19 0.52 1.64 0.90 1.40 0.18 0.06 1.60 1.71 95%
Product family E
1820 - - - - 2.48 1.00 0.85 0.55 1.64 0.37 0.77 1.48 1.71 95%
1813 325,994 235,151 90,843 96,430 2.44 1.00 0.58 0.55 1.64 0.94 1.44 0.17 0.55 1.44 1.71 95%
1806 1,685,156 1,135,556 549,599 341,888 3.32 1.61 0.59 0.55 1.64 1.61 2.11 0.43 0.40 2.32 1.71 95%
Product family F
999 1,229,283 575,327 653,956 231,720 2.48 1.00 0.56 1.68 1.64 2.82 3.32 0.35 0.44 1.48 1.71 95%
Product family G
444 569,550 334,922 234,628 81,688 4.10 2.39 0.41 1.00 1.64 2.87 3.37 0.34 0.22 3.10 1.71 95%
555 141,882 83,439 58,443 20,422 4.09 2.37 0.41 1.00 1.64 2.86 3.36 0.34 0.22 3.09 1.71 95%
Total 25,084,392 16,054,333 9,030,059 6,621,257
Master thesis project – Roy Hartevelt Page 57
[7]
Repl = Replenishment order to the component supplier
ROP = Re-Order-Point level
WIP = Work-in-process back-end component supplier
InT = In-Transit between component supplier and the LED assembly and test manufacturer
Stock = Physical stock availability at the LED assembly and test manufacturer
Prod = Production as planned for the selected time period
Figure 33 shows the results of the replenishment calculation model for the last quarter of 2010 (the first
full quarter that the model is fully used in the Lumileds organization), the selected products are
reflecting the overall way of working.
Figure 33 Weekly replenishment calculation
Figure 33 shows the weekly replenishment quantities per product type. Per week Usage, Supply Pipeline
Economical stock, Economical stock end of week and on-hand end of week are calculated. The usage is
based on the production plan of another key supplier as discussed in 5.3.4 Replenishment orders, in
case no starts are planned at the GGI, the ‘Buildplan /S&OP’ will be taken as guideline. At the supply
side, ‘Planning Mhz’ indicates the replenishment order based on the calculation as discussed. The
‘confirmed plan Mhz’ are the quantities as planned at the component supplier. The difference between
‘planning Mhz’ and ‘Confirmed plan Mhz’ is related to extra information as available in the chain of
Lumileds (be. Expected extra sales not yet integrated in the S&OP or a possible temporary shutdown at
the component supplier start 2011). The ‘Pipeline Econ. Stock’ is a summary of the work in process at
the component supplier (back-end), ‘goods in transit’ and the components as available at the
manufacturer. The ‘Pipeline econ. Stock’ is summarized as ‘Econ Stock (qty)’ and weeks of forecasted
consumption. In case the ‘Econ Stock (wks)’ is lower than two weeks this fields will be red colored. The
21 Replenishment Calculation based on GGI Starts
Wknr 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
Product family B, BIN 333
Backlog 0 0 0 0 0 0 0 0 0 0 0 0 0
Batchsize GGI Starts SGP 1,680,000 1,680,000 1,924,128 2,100,000 1,911,552 2,090,000 1,815,000 2,144,000 2,398,050 2,200,000 2,700,000 2,100,000 2,452,320
20960 Buildplan/S&OP GGI 1,512,000 1,512,000 1,731,715 1,890,000 1,720,397 1,881,000 1,633,500 1,929,600 2,158,245 1,980,000 2,430,000 1,890,000 2,207,088
90.0% Actual Pl/Attach
ROP 5,796,082 5,796,082 5,796,082 5,796,082 5,796,082 4,817,786 4,817,786 4,817,786 4,817,786 6,222,473 6,222,473 4,977,979 4,977,979
Batchsize Planning Mhz 1,416,000 1,062,000 944,000 944,000 236,000 0 0 0 0 0 0 0 0
118000 Confirmed plan Mhz 2,580,000 2,300,000 2,340,000 2,400,000 1,845,000 3,132,000 2,668,000 1,740,000 1,392,000 2,088,000 0 0 1,190,000
WIP Mhz 1,548,000 516,000 615,000 1,032,000 1,722,000 1,548,000 903,000 1,392,000 1,624,000 580,000 928,000 928,000 100,000
In-Transit 1,653,648 1,779,774 2,244,509 1,392,272 1,190,373 1,670,764 2,327,476 1,567,587 1,156,905 1,909,011 1,635,070 1,676,891 222,184
Stock take Penang 2,724,371 4,020,741 3,746,909 4,318,502 4,406,332 4,854,873 5,282,569 5,873,545 6,972,546 7,878,446 7,834,441 7,389,441 8,068,381
Econ. Stock (qty) 6,994,019 7,104,515 7,214,703 7,252,774 7,443,308 9,324,637 9,547,545 8,643,532 8,987,206 10,475,457 7,967,511 8,104,332 7,373,477
Econ. Stock (wks) 4.1 4.2 4.3 4.4 4.5 5.7 5.8 5.3 5.7 6.7 5.3 5.5 5.2
On-hand + InTransit 2,495,543 3,923,055 4,479,418 4,198,774 3,864,990 4,635,637 5,889,648 5,560,132 6,495,951 7,857,857 7,311,266 7,086,332 5,860,565
Product family B, BIN 555
Backlog 0 0 0 0 0 0 0 0 0 0 0 0 0
Batchsize GGI Starts SGP 1,120,000 1,120,000 1,282,752 1,400,000 1,400,000 1,482,000 1,320,000 2,600,000 2,930,950 2,500,000 2,700,000 2,100,000 2,431,360
20960 Buildplan/S&OP GGI 1,008,000 1,008,000 1,154,477 1,260,000 1,260,000 1,333,800 1,188,000 2,340,000 2,637,855 2,250,000 2,430,000 1,890,000 2,188,224
90.0% Actual Pl/Attach
ROP 3,932,191 3,932,191 3,932,191 3,932,191 3,932,191 3,554,717 3,554,717 3,554,717 3,554,717 6,473,904 6,473,904 7,768,685 7,768,685
Batchsize Planning Mhz 236,000 0 826,000 0 0 0 0 0 0 590,000 1,180,000 3,422,000 1,888,000
118000 Confirmed plan Mhz 1,677,000 1,700,000 1,710,000 2,200,000 1,368,000 1,710,000 1,710,000 2,166,000 2,736,000 2,280,000 2,808,000 2,574,000 2,925,000
WIP Mhz 0 1,677,000 615,000 903,000 1,161,000 1,032,000 1,419,000 1,140,000 1,140,000 2,052,000 1,856,000 812,000 1,972,000
In-Transit 2,222,499 81,398 1,241,114 1,590,318 1,223,946 2,127,451 511,366 1,143,207 1,082,338 1,224,475 1,973,875 2,231,481 2,153,156
Stock take Penang 2,554,185 3,185,424 2,491,348 3,711,629 4,260,958 3,952,499 4,886,614 4,461,075 4,280,424 4,871,116 3,931,361 3,228,353 3,962,697
Econ. Stock (qty) 5,445,684 5,635,822 4,902,985 7,144,947 6,753,904 7,488,150 7,338,980 6,570,282 6,600,907 8,177,591 8,139,236 6,955,834 8,824,629
Econ. Stock (wks) 2.4 2.4 2.0 2.8 2.5 2.7 2.6 2.3 2.3 2.8 2.8 2.3 2.9
On-hand + InTransit 3,678,573 2,011,838 2,724,462 4,293,947 4,330,427 4,819,950 4,137,980 4,270,482 4,174,762 3,755,591 3,267,381 3,209,834 3,685,853
Oct-10 Nov-10 Dec-10
Usage
Supply
Pipeline Econ.
stock
Econ. stock end of week
On-hand end of week
Usage
Supply
Pipeline Econ.
stock
Econ. stock end of week
On-hand end of week
Master thesis project – Roy Hartevelt Page 58
last row indicates the number of products on stock at the manufacturer including the number of
products in transit.
6.4.3 WIP back-end component supplier
The Work-In-Process (WIP) (A3 in Figure 31) is a result of the outcome of the weekly replenishment
calculation. Based on the replenishment calculation the production capacity is balanced over the several
product types. In case the total request is higher than the production capacity in that period a selection
out of the requested plan is made after alignment between the manufacturer and the component
supplier. After alignment the confirmed plan is entered in the model. The supermarket (the customer
decoupling point) is not part of the back-end or calculated in the WIP. One of the parameters for
calculating the safety stock is the lead time. In case the Kanban of the component supplier is not on
target level the lead-time for the back-end will be extend with the lead-time of the front-end. The
second pillar in the safety stock calculation is the uncertainty; in case the supermarket is not meeting
the required standards as calculated in the Kanban system the CV value in the SS calculation should be
revised too.
6.4.4 Generate In-transit overview
The in-transit (A4 in Figure 31) file is an in-between overview for managing the goods flow between
locations. A production batch is in-transit in case a product has left the factory of the component
supplier, but not yet received at the assembly and test manufacturer.
6.4.5 Calculate platelet stock position
Based on the ROP calculation the stock positions at the assembly and test manufacturer are set (A5 in
Figure 31). The stock transactions are administratively managed via an ERP system, the goods received
transaction finishes the in-transit phase (status).
6.5 Results Implementing the new way of working based on the design as written in this master thesis results in a
43.5 M pieces stock reduction (based on 12 M sales per week). The 43.5 M pieces of stock correspond
with 720 K Euro. This enormous stock reduction is a result of two main achievements: lead-time and
variance reduction. With the introduction of the supermarket at the component supplier (customer
decoupling point) and the use of the available supply chain information the reaction time of the
component supplier is reduced from 12 to 1.3 weeks. The variance reduction is mainly caused by
introduction of scheduling rules at the component supplier (first in first out) at the work centers and
buffer lanes. Secondly, one production step at the back-end of the component supplier is outsourced.
Updating the service level agreement with the supplier with logistic parameters like throughput-time,
pick-up /delivery schedules and a day to day communication structure including information sharing
about the midterm (coming 3 months) production schedule reduces supply chain variance. Finally, the
awareness of the effects of being non predictable at the component supplier in terms of stock levels at
the manufacturer brought an focus on doing the right things at the right time according the agreed plan
at the component supplier and the manufacturer as well.
Master thesis project – Roy Hartevelt Page 59
Setting up a control model based on design requirements as set by the organization and the business
characteristics ends with tools to manage the business in a proper way. At the other hand models
without embedded processes at the involved organizations will not bring success after closing the
project (Lee H.L a.o 1992). Figure 34 shows the implemented processes with the designed control model
as backbone and a meeting structure based on terms of references (TOR). The backbone is at the left
side of Figure 34, the monthly and weekly planning sheets are integrated in an overall control process
based on a fix structure. The daily execution is the result of the day to day scheduling based on the
weekly replenishment orders. The backbone starts with a forecast (Sales and Operations Planning),
based on the forecast a production planning is made. This production plan includes also the
performance of the production processes in the period before. Control loops are the linking pins
between the several meetings at monthly /weekly and daily rhythm as well. The performance input is a
bottom up process meanwhile the planning guidelines are top down managed. With the designed
models and the supporting processes supply chain becomes a competitive edge (making the difference).
Figure 34 Planning rules and deploy control model
6.6 Conclusion model design The advice models (ROP, Replenishment, Kanban) are designed to manage specific risks (lead-time and
uncertainties). The models generate advice based on these risks for a desired service level. The desired
service level is expressed in the Z factor. The models give product specific advice for safety stock,
economical stock levels and quantities to produce based on the requested service level and demand.
The outcomes of the models are shown in appendix F, G and H. The outcomes of the model have an
average safety stock of 1.85 week at the assembly and test manufacturer for a service level of 95%. For a
service level of 99% a safety stock of 2.6 weeks is required with a total average stock of 3.1 weeks. The
results of the advice model indicate that a product safety stock level has potential to optimize to the
most suitable stock positions for the Lumileds supply chain at Philips: meaning high customer service
with the right underpinning stock levels in the chain.
Activity
MEETING
MEETING
TOR
TOR
TOR
Day
Week
Month
Production orders
Plan
SCM improvementactions
MonthreportKPI’s
P
D
C
A
WeekreportKPI’s
Daily execution
Review KPI
S&OP ROP settings
DayreportKPI’s
Control ReportPlanForecast
ForecastHistoric sales
Updated
ROP
P
D
C
A
P
D
C
A
Master thesis project – Roy Hartevelt Page 60
Figure 35 shows the ROP development from a financial point of view. The difference between a 90%
service level and a 99.5% service level results in a 23% stock cost level increase with a stable demand
during lead-time and demand and supply variances.
Figure 35 ROP development financially
Finally, with chapter six sub question two (which planning model is most suitable for the LED supply
chain) is answered.
0%
20%
40%
60%
80%
100%
120%
140%
90.0% 92.5% 95.0% 97.5% 99.5%
In K
Eu
ro (
%)
Service level
ROP development
Safety stock
Demand L-T
Master thesis project – Roy Hartevelt Page 61
7 Summary design and modeling phase In this phase a stock level advice model is designed and modeled based on the design objectives and
boundaries as determined in the analysis phase. The Lumileds supply chain is represented as a model
based on three different replenishment strategies; Kanban, Re-Order-Point and Replenishment
calculations. The front-end (all production processes before the customer order decoupling point) of the
component supplier is optimized with the Kanban replenishment strategy. Per product group demand is
translated into Kanban tickets and enabling that the back-end is served with a 99% service level. The
boundaries as set in the Re-Order-Point calculation are leading in the replenishment calculation. Due to
reliable upfront information the inventory required during order lead-time is minimized. The challenge
for the component supplier is to minimize the back-end lead time in such a way that production and
transport lead time is shorter than the throughput time of the parallel process of one of the other key
suppliers of the LED manufacturer.
Designing a most suitable supply chain concept for the component supplier was complex due the fact
that a mass production environment is integrated with a development centre. Production and
development have by nature some conflicting objectives. The supply chain concept must be such that
both are optimally satisfied.
In the analysis phase the design objectives for the model are determined. The realization of these
objectives can be evaluated after the design and modeling phase (Table 1 requirements, criteria,
constraints & design option stakeholders).
With the input of the production planner of the component supplier the supply chain stock
optimization tool is directly linked with the front-end of the component factory. The integration
of the two planning tools enables a minimum of transaction time and is a pro in process
optimization.
One single source for the forecast of the entire supply chain was a key issue for the planner of
the LED manufacturer. With this basic rule all forecast models should use the same basic
information. The lead-time, variance and uncertainties should be worked out as a variable and
are a changeable qualifier in the model.
Design, build and implementation time were important boundaries for the improvement
manager. Secondly the model should be stable and predictable (easy to understand) and finally
calculations of the previous periods should be stored in the model as well.
Simulating different scenarios and reporting were the key topics of the material manager at the
LED manufacturer.
All design requirements and criteria can be evaluated with the model results. The success of the
models is strongly determined by the use in the standing organization. Monthly and weekly
meetings are planned and fully integrated in the business processes of the LED supply chain.
Master thesis project – Roy Hartevelt Page 62
The model designs are based on data of the component supplier and the LED assembly & test
manufacturer. The tools are handed over during three sessions to the end users (the planners at the
component supplier and the manufacturer) and next to this a manual and instructions which will
improve the usability. Possibilities to change /add /remove product characteristics increase the
generality of the tool and give the end user options to make adaptations when necessary. Appendix I
and J show the key information as used in the workshops to introduce the new way of working.
The second phase of design and modeling is finished. In the next and last phase the theoretical models
are transferred into sustainable decision taking models. The last phase ends with conclusions and
recommendations for the future.
Master thesis project – Roy Hartevelt Page 63
Evaluation and validation phase
The design and the models are setup in the analysis, design and modeling phase. The next step is to
evaluate the process and to evaluate in what degree the research has given insights in the research
objectives and questions. In chapter 8 the replenishment strategy models are discussed and
recommendations out of the translation into practice are indicated. The last chapter (9) is reserved for a
validation of the research questions including a personal reflection.
Master thesis project – Roy Hartevelt Page 64
8 From model to practice
8.1 Introduction The designed models are capable to generate a product specific stock level advice including the resulting
replenishment orders or number of Kanban tickets all based on a ROP calculations. The supply chain
uncertainties are covered in the ROP calculation. The outcomes of the models can be used to adapt the
current stock levels to the uncertainties in the supply chain. Better allocation of stock levels can
decrease the total stock level; shifting performance to a new level can improve service. The model
indicates these possibilities theoretically, but to ensure results the model must also be translated into
practice. This chapter describes this translation from model to practice and indicates possible obstacles
and chances. In section 8.2 there is attention to the generality of the model for other situations. Is it
possible to use the models for other situations and which problems can be expected? The production
planner at the manufacturer and at the component supplier both have a very important role in the
translation of the model in practice and must be convinced of the validity and use of the models, as the
conclusions of the stakeholder analysis show. In section 8.3 more about the pilot phase during model
introduction will be discussed. The relation and tension between these stakeholders is part of this
discussion.
8.2 General usability of the model The main goal of this research and the purpose of the model are to generate stock advice, based on lead
time and uncertainties in the supply chain. Combining information about the total supply chain of
Lumileds can help to find an optimal allocation of stocks. With this reallocation a higher service to the
next step in the Supply Chain can be achieved. Besides the main goal, the research and the models are
also a next step in a continuous improvement program to improve ‘customer’ service and lower the
costs. Openness in the Supply Chain is a first step to improve supply chain processes in a bigger context
and is a first step in the direction of collaborative planning. Actually, those models can be used as
support for further cooperation and openness for overall supply chain improvements.
The advice models have a threefold practical purpose:
1. Providing a product specific Kanban advice for the front end of the component supplier.
2. Proving a product specific replenishment stock advice model between the component supplier and
the manufacturer.
3. Supporting tool for further supply chain optimization of the Lumileds supply chain (based on re-
order-point calculation).
Transparency and cooperation between a manufacturer and component supplier have a definite
potential for supply chain optimization. Shifting risks and opportunistic behavior can be a result of the
openness, in the end the supply chain fits in one company, those risks have limited impact at the
bottom-line.
Master thesis project – Roy Hartevelt Page 65
This specific Supply chain optimization issue of Philips Lumileds is, at the end, solved with a standard
design approach based on the theory of Herder and Stikkelman (2004). The models as used within this
assignment are based on proven knowledge and customized to the Lumileds business situation. The
specific elements in this case are the organizational setting and the interdependencies between the
assembly and test manufacturer with the key component suppliers. In this situation the key component
suppliers are all part of one company. Due to the intercompany relationship between the suppliers and
the manufacturer the way of managing a sub contractor wasn’t strictly (according agreed guidelines
saved in a contract) arranged in execution from the past. Creating meeting rules (TOR, RACI) and setting
up decision criteria when calculating the ROP or replenishment orders were part of the maturity shift of
the Lumileds organization. The maturity shift was a need to implement a structured way of working
based on companywide agreements.
The applicability of the design model in industries outside Lumileds with or without seasonality is
enormous due the fact that the models are running on two planning cycles, the monthly and the weekly
planning cycle. The monthly planning cycle brings an outlook for the coming four quarters and based on
the MAT (moving average total) the demand as used in the next calculation steps takes already future
development into account. Based on the calculated system boundaries in the monthly planning process
the weekly replenishment orders are calculated between a component supplier and the manufacturer.
The setup with the monthly and the weekly planning cycles makes the design model as a generic
applicable supply chain control model for the product creation industry.
8.3 Using the model for a pilot
The advice models are based on theory and input from the component supplier and the LED assembly
and test manufacturer. Those models and the research where the models are based on, has provided
insights in the supply chain uncertainties. To improve the suitability for Lumileds the model is extended
into a tool. This tool fits well with the current supply chain processes of Lumileds and makes is possible
to adapt product specific characteristics.
Product and characteristics can change over time and maintenance is necessary to keep the tool useful.
The tools are handed over to the production planners of the component supplier and the LED
manufacturer. During workshops the handover was formalized in combination with a manual on how to
use and maintain the tools in the future. The manual describes the design of and theory behind the
replenishment strategy models and also explains how to make changes.
During the pilot phase (1 month) the models were used with 2 product families. Based on the
experiences of the field experts modifications were made before we extend the pilot with another
month. During the second pilot 4 extra product families were included. Based on the results out of the
second pilot the models were finalized and handover to operations ready. The last part of the hand-over
to operations was to imbed the models in the standard organization. Term-Of-References (TOR) were
made for creating meeting guidelines and meeting inputs & outputs /results. With the help of the TOR a
structure was set for having efficient and fruitful weekly and monthly alignment meetings. As a direct
Master thesis project – Roy Hartevelt Page 66
effect of the transparent meeting structure thinking in improvement opportunities were embed in the
day to day business easily.
Effects of the pilot on the stock positions were directly visible. Due to the information sharing and the
reduced lead-time no production was necessary from the component supplier in the first week after the
changeover (start of the pilot). After the first week the replenishment values were positive, meaning
new components were needed to fill the stock positions to the requested levels. After each introduction
or update of the ROP calculation the stock positions were discussed with the stakeholders during the
ROP meeting, actions were taken in the replenishment meeting. With this setup the content discussions
were based on boundaries as set during the ROP meeting. The ROP meeting was full of discussions
about demand and supply variances, batches sizes (yield), throughput-time and the total demand plan
for the coming period.
The design model of Herder and Stikkelman (2004) has a focus on design and less at the implementation
part of a project. The assignment of this master thesis was to design and to implement supply chain
controls. For that reasons the design model of Herder & Stikkelman (2004) was enriched with the
approaches of Shannon (1998) and Lee & Billington (1992) to enable the project to use an integrated
approach from the start of the project till implementation and finally a sound handover to operations.
The scope of the approach of Herderand Stikkelman (2004) is unlimited: additional building blocks can
be added (f.e. number of iterations) when desired. The flexibility of the model and on the other hand
the step by step approach was useful during the design and development phase of the projects. The
focus on the design parameters and the actor analysis during and before the design phase enables a
smooth implementation. The lack of structure in the implementation part of a model is the weak point
of the approach of Herder and Stikkelman (2004). No structure or implementation guidelines are
provided in the approach. For that reason additional research was necessary for closing the gap and to
run the project on a proven knowledge based design and implementation approach. For next design and
implementation projects the opportunity is there to enrich and integrate the design model of Herder
and Stikkelman (2004) with a standard implementation approach of (change) processes. The integration
of a design and implementation model was out of scope of this research /master thesis.
Table 1 shows the requirements of the most important stakeholders. Based on the stakeholders input
and the characteristics (no seasonality, fast growing and demanding customers) of the Lumileds
business the design phase was started. Evaluating the model characteristics based on the requirements
of the stakeholders (as provided by table 1), the implemented supply chain control model and the
related business control processes are from all perspectives better than initial requested. Strategic
behavior of one of the key stakeholders can always change the input of table 1, but will not harm the
designed and implemented control model because the business characteristics of the LED industry were
the first attention points; secondly the requirements of the stakeholders were checked on completeness
necessary for acceptation and hand-over to operations. In case a key stakeholder will change their
mindset it is not that difficult to change some output parameters, in this: changing a mindset will not
change a business setup.
Master thesis project – Roy Hartevelt Page 67
8.4 Summary from model to practice In this chapter the translation from the model into practice is described. The designs of the models are
based on theory and implemented with the field experts within the Lumileds supply chain. The designed
models are turned into tools for Philips Lumileds. Besides the several hand-over sessions a tutorial is
written were all possibilities to change /added /delete product are described.
Within this chapter sub question 6 (What actions are recommended to convince the actors about the
benefits of the new supply chain planning model and way of working) is answered. Using a pilot for a
limited part of the business was the right way to do in convincing stakeholders. With the results of the
pilot it was easy to share the benefits of an optimal supply chain control model.
Master thesis project – Roy Hartevelt Page 68
9 Conclusion and recommendations
9.1 Introduction In the previous chapter the model design is evaluated and the model results are validated. In this
chapter conclusions are drawn up based on the results of this research. The research questions
determined the direction and structure of the research. In the conclusions answers are provided on the
research questions. The research questions therefore form the structure of the conclusions section (9.2).
Recommendations are provided in section 9.3 for aspects out of the research scope and aspects which
deserve further research.
9.2 Conclusions of the research Supply chain optimization based on customer service and cost minimization is the most important
drivers for Philips Lumileds. Improving the service will result in fewer out-of-stocks at the back-end of
the component supplier and the manufacturer. This research has one objective.
Generate, implement and deploy supply chain planning controls for shop floor scheduling and
material replenishment for the LED products between a component supplier and the LED assembly and
test manufacturer.
The stock level optimization results in better service in the supply chain, which can be measured in cycle
time, variance and availability. The ROP model can be used for realizing service improvement by
managing the stock levels accordingly. The performed research gives an answer on the main research
question:
WHICH SUPPLY CHAIN PLANNING CONTROL IS NEEDED FOR AN MOST SUITABLE STOCK SITUATION TO SECURE THE
SAFETY STOCK LEVELS BETWEEN THE COMPONENT SUPPLIER AND THE LED ASSEMBLY & TEST MANUFACTURER.
Before an answer on the main question can be formulated, first the sub-questions will be answered,
based on the performed research. The sub-questions as set in section 1.2.3 will now be discussed and
answered systematically.
1. Which facts determine the most suitable supply chain planning method?
In the current situation the LED assembly and test manufacturer send in orders every week. The
order frequency is based on the demand pattern of the industrial customers of the manufacturer.
The difference per product makes that the models are generating a separate and specific advice per
product per part of the Supply chain as indicated in Figure 17 (section 5.1). The safety stock levels
are not only determined by uncertainties but also by the desired service level. In this research the
models generates advice for a service level of 97.5% (default) but can be manual changed. Searching
for the most suitable planning model is for the total Lumileds Supply chain ends in a model
configuration of best practices per part of the chain. A Kanban replenishment strategy system for
high volume and low costs products, replenishment orders based on supply chain stock availability
Master thesis project – Roy Hartevelt Page 69
to supply the manufacturer and an overall control based on Re-Order-Point calculation. Main inputs
for the three models are the Sales & Operations Planning, supply chain lead-times and the various
supply chain uncertainties.
2. Which planning model is most suitable for the Lumileds supply chain?
A most suitable planning method is determined by several aspects. Managing stock levels and Supply
Chain uncertainties are two aspects to cover. The Lumileds supply chain can be grouped into three sub
models, which are all influencing the stock positions at the LED assembly & test manufacturer (section
4.3).
Figure 36 Concept advice model
These three sub models in combination with a desired service level in the Supply Chain determine the
stock levels at the LED assembly & test manufacturer. The front-end (the processes at the component
supplier before the decoupling point) is controlled by the forecasted demand pattern translated into a
Kanban calculation at article level. Uncertainties (demand, lead-time and batch size) in this part of the
model are translated into factors in the formula. In the Re-Order-Point calculation the boundaries are
set for the overall component flow, meaning; replenishment levels with respect to the work in process
at the component supplier after the decoupling point and the in-transit. Uncertainties covered in the
ROP calculation are; demand variance, supply lead-time variance and supply quantity variance. In the
weekly replenishment process the output of calculations are used as set in the ROP. The supermarket
between the front- and back-end (decoupling-point) guarantees a 99% availability of available products
in relation with the replenishment orders. Based on literature and experts a list is created of
characteristics which possibly influence the choice of a supply chain control method.
The major characteristics which determine the planning method are:
Supply chain lead-time
Volume
Responsiveness component supplier
Demand variation
Supply variation
Re-order-point
calculation
Replenishment
orders
Stock
levels
Manuf
acturer+/- +/-
S&OP
Supply variance
Demand variance
Moving average
Lead-time
Service level
Product life-cycle
phase
Stock take
WIP
In-Transit
ROP
Kanban
component
supplier
S&OP
Supply variance
Lead-time
Batch size
Service level
+/-
+/-
Master thesis project – Roy Hartevelt Page 70
Stock value
3. Which processes are related to the supply chain planning processes and what are the
interdependencies?
To generate product specific advice the overall planning cycle should be incorporated. Figure 37 shows
the relationship between the two main component suppliers of the LED assembly and test manufacturer
which shows a (partly) parallel planning opportunity. Making component supplier1 leading in the Supply
chain saves 10 production days to be covered by stock including the uncertainties involved.
Week 1 Week 2 Week 3 Week 4
Assembly & test
manufacturer
Component supplier 1
Component supplier 2
Assembly & test
PP
PP
Production &
transport
Production & transport
PP = production plan
Difference in production lead
time to cover with stock
Figure 37 Planning of replenishment orders
Secondly, the ROP calculation affects the two other planning tools directly with a weighted average
demand for the coming 2 till 3 months and replenishment levels besides the overall used CV values for
covering the Supply chain uncertainties.
4. What are the relevant actors, what role do they currently have and what role should they have
according the new model?
In the current situation the production planner of the component supplier is managing the stock levels
of the LED assembly and test manufacturer. Due to the implementations of the Supply chain control
models the planning method is changed into a format in which the stock owner (production planner of
the manufacturer) is controlling the Supply chain from the decoupling point of the component supplier
till the goods are on stock at the LED assembly & test manufacturer.
Roles and responsibilities are changed in this new control setup (see also appendix G) from a known
uncontrolled situation to a known controlled situation. In this new setup the production planner of the
component supplier and the LED assembly & test manufacturer are the first in line for managing the day
to day operations. Escalation in case an issue is not solvable within the boundaries which are set in the
ROP is in line with the organization hierarchy.
Master thesis project – Roy Hartevelt Page 71
Figure 38 stakeholder diagram
Figure 38 shows the overview of the involved stakeholders. All critical marked stakeholders have any
relationship with the new control setup according the RACI overview (as presented in appendix G).
5. Which key performance indicators are relevant to evaluate the supply chain plan model and
what are the critical success factors?
Based on research and the outcome of discussions with field experts the following metrics are
introduced:
1. CT measurement per sub process including the variation and target settings
2. Stock levels (at the component supplier (decoupling point) and at the LED assembly and test
manufacturer) are judged against the safety stock levels.
3. Delivery performance component supplier towards the LED assembly & test manufacturer at
component and quantity level.
Above metrics are new in the supply chain of Lumileds and are monitored at a weekly basis. Actions are
taken out of the measurement and filtered on incidents and structural (repeating) failures. A key
element in those measurements is besides a reliable data source and ownership continuous
improvement cycles in which the process improvements regarding cycle time and variance reduction are
part off.
6. What actions are recommended to convince the actors about the benefits of the new supply
chain planning model and way of working?
Benefits for all related actors are the best arguments to convince stakeholders about the improvement
possibilities in there Supply chain. Nevertheless, besides selling all the goods out of the advice models
different workshops were organized to inform all stakeholders about the supposed changes before we
CriticalNon-Critical
Dedicated
Non-Dedicated
Production planner
Maarheeze
Production
manager
Maarheeze
Engineering
manager
Maarheeze
Development
manager
Maarheeze
Product manager
MaarheezeSupply manager
PenangProduction planner
Penang
Component planner
Penang
Production
manager Penang
Supply chain
excellence
manager
Planning manager
Penang
Master thesis project – Roy Hartevelt Page 72
went live with the new planning concept. The results out of the pilot were useful examples to explain
the new ideas and to show the actors the effects based on recognized system information. After
introducing of the new concept an after care (managed by experienced consultants) period of eight
weeks was arranged upfront to safeguard the organization for any fall back due to lack of knowledge
regarding the tools and /or calculations rules.
With help of the sub-questions it’s possible to formulate an answer on the main question:
WHICH SUPPLY CHAIN PLANNING CONTROL IS NEEDED FOR AN MOST SUITABLE STOCK SITUATION TO SECURE THE
SAFETY STOCK LEVELS BETWEEN A COMPONENT SUPPLIER AND THE LED ASSEMBLY & TEST MANUFACTURER.
The advice models gives a product specific advice based on the different lead-times and the Supply chain
uncertainties including a specific demand pattern. Based on the current situation the models are
calculating an advice related to the front-end of the component supplier. The back-end of the
component supplier is controlled via a replenishment calculation al managed by a Re-Order-Point
calculation. The ROP and Kanban boundaries are set ones a month, the replenishment orders are
calculated weekly. As a result of above structure the organization is able to measure the different parts
of the Supply chain and the overall Supply chain as well. The measurements about the sub parts of the
Supply chain are a result of the new control model. The models results in transparency and the
opportunity to simulate the impact of variance or lead-time reduction on stock levels. Based on the
simulation results the model support improvement potential and proves that an equal service is possible
with lower stock levels. The influence of the component supplier and the LED assembly & test
manufacturer can be used to improve and focus on cycle time reduction and controlling the
uncertainties in the Supply chain as much as possible. Finally, the models are the first step to further
improvements & cooperation between the component supplier and the manufacturer.
9.3 Recommendation for further research Although this research project is finished; the insights of this research have identified other future work
which can be performed. These recommendations are divided into three recommendations for future
research.
The first and most important recommendation is to establish, improve and extend the relation between
the component supplier and the LED assembly & test manufacturer. As described in the previous
chapter the advice models are one step in the right direction of a fully controlled Supply chain. The next
step is steering together on improving performance indicators and to start up a variance reduction
program. After that Supply chain integration can be one of the next steps to go in further optimization
and drive together integral Supply chain improvement projects. In a couple of years from now the
situation can be ready for the next steps into system integration between the component supplier and
the manufacturer. Working in one system environment will bring a closer cooperation and opens
especially options for examining planning and forecast root causes like demand /supply
underestimation, phase in or phase out of products and elephant orders for special customers.
Master thesis project – Roy Hartevelt Page 73
The second recommendation is aimed at the goods supply at the component supplier. The source part
of the component supplier was not in scope of this research. It would be worthwhile to also investigate
in the order pattern and purchasing activities in the several articles which are used at the component
supplier. Using the Philips buyer power and setting up system control limits will enable the component
supplier to enter the next level Supply chain control at their source side.
The last recommendation is an extension of the models with a sensitivity check on the several CV factors
and lead-times with the effects on the safety stock levels. Stock and service level decisions are related
with cost considerations. Costs are not part of the model yet. The models can support decision taking
based on service and Supply chain uncertainties, implementing a cost aspect improves the quality of the
decision support power of the models.
9.4 Reflection
9.4.1 Reflection on theory and methodology
A benefit of the design set up is that the models including the way of thinking behind can be indicated as
a general approach to solve comparable issues. The general approach based on the model of Herder and
Stikkelmans (Figure 4) methodology was a guideline to organize researches like this. A missing part of
the methodology of Herder and Stikkelmans is the implementation approach. Therefore input of the
field experts and the publication of D.J. Bowersox ao, 1999 – managing introduction risk through
response-based logistics was added to shape the implementation part of the project.
9.4.2 Reflection on results
Creating a tool for optimizing the controls in the Supply chain was the assignment. Finally a theory based
model was designed and implemented in the Lumileds supply chain. After two iterations the final
version of the models was implemented within the timeframe of the aftercare period. In the last four
months the models are running quite smoothly and are embedded in the running organization. The
models have drastically improved business decision taking. The results of this project move the Lumileds
Supply chain control to a higher platform, by managing the short- and medium term supply and demand
balance.
9.4.3 Personal reflection
The understanding and knowledge gained in my personal development during this research was
enormous. Working with knowledgeable people during this research was really fun and brought me
besides this Master Thesis to a new direction of using science articles for solving day to day business
issues.
Master thesis project – Roy Hartevelt Page 74
References Affonso R. , Grabor B., Marcotte F.(2009). Coordination Model for Supply Chain Management, University
of Toulouse.
Baganha M. P. and Cohen M. A. (1998). The Stabilizing Effect of Inventory in Supply Chains, Operations
Research, Vol. 46, No. 3, Supplement: Focus Issue on Manufacturing Logistics (May - Jun., 1998), pp.
S72-S83.
Bowersox, D.J. (1990). The strategic Benefits of Logitics Alliances. Harvard Business
Review, 86 (4):p.36-43.
Bowersox D.J., Stank T.P. and Daugherty P.J. (1999). Managing introduction risk through response-based
logistics.
Chopra, S., Dada, M. and Reinhardt, G. (2004). The effect of lead time uncertainty on
safety stocks. Decision sciences, 35 (1):p.1-24.
Corsten, D. and Gruen, T.W. (2004). Stock-Outs Cause Walkouts. Harvard Business
Review:p.26-28.
Davis T. (1993). Effective supply chain management.
Enserink, B., Koppenjan, J.F.M., Thissen, W.A.H., Kamps, D.P. and Bekebrede, G.
(2008). Analyse van complexe omgevingen, Delft: Delft University Press.
Eppen, G.D. and Martin, R.K. (1988). Determining safety stock in the presence of
stochastic lead time and demand. Management Science, 34 (11):p.1380-1390.
Fisher, M.L., Hammond, J.H., Obermeyer, W.R. and Raman, A. (1994). Making Supply
Meet Demand in An Uncertain World. Harvard Business Review, 72 (3):p.83-93.
Fleisch, E. and Tellkamp, C. (2005). Inventory inaccuracy and supply chain
performance: a simulation study of a retail supply chain. International Journal of Production
Economics, 95 (3):p.273-385.
Graham C. Stevens (2007). Integrating the Supply Chain, Peat Marwick McLintock, London.
Herder, P.M. and Stikkelman, R.M. (2004). Methanol-based industrial cluster design: a
study of design options and the design process. Industrial & Engineering Chemistry Research,
42:p.3879-3885.
Master thesis project – Roy Hartevelt Page 75
Hengst M. den and Herder P.M. (2004). Quick and Dirty’ Modeling in a Decision Support Tool for Supply
Chain Design, international Confcrence on Systems, Man and Cybernetics 445-450.
Jammernegg, W. and Reiner, G. (2007). Performance improvement of supply chain
processes by coordinated inventory and capacity management. International Journal
Production Economics, 108:p.183-190.
Johson J. D. (2005). Managing variability in the semiconductor supply chain.
Kok T.G. de and Fransoo J.C. (2002). Planning supply chain operations: definition and comparison of
planning concepts.
Kok T. de, Janssen F., J. van Dorenmalen, E. van Wachem, M. Clerkx and W. Peeters (2005). Philips
Electronics Synchronizes Its Supply Chain to End the Bullwhip Effect.
Lee H.L. and Billington C. (1992). Managing supply chain inventory: pitfalls and opportunities.
Lutza S., Loeddingb H., Wiendahl H-P. (2003), Logistics-oriented inventory analysis, Int. J. Production
Economics 85 (2003) 217–231.
Marmostein, H. and Zinn, W. (1990). Comparing two alternative methods of
determining safety stock levels: the demand and the forecast systems. Journal of Business
Logistics, 11 (1):p.95-110.
H. Mintzberg (1983). Structure in Fives: designing effective organizations: p. 121-156.
Orcun S., Çetinkaya S. and Uzsoy R. (2007). Determining safety stocks in the presence of workload-
dependent lead-times.
Ortmanns J., Mooren J., Rulkens H., Mingels J. (2010). Hikari industrialization assesment.
Shannon R. E. (1998). Introduction to the art and science of simulation, Industrial Engineering, Texas
A&M University, College Station, Texas 77843-3131.
Schneider, M. (2009). Out-of-Stock out of business, Enschede, TU Twente.
Sanajian N. and Balcıoglu B. (2009). The impact of production time variability on make-to-stock queue
performance, European Journal of Operational Research 194 (2009) 847–855.
SUWANRUJI P. and ENNS S.T. (2005). Evaluating the effects of capacity constraints and demand patterns
on supply chain replenishment strategies, International Journal of Production Research, Vol. 44, No. 21,
1 November 2006, 4607–4629.
Master thesis project – Roy Hartevelt Page 76
Tien-Yu Lin (2010). An economic order quantity with imperfect quality and quantity discounts. Applied
Mathematical Modelling 34 (2010) 3158–3165.
Zhao, X., Xie, J. and Leung, J. (2002). The impact of information sharing and orderingco-ordination on
supply chain performance. Supply Chain Management: an International
Journal, 7 (1):p.24-40.
Zinn W. and Marmorstein H. (1990). Comparing two alternative methods of determining safety stocks
levels: the demand and the forecast systems.
Master thesis project – Roy Hartevelt Page 77
Appendices
Supply chain planning at Philips Lighting Lumileds
How secure do we like to be?
A design and implementation of a stock control model to balance customer service and stock levels in an
end 2 end environment to improve product availability.
Author:
R. Hartevelt
Confidential
Master thesis project – Roy Hartevelt Page 78
Appendix A Product families
Product family SKU component
supplier
SKU assembly & test
manufacturer
Luxeon Flash PWF5 3222 023 62010 LL60.0163 Bin 330
3222 023 63010 LL60.0163 Bin 440
Luxeon Flash PWF6 3222 023 63310 LL60.0134 Bin 222
3222 023 63410 LL60.0134 Bin 333
3222 023 63510 LL60.0134 Bin 555
3222 023 63610 LL60.0134 Bin 444
Luxeon Rebel Warm White 3222 023 61200 LL60.0187 Bin 231
3222 023 61200 LL60.0187 Bin 232
3222 023 61200 LL60.0187 Bin 241
3222 023 61200 LL60.0187 Bin 242
3222 023 61200 LL60.0187 Bin 251
3222 023 61200 LL60.0187 Bin 252
3222 023 61200 LL60.0187 Bin 261
3222 023 61200 LL60.0187 Bin 262
Luxeon Altilon C2 3222 023 60100 LL60-156-018 Bin 1820
3222 023 60400 LL60-156-018 Bin 1813
3222 023 60300 LL60-156-018 Bin 1806
Luxeon Rebel PC Amber 3222 023 64200 LL60.125 Bin 3 - 5
Luxeon Rebel Hikari 2700K 90CRI T&R 3222 023 6500 LL60.0171 Bin 3 – 5
Luxeon Rebel Hikari 3000K 80CRI T&R 3222 023 65100 LL60.0174 Bin 3 – 5
Luxeon Rebel Hikari 4000K 70CRI T&R 3222 023 65200 LL60.0177 Bin 3 - 5
Master thesis project – Roy Hartevelt Page 79
Appendix B Relation between stock level and service level
Correlations
Stock level Customer line item performance
Stock level Pearson Correlation
N
1,000
26
0,072
26
Customer line item
performance
Pearson Correlation
N
0,072
26
1,000
26
Table 4 Results correlation analysis stock level and CLIP
The correlation between stock level and CLIP can help to control if a higher stock level results in a better
CLIP. There is a significant and relation between stock level and CLIP. The relation is 0.072, which means
a high correlation coefficient.
Master thesis project – Roy Hartevelt Page 80
Appendix C Philips analysis Royal Philips Electronics of the Netherlands is a diversified Health and Well-being company, focused on
improving people’s lives through timely innovations. As a world leader in healthcare, lifestyle and
lighting, Philips integrates technologies and design into people-centric solutions, based on fundamental
customer insights and the brand promise of “sense and simplicity”. Headquartered in the Netherlands,
Philips employs approximately 116,000 employees in more than 60 countries worldwide. With sales of
EUR 23 billion in 2009, the company is a market leader in cardiac care, acute care and home healthcare,
energy efficient lighting solutions and new lighting applications, as well as lifestyle products for personal
well-being and pleasure with strong leadership positions in flat TV, male shaving and grooming, portable
entertainment and oral healthcare.
Sustainability is at the center of Philips’ strategy. Philips is committed to reducing its environmental
footprint in all aspects of its business: in the products, manufacturing, procurement, as well as in the
communities where the company acts and in the working practices of its employees. All Philips products
go through an EcoDesign process, identifying environmental impact in terms of energy efficiency,
hazardous substances, take-back and recycling, weight and lifetime reliability. Philips’ processes on
Green Product sales are verified annually by an independent third party and published in the Annual
Report. Philips aims to combat global healthcare challenges by focusing on delivering better quality
healthcare at lower costs, also in the emerging markets, such as China and India. Philips also takes a
leading position in educational programs, showing its stakeholders that energy efficient solutions are
simple, easy and actionable and make economic sense for national and local governments, businesses,
schools and individuals.
Philips has 3 sectors (Healthcare, Lighting and Consumer Lifestyle) a design- and a research unit. In 2009
Philips invested EUR 1.6 billion in research and development.
Philips Lighting
Philips Lighting is a leading provider of solutions and applications for both professional and consumer
markets. We address lighting needs in a full range of environments – indoors (homes, shops, offices,
schools, hotels, factories, and hospitals) as well as outdoors (public places, residential areas and sports
arenas). We also meet people’s needs on the road, by providing safe lighting in traffic (car lighting and
street lighting). In addition, we deliver light-inspired experiences through architectural and city
beautification projects. Our lighting is also used for specific applications, including horticulture,
refrigeration lighting and signage, as well as heating, air and water purification, and healthcare. With the
new lighting technologies, such as LED technology, and the increasing demand for energy efficient
solutions, Philips will continue shaping the future with groundbreaking new lighting applications.
Philips Lighting Lumileds
Philips Lumileds Lighting Company (founded 1999) is the world's leading manufacturer of high-power
LEDs and a pioneer in the use of solid-state lighting solutions for everyday purposes including
automotive lighting, computer displays, LCD televisions, signage and signaling and general lighting. The
company's patented LUXEON® Power Light Sources are the first to combine the brightness of
Master thesis project – Roy Hartevelt Page 81
conventional lighting with the small footprint, long life and other advantages of LEDs. The company also
supplies core LED material and LED packaging, manufacturing billions of LEDs annually, and ranks as the
producer of the world's brightest red, amber, blue, green and white LEDs.
Lumileds began as the optoelectronics division in Hewlett-Packard (HP) almost 40 years ago. Hewlett-
Packards' experts literally wrote the book on LEDs. In the late 1990's, recognizing the potential for solid-
state lighting, HP and Philips, one of the world's leading lighting companies, began exploring how they
could work together and deliver a new solid-state lighting solution to the market. In 1999 HP split its
company into two, and the optoelectronics group was assigned to the new Agilent Technologies. In
November of the same year, recognizing the enormous potential for LEDs, Agilent Technologies and
Philips formed Lumileds and assigned it the responsibility of developing and marketing the world's
brightest LEDs and enabling a new world of light. In 2005, Philips acquired Agilent Technologies’ interest
in Lumileds. Today, the company continues to lead the industry in the development and release of
increasingly brighter and technically advanced solid-state lighting technology.
Master thesis project – Roy Hartevelt Page 82
Appendix D Details sub processes front-end component supplier
Monthly update based on
customer demand Penang
KANBAN
MES replenishment report
SlurryFIFO
GranulationFIFO
UAP BBO SinterFIFO FIFO
Pre-GrindingFIFO FIFO
WIPOverview_replenishmentorders
Median = 19:12
Min = 8:48
Max = 62:42
Buffer-time = 0
Yield = 100%
OXOX
Median = 17:18
Min = 2:36
Max = 63:30
Buffer-time = 17:42
Yield = 88.13%
Median = 1:18
Min = 0:30
Max = 2:12
Buffer-time = 166:00
Yield = 99.83%
Median = 25:46
Min = 20:46
Max = 45:36
Buffer-time = 36:54
Yield = 99.95%
Median = 55:24
Min = 37:36
Max = 83:06
Buffer-time = 0:12
Yield = 99.95%
Median = 4:00
Min = 2:30
Max = 12:48
Buffer-time = 23:30
Yield = 99.18%
PWF6 & HIKARI
Process Time: 122:40 hours
VSM Lead Time: 244:54 hours
19:12 hr 17:18 hr 1:18 hr 25:46 hr 55:24 hr 4:00 hr
Master thesis project – Roy Hartevelt Page 83
Appendix E Details sub processes front-end component supplier
Weekly update based on planned
consumption & stock take Penang,
in-transit and Re-Order-Point
calculation
Replenishment
GrintapingFIFO
GrindingFIFO
Measuring DCF Coat & CureFIFO FIFO
SepTap
Median = 2:06
Min = 0:42
Max = 4:12
Buffer-time = 0
Yield = 99.82%
Median = 15:42
Min = 2:48
Max = 24:48
Buffer-time =13
Yield = 93.72%
Median = 0:12
Min = 0:06
Max = 5:54
Buffer-time = 0
Yield = 100%
Median = 120
Min =
Max =
Buffer-time =
Yield = 90%
Median = 33:24
Min = 4:54
Max = 110:54
Buffer-time =
Yield =
Median = 3:42
Min = 1:36
Max = 11:00
Buffer-time = 3:42
Yield = 81.56%
HIKARI
Process Time: 231:36 hours
VSM Lead Time: 545:36 hours
Seperation Visual Insp. CTISFIFO FIFO FIFO FIFO
Median = 10:12
Min = 2:12
Max = 16:36
Buffer-time = 4:18
Yield = 100%
Median = 10:24
Min = 3:48
Max = 34:48
Buffer-time = 41:00
Yield = 100%
Median = 35:54
Min = 25:24
Max = 50:00
Buffer-time = 6:12
Yield = 100%
OXOX
2:06 hr 15:42 hr 0:12 hr 120 hr 33:24 hr 3:42 hr 35:54 hr 10:24 hr 10:12 hr
GrintapingFIFO
GrindingFIFO
Measuring SepTap
Median = 1:42
Min = 0:12
Max = 4:12
Buffer-time = 0
Yield = 99.73%
Median = 13:06
Min = 1:42
Max = 30:18
Buffer-time = 10:00
Yield = 80.72%
Median = 0:06
Min = 0:06
Max = 24:48
Buffer-time = 0
Yield = 100%
Median = 1:12
Min = 0:30
Max = 4:24
Buffer-time = 26:36
Yield = 77.84%
PWF6
Process Time: 78:48 hours
VSM Lead Time: 130:24 hours
Seperation Visual Insp. CTISFIFO FIFO FIFO FIFO
Median = 6:30
Min = 0:24
Max = 30:18
Buffer-time = 2:30
Yield = 100%
Median = 3:48
Min = 0:30
Max = 33:30
Buffer-time = 7:30
Yield = 100%
Median = 32:06
Min = 7:12
Max = 145:24
Buffer-time = 5:00
Yield = 100%
Weekly update based on planned
consumption & stock take Penang,
in-transit and Re-Order-Point
calculation
Replenishment
OXOX
1:42 hr 33:24 hr 0:06 hr 1:12 hr 32:06 hr 3:48 hr 6:30 hr
Master thesis project – Roy Hartevelt Page 84
Appendix F Model results Kanban front-end component supplier
Product family Product Demand CV
Lead-time
front-end Batch size
Lead-
time
demand
Safety
stock
95%
Safety
stock
95%
(days)
Total
Kanban
95%
Total
Kanban
95%
(days)
Safety
stock
97.5%
Safety
stock
97.5%
(days)
Total
Kanban
97.5%
Total
Kanban
97.5%
(days)
Safety
stock
99%
Safety
stock
99%
(days)
Total
Kanban
99%
Total
Kanban
99%
(days)
Luxeon Flash5 3222 023 62010 228,538 0.34 2.94 110,000 7 2 9 2 9 2 9
Luxeon Flash5 3222 023 63010 533,256 0.34 2.53 110,000 13 3 16 4 17 4 17
Luxeon Flash6 3222 023 63310 116,751 0.71 2.34 117,000 3 2 5 2 5 2 5
Luxeon Flash6 3222 023 63410 817,259 0.71 2.10 117,000 15 9 24 10 25 12 27
Luxeon Flash6 3222 023 63610 1,050,762 0.71 2.14 117,000 20 11 31 13 33 15 35
Luxeon Flash6 3222 023 63510 350,254 0.71 2.17 117,000 7 4 11 5 12 5 12
Luxeon Rebel WW 3222 023 61200 98,129 1.36 3.43 90,000 4 3 7 3 7 4 8
Luxeon Rebel Hikari 2700K 80CRI T&R 3222 023 65000 91,847 1.00 3.47 60,000 6 3 9 4 10 4 10
Luxeon Rebel Hikari 3000K 80CRI T&R 3222 023 65100 147,404 1.00 3.56 60,000 9 5 14 5 14 6 15
Luxeon Altilon C2 3222 023 60100 11,886 0.79 2.93 40,000 1 1 2 1 2 1 2
Luxeon Altilon C2 3222 023 60400 62,401 0.79 2.61 40,000 5 3 8 3 8 3 8
Luxeon Altilon C2 3222 023 60300 72,801 0.79 2.36 40,000 5 3 8 3 8 4 9
Luxeon Rebel PC Amber 3222 023 64200 159,863 0.84 6.00 95,000 11 3 14 3 14 4 15
3,741,151 106 52 5.2 158 15.8 58 5.8 164 16.4 66 6.6 172 17.2
Master thesis project – Roy Hartevelt Page 85
Appendix G RACI model Supply chain control model
RACI ROP process lumiramics
Activity /Decision in process
SC im
prove
men
t man
ager P
enang
Mate
rial m
anage
r Penan
g
Inte
gral p
lanner M
aarheeze
Plate
let p
lanner
Penan
g
Operatio
ns man
ager l
umira
mics
Mhz
Sr. e
ngineer
ing p
lannin
g offi
cer P
enang
WW
supply
pla
nning
man
ager
Monthly process1 Orientation sheet
Update S&OP R/A C I I
Update bin distribution C A I R
2.1 Copy JDE Sales order detail report into Monthly ROP A R
2.2 Verify Shipped table on completeness A R
2.3 Shipped Pivot
Update product family A R
Determine selection period A R
3 Determine average demand period A R
4 ROP calculation
Determine service level C R/A I I
Determine average lead time back-end Mhz I R/A I
Determine average lead time variation back-end Mhz I R/A I
5 Update Product Life Cycle overview A I I R
Weekly process0 Prepare KPI reports
Penang part A R
Maarheeze part R A
1a Copy JDE reports Stock Aging report into weekly ROP A R
1b Verify Platelet stock table on completeness A R
1 Copy Stock take values into table A R
2 Determine WIP back-end Mhz R A
3 In transit
Update shipped goods R A
Copy JDE report received goods A R
4 Copy ROP from monthly to weekly ROP sheet A R
5 MD update A R
6 Copy BP Penang into weekly ROP A R
7 Copy # of GGI starts into weekly ROP A R
8 Verify GGI history sheet on completeness A R
9a Verify Calc replenishment on completeness /rightness A R
9b Verify Production Orders MHZ on completeness /rightness A R
General activities1 Maintenance of weekly & monthly ROP calculation sheets A C C C I R
2 S&OP forecast Png -> Mhz R/A I I I I
3 Capacity commitment process Mhz -> Png I I R I A
4 Escalation process capacity commitment A C R R A
Allocation decisions to deal with short term constraints R C C C C A
Plan to increase capacity to fulfil capacity need on mid term
Inventory build up decission to avoid capacity constraints
Phase in /phase out products in weekly /monthly process C A C R I I
5 yearly process review /maintenance platelet planning A R C C I C
Legend:
Legend:
- Improvement & Process Responsibility - Process/Improvement Consult
- Process Accountability - Process/Improvement Inform
R
A
C
I
Master thesis project – Roy Hartevelt Page 86
Appendix H IDEF0 level 2
TITLE:NODE: NO.: 2level A1 SADT Lumileds /lumiramics
S&OP
Finished product
shipments
Service level
MAT
Lead-time platelets
Product-life-
cycle
A1.1
Transform
orientation from
monthly to
weekly
A1.2
Shipped data
consolidated per
prod family
A1.3
Calculate
average platelet
demand
A1.4
Calculate
demand variation
A1.5
ROP calculationRe-order-point
Selection period
Platelet distribution
Weekly bin
forecast
Platelet
demand
Weekly qty
per prod
family
Demand variation per prod
family
Prod family table
TITLE:NODE: NO.: 3level A2 SADT Lumileds /lumiramics
WIP MaarheezeIn-transit
Consolidated stock overview
GGI start
ROP
A2.1
Replenishment
calc.
Confirmed plan
Yield back-end Maarheeze
A2.2
Production
planning MHZ
Cap
constraints
Production request
Master thesis project – Roy Hartevelt Page 87
TITLE:NODE: NO.: 4level A3 SADT Lumileds /lumiramics
Stock pre grinded wafers
Confirmed plan
A3.1
Planned orders
MES
Goods receipt production order
Supermarket front-end
Goods Issue back-end Mhz
Production
planning
A3.2
Back-end
production Mhz
Confirmed plan
TITLE:NODE: NO.: 5level A4 SADT Lumileds /lumiramics
Carrier lead-time
Platelets
A4.1
Pack goods
A4.2
Transport
A4.3
Prepare platelet
stock report PngStock data overview
A0
Generate In-
transit report
Platelets on transport
Platelets received
Platelets In-transit
Goods
Ready
for transport
Master thesis project – Roy Hartevelt Page 88
TITLE:NODE: NO.: 6level A5 SADT Lumileds /lumiramics
Platelets In-transit
JD Edwards stock data
ROP
Aggregated stock data overview
A5.1
generate platelet
stock report
Master thesis project – Roy Hartevelt Page 89
Appendix I Monthly Re-Order-Point process
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Monthly process
1. Review product life-
cycle
2. Prepare orientation
sheet
3. Download shipped data
from JDE
4. Determine selection period CV
5. Determine moving average period
6. Update average LT
& LT var and determine
service level
7. Copy new ROP values
in weekly ROP sheet
8
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 1a. Review product life-cycle
• Output of S&OP meeting (tab 5.)
9
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 1b. Product family determination
• At tab 5
– All used item numbers are listed including the product family name
– Like:
– Based on the translation table the product families are grouped in
the pivot
10
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Item Number Product Family
LXCL-PWF4N Flash PWF5
LXCL-PWF4W Flash PWF5
LXCL-PWF4W-3000 Flash PWF5
LXCL-PWF5 Flash PWF5
LXCL-PWF5D Flash PWF5
Sum of Quantity Shipped Column Labels
Row Labels 1008 1009
Altilon 4,266 10,038
Flash PWF5 250,000 310,760
Flash PWF6 2,800 1,010,200
Rebel PC Amber 310 135,090
Rebel WW Lumiramic 11,500 58,000
Platelet, Hikari 2700K 80CRI
Platelet, Hikari 3000K 80CRI
Grand Total 268,876 1,524,088
Master thesis project – Roy Hartevelt Page 90
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 2. Prepare orientation sheet
• Lumiramics forecast Maarheeze
11
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 3. Download shipped data from JDE
• Raw data download from JDE
• Transfer into pivot ready
• Shipped quantity per product family
12
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Master thesis project – Roy Hartevelt Page 91
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 4. Determine selection period CV
• Product family
– Select row from pivot
– Determine selection period
– Evaluate effect of elephant orders
13
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Sum of Quantity Shipped Column Labels
Row Labels 1008 1009 1010
Altilon 4,266 10,038 3,922
Flash PWF5 250,000 310,760 400,020
Flash PWF6 2,800 1,010,200 251,450
Rebel PC Amber 310 135,090 44,275
Rebel WW Lumiramic 11,500 58,000 30,000
Platelet, Hikari 2700K 80CRI
Platelet, Hikari 3000K 80CRI
Grand Total 268,876 1,524,088 729,667
Product Family Row Avg Stdev CV Kolom Week Kolom Week
Rebel WW Lumiramic 33 $I$33:$N$33 111,098 151,008 1.36 9 1015 14 1020
Flash PWF5 30 $B$30:$N$30495,145 169,351 0.34 2 1008 14 1020
Flash PWF6 31 $B$31:$N$31678,072 479,877 0.71 2 1008 14 1020
Rebel Hikari 2700K 80CRI T&R 34 $B$34:$N$34 2,000 #DIV/0! 1.00 2 1008 14 1020
Rebel Hikari 3000K 80CRI T&R 35 $B$35:$N$35 2,000 #DIV/0! 1.00 2 1008 14 1020
Altilon C2 29 $B$29:$N$29 7,918 6,218 0.79 2 1008 14 1020
Rebel PC Amber 32 $B$32:$N$32 94,471 79,168 0.84 2 1008 14 1020
Start week End Week
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 5. Determine moving average period
• By changing the moving
average in a higher value
demand development of the
coming three months are
taken into account.
• Default value is 2 (outlook of
2 months) - (production lead
time and safety stock levels
are at the right level to
anticipate within a period of 2
months).
14
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Master thesis project – Roy Hartevelt Page 92
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 6. Update average LT & LT variance and
determine service level
• If applicable change the production lead time of the back-end including
transport of Maarheeze including the variance.
• Default value of the service level is 95% -
15
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
15
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 7. Copy new ROP values in weekly
ROP sheet
• Copy ROP values into the
weekly ROP calculation
sheet.
16
1. Review product
life-cycle2.
Prepare orientation sheet
3. Downloa
d shipped
data from JDE
4. Determi
ne selection period
CV
5. Determi
ne moving
average period
6. Update
average LT & LT var and
determine
service level
7. Copy new ROP
values in
weekly ROP sheet
Master thesis project – Roy Hartevelt Page 93
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
TO
R m
on
thly
RO
P m
ee
tin
g
18
Terms Of Reference SCM monthly review meeting – pilot PWF5
Meeting Frequency Time
Duration Location
Every 3nd Tuesday of the (Ph) month 4 PM (local time Penang) 60 minutes Telephone conference (participant code *****)
Participants SC improvement manager Penang (chair) Planning manager Penang Sr. planning officer (rebel Lumiramic)
Penang Sr. engineering planning officer Penang Integral planner Maarheeze
Objectives Review target settings KPIs incl yield & lead-
times Review target setting ROP Review the balance between the platelet – end
the pump bin distribution Review safety stock targets Review stock levels supermarkets front & back
end Product lifecycle review (product classification)
Input Stock levels front/back -end supermarket Yield Mhz, Png, Sgp Lead-times Mhz, Png, Sgp S&OP sheet KPIs performance overviews Proposal new ROP levels (Sr. engineering planning officer Penang) Future capacity constraints (Lumiramics, or GGI) Output New ROP levels Updated product life cycle table (NPI, Ramp Up, Mature, EOL)
mapping to scheduling method (ROP, on order)
Leading questions KPIs on target? Stock levels supermarket front -end on target level (2 wks of prod
fc(?))? Supply & demand in balance; safety stock -level performance? Platelet distribution in balance with the pump bin availability
/distribution? Sales forecast reliability above 80%? Is everybody agreeing on the new ROP levels?
Agenda KPI development & trend last
period Product lifecycle classification Review platelet distribution Stock development (bin level)
back /front end supermarket incl safety levels
Proposal new ROP levels Any other relevant business
KPIs Confirmed volume performance Maarheeze CLIP Maarheeze, work order (product & bin)
level CRSD, customer requested shipping date
Lumiramics based products Inventory days platelets (current inventory
platelets Penang : daily run rate lumiramic based fc)
Inactive platelets as % of total platelet stock
Basic values Meeting will start and finish on time Stick to the agenda One meeting The appointed employee is responsible for
his/her actions Open and fair discussion Trust each other No reaction = agreeing Be eager to achieve the required result.
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
RACI – monthly process
19
RACI ROP process lumiramics
Activity /Decision in process
SC im
prove
men
t man
ager P
enang
Mate
rial m
anage
r Penan
g
Inte
gral p
lanner M
aarheeze
Plate
let p
lanner
Penan
g
Operatio
ns man
ager l
umira
mics
Mhz
Senio
r Busin
ess A
nalyst
Penan
g
WW
supply
pla
nning
man
ager
WW
supply
pla
nner
Purchas
ing m
anag
er Pen
ang
Purchas
ing o
ffice
r Penan
g
KHOR, LEA
N KIM
Monthly process1 Orientation sheet
Update S&OP C I I A R
Update bin distribution A I R I C
2.1 Copy JDE Sales order detail report into Monthly ROP A R
2.2 Verify Shipped table on completeness A R
2.3 Shipped Pivot
Update product family A R
Determine selection period A R
3 Determine average demand period A R
4 ROP calculation
Determine service level A I R
Determine average lead time back-end Mhz I R I A
Determine average lead time variation back-end Mhz I R I A
5 Update Product Life Cycle overview I I R A
Legend:As discussed during WS3 - Penang W20
Legend:
- Improvement & Process Responsibility - Process/Improvement Consult
- Process Accountability - Process/Improvement Inform
R
A
C
I
Master thesis project – Roy Hartevelt Page 94
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Run book monthly ROP meeting
• Meeting will take place at the 3rd Tuesday of the Philips month
• Preparation of the meeting and a pre alignment of the outcome of the
calculation together with Maarheeze is planned at the day before
20
Master thesis project – Roy Hartevelt Page 95
Appendix J Weekly replenishment process
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Weekly process
21
1. Import platelet raw data from
JDE in excel2. Update pivot and plot the
stock levels in the tabel
3. Determine
WIP backend
Maarheeze
4. Update in transit sheet
5. Check masterdata
6. Import build plan &
review distribution
7. Load GGI starts
8. Align calculation results with
Mhz
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 1. Import platelet raw data from JDE
in excel
• Copy download of JDE into the weekly excel
• Don’t change anything at the format
• The next step will be arrange via formulas – to become pivot ready
22
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Action
1
Action
2
Master thesis project – Roy Hartevelt Page 96
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 2. Update pivot and plot the stock
levels in the table
• Update pivot
• The values are copied automatically (by macro)
23
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
• Be aware that if there is
no bin identification the
stock pieces are not taken
into account.
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 3. Determine WIP backend Maarheeze
• Maarheeze back-end figures - provided by Adrian at (target) bin level
24
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Master thesis project – Roy Hartevelt Page 97
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 4. Update in transit sheet
• Based on the information we
have in the in transit sheet
(copy at our share point with
a weekly update).
• Input managed by Adrian
25
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 5. Check master data
• All MD as used in the
calculation is collected at
one tab (5 MD).
• Based on those values
calculations will run.
26
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 6. Import S&OP & review distribution
27
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
• Copy monthly S&OP and translate into weekly demand, bin distribution
included.
S&OP
Product Family 12nc - Maarheeze Lumi - Penang Wknr 1027 1028 1029 1030 1031
Flash PWF5 3222 023 62010 LL60.0163 B330 264,613 264,613 264,613 264,613 264,613
Flash PWF5 3222 023 63010 LL60.0163 B440 617,431 617,431 617,431 617,431 617,431
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B231 9,540 9,540 9,540 9,540 9,540
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B232 18,535 18,535 18,535 18,535 18,535
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B241 28,601 28,601 28,601 28,601 28,601
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B242 26,151 26,151 26,151 26,151 26,151
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B251 10,859 10,859 10,859 10,859 10,859
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B252 3,312 3,312 3,312 3,312 3,312
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B261 2,003 2,003 2,003 2,003 2,003
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B262 159 159 159 159 159
Flash PWF6 3222 023 63310 LL60.0134 B222 255,592 255,592 255,592 255,592 255,592
Flash PWF6 3222 023 63410 LL60.0134 B333 766,777 766,777 766,777 766,777 766,777
Flash PWF6 3222 023 63610 LL60.0134 B444 1,150,166 1,150,166 1,150,166 1,150,166 1,150,166
Flash PWF6 3222 023 63510 LL60.0134 B555 383,389 383,389 383,389 383,389 383,389
Jul-10
Master thesis project – Roy Hartevelt Page 98
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 7. – GGI starts
• Copy GGI starts, planned GGI starts for next week are the expected
consumption 2 weeks later.
28
GGI Starts (history)
Product Family 12nc - Maarheeze Lumi - PenangWknr 1023 1024 1025 1026
Flash PWF5 3222 023 62010 LL60.0163 B330 209600 209600 209600 209,600
Flash PWF5 3222 023 63010 LL60.0163 B440 503040 503040 503,040 503,040
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B231 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B232 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B241 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B242 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B251 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B252 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B261 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B262 0 0 0 0
Flash PWF6 3222 023 63310 LL60.0134 B222 272480 272,480 544,960 230,560
Flash PWF6 3222 023 63410 LL60.0134 B333 1655840 838,400 838,400 691,680
Flash PWF6 3222 023 63610 LL60.0134 B444 544960 1,236,640 964,160 1,048,000
Flash PWF6 3222 023 63510 LL60.0134 B555 272480 419,200 419,200 356,320
Jun-10
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Step 8. Input overview & Calculate replenishment
• Replenishment = ROP – (Pipeline – Usage) rounded to next multiple of 110000
• (G10); 1,672,168 – (200,000+101,519+997,250-115,161) = 488560, rounded: 550000
• If GGI starts are entered in G6, e.g. 10, then G7 will use 10*21840 = 218400, instead of 115,161
• Econ Stock = current pipeline + planned / actual supply – planned / actual usage (actual overwrites
planned)
• (G15); 1,298,769 + 200,000 - 115,161 = 1,383,608 (we now use G11 as actual, not G10 as planned)
• Purpose : to calculate end of period econ stock and project out into the future with MRP build plan and
determine how many weeks we can continue production with the current pipeline stock.
29
If actual is empty
(G8), then G7 is
used
If actual is empty
(G11), then G10
is used. In this
case G11 is
used.
Master thesis project – Roy Hartevelt Page 99
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 9a. Align calculation results with Mhz
• Step 9a shows the original replenishment quantities
30
1. Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Overview Replenishment calculations
Wknr 1027 1028 1029 1030 1031
Flash PWF5 3222 023 62010 LL60.0163 Econ Stock 410,191 575,578 420,965 376,351 441,738
B330 ROP 639,058 639,058 639,058 639,058 639,058
Replenishment 228,867 63,480 218,093 262,707 197,320
Flash PWF5 3222 023 63010 LL60.0163 Econ Stock 852,837 918,975 961,544 894,114 936,683
B440 ROP 1,491,134 1,491,134 1,491,134 1,491,134 1,491,134
Replenishment 638,297 572,159 529,590 597,020 554,451
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 Econ Stock 277,679 258,599 249,058 239,518 229,978
B231 ROP 30,549 30,549 30,549 30,549 30,549
Replenishment 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 Econ Stock 507,344 470,274 451,739 433,204 414,669
B232 ROP 59,352 59,352 59,352 59,352 59,352
Replenishment 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 Econ Stock 328,398 271,196 242,595 213,995 185,394
B241 ROP 91,584 91,584 91,584 91,584 91,584
Replenishment 0 0 0 0 0
Jul -10
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Rounding Replenishment to Production Batches Jul -10
Product Family 12nc - Maarheeze Lumi - Penang Wknr 1027 1028 1029 1030 1031
Flash PWF5 3222 023 62010 LL60.0163 B330 330,000 110,000 220,000 330,000 220,000
Flash PWF5 3222 023 63010 LL60.0163 B440 660,000 660,000 550,000 660,000 660,000
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B231 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B232 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B241 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B242 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B251 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B252 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B261 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B262 0 0 0 0 0
Flash PWF6 3222 023 63310 LL60.0134 B222 0 0 0 236,000 236,000
Flash PWF6 3222 023 63410 LL60.0134 B333 590,000 708,000 826,000 708,000 826,000
Flash PWF6 3222 023 63610 LL60.0134 B444 1,298,000 1,180,000 1,062,000 1,180,000 1,180,000
Flash PWF6 3222 023 63510 LL60.0134 B555 0 236,000 354,000 472,000 354,000
Step 9b. Align calculation results with Mhz
• Step 9b does the rounding to the nearest production batch sizes
31
1. Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Master thesis project – Roy Hartevelt Page 100
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Step 9c. Align calculation results with Mhz
• In Step 9c the requested (and rounded) replenishment quantities can be
confirmed (and modified) by Maarheeze.
32
1. Import platelet
raw data from
JDE in excel
2. Update pivot
and plot the
stock levels in
the tabel 3.
Determine WIP backen
d Maarhe
eze
4. Update
in transit sheet5.
Check masterd
ata
6. Import build
plan & review
distribution
7. Load GGI
starts
8. Align calculati
on results
with Mhz
Confirmed Orders MHZ < 1 wk < 2 wks < 3 wks not used
12nc Lumi Wknr 1027 1028 1029 1030 1031
Flash PWF5 3222 023 62010 LL60.0163 B330 430,000 110,000 220,000 330,000 220,000
Flash PWF5 3222 023 63010 LL60.0163 B440 861,000 660,000 550,000 660,000 660,000
Rebel WW Lumiramic not existing LUMI.0187 B221 0 0 0 0 0
Rebel WW Lumiramic not existing LUMI.0187 B222 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B231 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B232 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B241 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B242 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B251 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B252 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B261 0 0 0 0 0
Rebel WW Lumiramic 3222 023 61200 LUMI.0187 B262 0 0 0 0 0
Flash PWF6 3222 023 63310 LL60.0134 B222 0 0 0 236,000 236,000
Flash PWF6 3222 023 63410 LL60.0134 B333 948,000 708,000 826,000 708,000 826,000
Flash PWF6 3222 023 63610 LL60.0134 B444 1,300,000 1,180,000 1,062,000 1,180,000 1,180,000
Flash PWF6 3222 023 63510 LL60.0134 B555 356,000 236,000 354,000 472,000 354,000
Jul -10
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
TOR weekly ROP meeting
44
Terms Of Reference SCM weekly review meeting
Meeting Frequency Time Duration Location
Weekly Monday - 5 PM (local time Penang) 60 minutes Telephone conference (participant code *****)
Participants SC improvement manager Penang
(chair) Planning manager Penang Sr. planning officer (rebel Lumiramic)
Penang Sr. engineering planning officer
Penang Integral planner Maarheeze
Objectives Discuss result production Penang, Maarheeze & Singapore o f
the last week by reviewing defined KPIs . Based on KPI review determine improvement areas including
corrective actions Set priorities for quality & quantity related topics Discuss progress running actions linked to this meeting
Input Action list SCM review meeting Weekly platelet planning sheet
previous week KPIs performance overviews Output Updated weekly platelet planning
sheet Updated action list SCM review
meeting
Leading questions Winning strikes of last week Attention areas of last week Taken actions and results Actions taken to prevent our organization for the bleeders of today Short term planned actions
Agenda Results last week & results next
week o Maarheeze o Penang
Stock levels back-end Maarheeze Outcome weekly platelet planning
sheet Review main actions Action list update Any other relevant business
KPIs Confirmed volume performance Maarheeze CLIP Maarheeze, work order (product & bin)
level CRSD, customer requested shipping date
Lumiramics based products Inventory days platelets (current inventory
platelets Penang : daily run rate lumiramic based fc)
Inactive platelets as % of total platelet stock
Basic values Meeting will start and finish on time Stick to the agenda One meeting The appointed employee is responsible fo r his/her
actions Open and fair discussion Trust each other No reaction = agreeing Be eager to achieve the required result.
Master thesis project – Roy Hartevelt Page 101
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
RACI - weekly
45
RACI ROP process lumiramics
Activity /Decision in process
SC im
prove
men
t man
ager P
enang
Mate
rial m
anage
r Penan
g
Inte
gral p
lanner M
aarheeze
Plate
let p
lanner
Penan
g
Operatio
ns man
ager l
umira
mics
Mhz
Senio
r Busin
ess A
nalyst
Penan
g
WW
supply
pla
nning
man
ager
WW
supply
pla
nner
Purchas
ing m
anag
er Pen
ang
Purchas
ing o
ffice
r Penan
g
KHOR, LEA
N KIM
Weekly process0 Prepare KPI reports
Penang part A R
Maarheeze part R A
1a Copy JDE reports Stock Aging report into weekly ROP A R
1b Verify Platelet stock table on completeness A R
1 Copy Stock take values into table A R
2 Determine WIP back-end Mhz R A
3 In transit
Update shipped goods R A
Copy JDE report received goods A R
4 Copy ROP from monthly to weekly ROP sheet A R
5 MD update A R
6 Copy GGI schedule Penang into weekly ROP A R
7 Copy # of GGI starts into weekly ROP A R
8 Verify GGI history sheet on completeness A R
9a Verify Calc replenishment on completeness /rightness A R
9b Verify Production Orders MHZ on completeness /rightness A R
Legend:As discussed during WS3 - Penang W20
Legend:
- Improvement & Process Responsibility - Process/Improvement Consult
- Process Accountability - Process/Improvement Inform
R
A
C
I
Confidential Philips Applied Technologies – Industry Consulting, Juli 2010
Run book weekly ROP meeting
• Meeting will take place at every Thursday (5pm Penang local time)
• Preparation and pre alignment planned at Wednesday (Maarheeze)
/Thursday Morning (Penang)
46
Master thesis project – Roy Hartevelt Page 102
Appendix K Tutorial Supply chain control models
PHILIPS LIGHTING, BU LUMILEDS
Tutorial Re-Order-Point
calculation platelets
Penang Planning method for lumiramic based products
Penang and Maarheeze
[July 2010]
This document will help you to understand the re-Order-Point calculation method including the
independencies between the organizations involved.
Master thesis project – Roy Hartevelt Page 103
Purpose of this document
This document will explain how to use and maintain the Re-Order-Point calculation excel sheets.
Re-Order-Point process
Overall context
The figure below shows the current lumiramics business process at high level. The numbers in the figure
(described in the text below) highlights the most critical process steps. Secondly, the black line indicates
the situation before the project the red line indicates the situation at the end of the project. The design
of the ROP calculation will cover the back-end processes of Maarheeze including the GGI starts. The GGI
starts is the main trigger for starting production batches in the back-end of Maarheeze. The front-end of
Maarheeze is managed based on the Kanban principle. The level of Kanban tickets in the front-end of
Maarheeze is directly linked with the ROP calculation.
ROP GGI/Saber/Assy&Test
1. In this phase of the to-be we’ll keep the GGI starts planning the same as in the as-is (MRP build-
plan GGI),
2. The same for the EPI starts (so no changes here).
Build plan GGI
Determine production batches on pump-bin level
(checking pump-bin and platelet bin availabilities, finding combinations wafer types, pump-bins,
Master thesis project – Roy Hartevelt Page 104
platelets, among competing requirements other products)
will be investigated in a broader scope (not just lumiramics) and supported by a LP tool.
Replenish platelet bank
3. ROP on platelet bank in Penang
ROP calculation based on forecast and bin distribution
ROP execution based on GGI production orders accumulated per internal partnr / bin, platelet
stock, WIP in Maarheeze and in-transit
4. Replenish the pre-grinded wafer bank
(using Kanban, as product costs in front-end are relatively low)
5. Anticipate on started GGI production orders from (1)
data flow from GGI starts to platelet bank (as future consumption)
Focus areas
As mentioned in the introduction of this chapter we will limited ourselves to the supply of lumiramics
from Maarheeze to Penang. In the IDEF0 (Integration Definition for Function Modeling) diagram below
the overall stock calculation process is decrypted in a structured matter.
Capacity
constraints
Product-life-
cycle
TITLE:NODE: NO.: 1To be
A0SADT Lumileds /lumiramics
A3
WIP back-end
Maarheeze
A1
Monthly ROP
calculation
A4
Generate In-
transit overview
A2
Weekly
replenishment
S&OP
Shipments
Service levelMAT
Lead-time platelets
WIP Maarheeze
In-transit
Consolidated stock data overview per bin
Yield back-end Maarheeze
Stock pre grinded wafers
Carrier lead-time
Completions to inventory
Consolidated stock data
overview
GGI start
Period
Platelet distribution
GI /Invoice
MES
Supermarket front-end
Maarheeze
Stock data overview
MES
A5
Calculate platelet
stock position
Png
ROP
values
WIP Back-end
Mhz
Confirmed
plan
Requested
plan
In-transit
overview
Prod family table
Master thesis project – Roy Hartevelt Page 105
A1. The process design starts with a Re-Order-Point calculation, to be executed on a monthly frequency.
As input for the Monthly ROP calculation the following elements are necessary;
The platelet distribution,
The platelet orientation out of the S&OP,
Shipments of the last 6 months for determination of the variance,
The transport lead-time and his variation and
A product family table Control elements are;
Selected period of customer shipments (for variance determination)
The MAT (moving average total) which incorporate the demand development for the coming periods
The product lifecycle, the ROP calculation method is designed for all phases in the product lifecycle but has a best fit in the maturity phase of a product.
Service level, with the service level indicator you will infect directly the level of safety stocks. The higher the services level the higher the safety stock.
Output element;
The ROP value per product family.
A2. Replenishment process
Input elements
Work-In-Process (WIP) back-end Maarheeze, all back-end started production batches but not
yet completed.
In-transit, production batches which has left Maarheeze but without a goods receipt in Penang.
Consolidated stock overview per bin, all lumiramics booked on stock in Penang
Yield back-end Maarheeze, on product level
GGI starts, number of blue bin batches planned for startup in the GGI process. Based on this
information we know almost exactly the consumption in two weeks from now. See also picture
below;
Master thesis project – Roy Hartevelt Page 106
Control elements
ROP values as set in the monthly ROP calculation (as described at A1)
Capacity constraints lumiramic based products – max capacity limits at Maarheeze.
Output element
Requested plan
Confirmed plan
A3. WIP back-end Maarheeze
Input elements
GI/Invoice – for measuring output of the back-end of Maarheeze the SAP Invoice or Goods Issue
information is leading. Without any SAP Invoice no delivery towards Penang.
Supermarket front-end Maarheeze, the basic material for the final product of Maarheeze.
Without stock in the front-end supermarket (on product bin level) no start up of production
batches in the back-end is possible.
Control elements
Requested plan, the first parameter is set by the initial request out of the replenishment
calculation. This calculation is mainly based on the boundaries as set in the ROP calculation and
the GGI starts.
Confirmed plan, based on the request Maarheeze will plan the back-end with as input the
supermarket (available raw materials and the available capacity) a production plan is
communicated towards Penang. This (confirmed) plan is the benchmark of the back-end
production of Maarheeze.
2010-05-02 2010-06-06
5-9 5-16 5-23 5-30
Singapore5-3 - 5-5
GGI /Saber
planning
2010-05-02 2010-06-06
5-9 5-16 5-23 5-30
Penang
2010-05-02 2010-06-06
5-9 5-16 5-23 5-30
Maarheeze5-6 - 5-7
B-E
planning
5-10 - 5-21
Back-end production
5-21 - 5-25
Transport
5-10 - 5-23
GGI /Saber production (throughput time; 3 + 7 days)
5-23 - 5-25
Transport
GGI /Saber is leading
Bin distribution based on fc
5-30 - 6-6
Start production
EXAMPLE
Master thesis project – Roy Hartevelt Page 107
A4. Generate In-transit overview
Input elements
WIP (work-in-process) back-end Maarheeze, the batches of the final products of the back-end of
Maarheeze are transferred to the in-transit sheet.
Stock data overview, based on the match at batch level between the goods receipt entries
(physical available stock of Penang) and the output of the back-end Maarheeze the final
overview can be generated.
Control elements
Carrier lead-time is the control element in this process, the longer the lead-time the bigger the
stock quantities in-transit.
A5. Calculate platelet stock position Penang
Input elements
In-transit overview, output of the process as descript at A4.
Completions to inventory, all production batches which are booked in.
Control elements
ROP values, as described in A1 the ROP boundaries are set in the calculation. Physical and
average available stock level in Penang according the model is the safety stock and approx 50%
of the demand during lead-time.
Master thesis project – Roy Hartevelt Page 108
Monthly Re-Order-Point calculation
Model setup
Step 1 – review product life cycle
On final product level, meaning one coefficient of variance per final product is valid for multiple
(lumiramic) bins. In the table below the threshold values are mentioned, based on the threshold value
the calculation can be followed freely without too much manual intervention.
1. Review product life-
cycle
2. Prepare orientation
sheet
3. Download shipped data
from JDE
4. Determine selection period CV
5. Determine moving average period
6. Update average LT &
LT var and determine
service level
7. Copy new ROP values in weekly ROP sheet
Product life cycle Characteristics Threshold value Planning method
New product
introduction
Slow volumes to start
and demand has to be
created
CV > 1 Re-order-point &
manual intervention
Ramp up (growth) Sales volume
increases significantly
CV > 1 Re-order-point &
manual intervention
Maturity Sales volume peaks
and market saturation
is reached
CV < 1 Re-order-point
calculation
Decline (End of
Live)
Sales volume decline
or stabilize
CV > 1 Re-order-point &
manual intervention
Master thesis project – Roy Hartevelt Page 109
Step 2 – prepare orientation sheet
Based on the S&OP a lumiramics supply plan is generated. The yellow fields are changeable (period,
total need and distribution) the rest of the fields are filled with formulas. None yellow fields are write
protected.
Step 3 – download shipped data from JDE
Download the standard report from JD Edwards (R554211C) – no changes in format, deleting columns or
whatever.
At the next modifications are made, empty columns are deleted. Column Q and R are calculated fields
(Wknr based on actual shipped date [column M] & Prod fam based on 2nd item number [column H]), be
aware in case of many shipments (more than 27 k) the formulas must be copied down (A ..R).
Regarding the product family as mentioned in column R – the 2nd item number of column H is linked
with a table at tab.5
Orientation based on monthly S&OP update
Phosphor Product Line Product Family Lumi nmbr Penang 12nc Maarheeze Bin Distribution Jul-10 Aug-10 Sep-10 Oct-10
LUMIRAMIC Luxeon Flash Flash PWF5 4,410,220 2,566,173 2,688,853 3,931,685
LL60.0163 3222 023 62010 330 30% 1,323,066 769,852 806,656 1,179,505
LL60.0163 3222 023 63010 440 70% 3,087,154 1,796,321 1,882,197 2,752,179
Luxeon Flash Flash PWF6 12,779,618 8,456,514 8,036,564 10,555,193
LL60.0134 3222 023 63310 222 5% 638,981 422,826 401,828 527,760
LL60.0134 3222 023 63410 333 35% 4,472,866 2,959,780 2,812,798 3,694,317
LL60.0134 3222 023 63610 444 45% 5,750,828 3,805,431 3,616,454 4,749,837
LL60.0134 3222 023 63510 555 15% 1,916,943 1,268,477 1,205,485 1,583,279
Luxeon Rebel Rebel WW Lumiramic 495,854 388,353 459,362 573,198
LL60.0187 3222 023 61200 221 0% - - - -
LL60.0187 3222 023 61200 222 0% - - - -
LL60.0187 3222 023 61200 231 10% 47,706 37,363 44,195 55,147
LL60.0187 3222 023 61200 232 19% 92,684 72,590 85,863 107,142
LL60.0187 3222 023 61200 241 29% 143,019 112,012 132,493 165,327
LL60.0187 3222 023 61200 242 26% 130,770 102,419 121,146 151,168
LL60.0187 3222 023 61200 251 11% 54,301 42,529 50,305 62,772
LL60.0187 3222 023 61200 252 3% 16,563 12,972 15,344 19,147
LL60.0187 3222 023 61200 261 2% 10,017 7,846 9,280 11,580
LL60.0187 3222 023 61200 262 0% 793 621 735 917
R554211C Philips LumiLeds Lighting
Sales Order Detail Report
Order Number Or Ty Order Co Line Number Hd CD Request Date Sched Pick
363961 SI 10 50 2009-12-15 ########
363961 SI 10 60 2010-01-08 ########
363961 SI 10 70 2010-02-12 ########
363961 SI 10 80 2010-03-12 ########
Master thesis project – Roy Hartevelt Page 110
Step 4 – determine selection period CV
The pivot will refresh automatically after entering the tab. Check after refresh if the total data source is
selected via pivotTable Tools| change data source. In case an unknown family name is popping up as
part of the row labels; the product family table at tab 5 is outdated. Actions to take- update product
family table at tab 5 and refresh the pivot.
2nd step is to determine the CV value (coefficient of variance) is the standard deviation divided by the
average. In column Row [B] you have to select the right row of the product family in the pivot. For
selecting the requested time window use the yellow columns at the right side [G&I].
R554211C
Order
Number Or Ty Order Co
Line
Number
Request
Date
Sched
Pick
Order
Date
363961 SI 10 50 40162 40221 40162
363961 SI 10 60 40186 40221 40162
363961 SI 10 70 40221 40221 40162
363961 SI 10 80 40249 40249 40162
Sum of Quantity Shipped Column Labels
Row Labels 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 Grand Total
Altilon 4,266 10,038 3,922 12,436 30 18,396 4,430 35 7,943 9,544 14,304 1,102 16,493 6,283 109,222
Flash PWF5 250,000 310,760 400,020 885,150 585,000 580,000 455,000 410,000 490,900 375,000 538,050 447,000 710,000 265,000 6,701,880
Flash PWF6 2,800 1,010,200 251,450 650,000 320,100 960,000 1,102,000 333,150 500,150 1,045,000 1,600,020 50,064 990,000 1,312,000 10,126,934
Rebel PC Amber 310 135,090 44,275 59,070 75,000 112,150 109,020 116,304 276,040 100,010 194,100 6,700 55 106,095 1,334,219
Rebel WW Lumiramic 11,500 58,000 30,000 35,000 32,500 9,050 39,040 410,000 32,500 124,050 22,000 39,000 98,000 940,640
Platelet, Hikari 2700K 80CRI 2,000 450 2,450
Platelet, Hikari 3000K 80CRI 2,000 2,000
Grand Total 268,876 1,524,088 729,667 1,641,656 1,012,630 1,670,546 1,679,500 898,529 1,685,033 1,562,054 2,470,524 526,866 1,759,548 1,787,828 19,217,345
Product Family Row Avg Stdev CV Kolom Week Kolom Week
Rebel WW Lumiramic 33 $I$33:$N$33 111,098 151,008 1.36 9 1015 14 1020
Flash PWF5 30 $B$30:$N$30 495,145 169,351 0.34 2 1008 14 1020
Flash PWF6 31 $B$31:$N$31 678,072 479,877 0.71 2 1008 14 1020
Rebel Hikari 2500K 80CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 2500K 90CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 2700K 75CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 2700K 80CRI T&R 34 $B$34:$N$34 2,000 #DIV/0! 1.00 2 1008 14 1020
Rebel Hikari 2700K 90CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 3000K 65CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 3000K 75CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 3000K 80CRI T&R 35 $B$35:$N$35 2,000 #DIV/0! 1.00 2 1008 14 1020
Rebel Hikari 3000K 90CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 3500K 80CRI T&R - NA NA NA 1.00 2 1008 14 1020
RebelHikari 4000K 70CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 4000K 80CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 5000K 80CRI T&R - NA NA NA 1.00 2 1008 14 1020
Rebel Hikari 5700K 65CRI T&R - NA NA NA 1.00 2 1008 14 1020
Altilon C2 29 $B$29:$N$29 7,918 6,218 0.79 2 1008 14 1020
Rebel PC Amber 32 $B$32:$N$32 94,471 79,168 0.84 2 1008 14 1020
Start week End Week
Master thesis project – Roy Hartevelt Page 111
Step 5 – determine moving average period
Use column MAT [Q] for changing the moving average total, the MAT takes care about the future
changes in the S&OP. The default value for the MAT is 2, meaning the forecast of the 2 coming months
are taken into account. In case the value is set at 3 the coming 3 months are taken into account,
etcetera. The MAT is used as weekly demand in the ROP calculation.
Step 6 – parameters to set for calculating the economical stock levels (ROP) are in (the yellow marked
fields) column S, T, U, V, W.
Column S: supply variance back-end Maarheeze measured on planning reliability.
Column T: supply variance back-end maarheeze measured on quantity reliability.
Column U: lead-time back-end Maarheeze in weeks.
Column V: lead-time GGI/Saber Singapore in weeks.
Column W: requested service level as agreed at management level between Maarheeze and
Penang.
Column S, T, U requires input of the integral planner of Maarheeze. Column V is the responsibility of the
platelet planner of Penang. Column W is the responsibility of the demand manager Penang.
The ROP is the sum of the safety stock and the demand during lead-time.
Column A – G are part the standard format repeating at at tab.
Column H, ROP calculation = I + J
Column I, Demand during leadtime = K (weekly demand) * L (LT1 = U + transport lead-time of 1 week).
Column J, Safety stock = P * SQRT(M*(O*K)^2+K^2*(N*M)^2)
M refer to the demand lead-time of GGI/Saber versus LT1 with a minimum of 1 week
(replenishment lead-time)
N refer to the supply variance = SQRT(S^2+T^2)
O refer to the demand variation as calculated at tab 2.3
P refer to the service level as indicated in column W
Phosphor Product Line Product Family Lumi nmbr Penang12nc Maarheeze Bin Distribution MAT Period 1 MAT period 1 Period 2 MAT period 2 Period 3 MAT period 3LUMIRAMIC Luxeon Flash Flash PWF5 2 Jul-10 / Aug-10 761,794 Aug-10 / Sep-10 656,878 Sep-10 / Oct-10 729,275
LL60.0163 3222 023 62010 330 30% 228,538 197,063 218,783
LL60.0163 3222 023 63010 440 70% 533,256 459,815 510,493
Luxeon Flash Flash PWF6 2 Jul-10 / Aug-10 2,335,026 Aug-10 / Sep-10 2,061,635 Sep-10 / Oct-10 2,060,090
LL60.0134 3222 023 63310 222 5% 116,751 103,082 103,004
LL60.0134 3222 023 63410 333 35% 817,259 721,572 721,031
LL60.0134 3222 023 63610 444 45% 1,050,762 927,736 927,040
LL60.0134 3222 023 63510 555 15% 350,254 309,245 309,013
Moving average total
ROP calculation sheet E F G H I J K L M N O P Q R S T U V W
(MAT)
Phosphor Product Line Product Family Lumi nmbr Penang 12nc Maarheeze Bin ROP Dem LT. Safety stock Weekly dem. LT 1 LT 2 Suppl var. Dem var. Z Safety[wks] Avg stock [wks] Supply var. t Supply var.q LT BE LT GGI Service l.
LUMIRAMICLuxeon Flash Flash PWF5
LL60.0163 3222 023 62010 330 766,037 576,569 189,468 228,538 2.52 1.00 0.35 0.36 1.64 0.83 1.33 0.33 0.13 1.52 1.71 95%
LL60.0163 3222 023 63010 440 1,845,187 1,374,276 470,911 533,256 2.58 1.00 0.42 0.36 1.64 0.88 1.38 0.40 0.13 1.58 1.71 95%
Luxeon Flash Flash PWF6
LL60.0134 3222 023 63310 222 482,663 331,740 150,922 116,751 2.84 1.13 0.45 0.61 1.64 1.29 1.79 0.40 0.20 1.84 1.71 95%
LL60.0134 3222 023 63410 333 3,716,094 2,317,513 1,398,581 817,259 2.84 1.12 0.76 0.61 1.64 1.71 2.21 0.73 0.20 1.84 1.71 95%
LL60.0134 3222 023 63610 444 4,592,884 3,068,224 1,524,660 1,050,762 2.92 1.21 0.52 0.61 1.64 1.45 1.95 0.48 0.20 1.92 1.71 95%
LL60.0134 3222 023 63510 555 1,548,109 1,049,761 498,348 350,254 3.00 1.28 0.46 0.61 1.64 1.42 1.92 0.41 0.20 2.00 1.71 95%
Master thesis project – Roy Hartevelt Page 112
Q is a calculation field = J/K representing the safety stock levels in terms of weeks of demand as
presented in column K.
R present the number of weeks physical available stock in Penang = the safety stock + ½ week.
The ½ week representing the average stock level due to consumption in a replenishment period
of 1 week.
Step 7 – copy new ROP levels in weekly ROP (replenishment) sheet
Copy column *H+ ‘ROP’ to the weekly replenishment sheet and save close the monthly_lumiramics.xls.
Maintain procedure
The Monthly_lumiramics.xls is a protected sheet; the password of each tab is drop. In the protected
mode users are only allowed to modify the yellow colored fields.
The standard fields like article number, bin distribution, month and etcetera are all linked to the
corresponding fields at tab 1 [Orientation sheet].
In case of adding or changing information at material master level, start at tab1. In case of a new
commercial type don’t forget to update the product family table at tab5.
In case of adding a new product to the ROP calculation sheet, start at tab1 with insert new row. Repeat
inserting new row (same position) at tab2.2, tab3, tab4 and copy the formulas from the row above.
FAQ
1. Q: Replenishment is arriving too late at Penang. Q: Increase transport lead-time and/or
transport lead-time variation at tab4 [column K and M].
2. Q: Request of weekly replenishment is more than we can deliver. A: Compare demand at
tab1.[orientation] & tab3.[avg.dem] with the requested quantities based on the GGI starts in the
weekly replenishment sheet. If the deviation is more than 10% re-consider the ROP by a manual
update of the CV value at tab 2.3 (the higher the CV value the higher the ROP value). And copy
the revised ROP value to the replenishment sheet. Communicate the issue to the S&OP
/Demand planner.
3. …
Master thesis project – Roy Hartevelt Page 113
Weekly Replenishment calculation
During the weekly replenishment cycle we determine the actual quantities of platelets to be replenished
by Maarheeze. The timing is chosen close to the end of the week (N-1) in order to get the best estimate
of the begin-of-the-week N status with respect to WIP, In-Transit and On-hand Inventory. This gives us
the economical stock at the start of week N. The planned GGI starts for week N give us the demand for
week N. The required replenishment for week N is easily determined by ROP – economical stock +
demand week N (1000 – 800 + 300 = 500; 800 – 1000 + 300 = 100)
The main efforts in the weekly cycle are to get all the inputs and in a synchronized way (in order to avoid
double counting). Typical inputs: Stock take of platelets in Penang, Work in process in Maarheeze, In-
Transit between Maarheeze and Penang, S&OP monthly quantities and the GGI starts for next week.
With this input the replenishments will be calculated automatically. The requested quantities are given
to Maarheeze, and Maarheeze will respond with the confirmed quantities based on their current
capabilities (capacity, front-end supermarket, yield situation, etc.)
Model setup
1. Import platelet raw
data from JDE in excel
2. Update pivot and
plot the stock levels in the
tabel
3. Determine WIP backend Maarheeze
4. Update in transit sheet
5. Check masterdata
6. Import build plan &
review distribution
7. Load GGI starts
8. Align calculation results with
Mhz
Master thesis project – Roy Hartevelt Page 114
These process steps are supported by an excel workbook with the following sheets:
Name Description
Action list
Instructions
1a Platelets raw data Input from JDE
1b Platelet stock Transform 1a
1c Stock Pivot Pivot of 1b
1 Stock take Select 1c into one column
2 WIP Input from MHZ
3 In Transit InTransit
4 ROPs Input Monthly ROPS
5 MD Master Data
6 SnOP Input Constrained Lumiramics S&OP
7 GGI Starts Input Planned GGI starts
7a Backlog Input for GGI backlog
8 Replmts Replenishment Calculation and input summary
9a Calc Replenishment Replenishment details
9b Production Orders MHZ Replenishment rounded
9c Confirmed Orders MHZ Confirmed replenishment
8 StockDays Auxiliary sheet for highlighting 9b & 9c
Summary On-hand projection (still in development)
WIP Reports Input from Sorting (LL60 to Lumi)
Step 1 Import platelet bin raw data from JDE into Excel
The standard JDE report “Stock Aging Report By BIN (excel Format) with Sales Cat Code” is loaded into
sheet “1a Platelets raw data”. This should be the exact standard report without any modifications. The
reason for this: we will transform this report into the required bin information automatically. However,
in order to do this we need to rely on a stable (not likely to change) report layout.
The transformation consists of the following steps:
Eliminate the empty columns, otherwise we cannot make a pivottable. This is done in sheet “1b
Platelet stock”.
In this sheet we also use the tLumi function in column B. This transforms the LL60 names to
LUMI, in order to simplify the summation of the stock take in the pivot table (except for Flash
LL60.0134 and LL60.0163)
In sheet “1c Stock Pivot” the data of “1b Platelet stock” is summarized in a pivot table by Item
number and by bin.
Master thesis project – Roy Hartevelt Page 115
Finally, in sheet “1 Stock take” the content of the pivot table is transferred to the planning week
column. When opening this sheet the pivot table “1c Stock Pivot” is automatically refreshed and
the content is copied to the planning week automatically. The macro which enables this
functionality is located on the sheet tab (right-click the sheet tab)
Step 2 Determine WIP backend Maarheeze
Based on the input from Maarheeze we fill-in the WIP on the “2 WIP” sheet. This is the work in process
of the backend in Maarheeze.
Master thesis project – Roy Hartevelt Page 116
Step 3 Update in-transit sheet
When platelet shipments leave Maarheeze the shipment information is logged into the “Lumiramics-
Intransit” workbook. Before updating the intransit information on the “3 Intransit” the “Lumiramics-
Intranist” workbook needs to be updated with the JDE report “Completions to Inventory - Transaction
Details”. The JDE report indicates the shipments which have arrived and have been added to the
warehouse. By comparing the JDE report with the Lumiramics-Intransit workbook, the actual shipments
still in transit become visible. This process is also supported by excel-macros, in order to minimize the
manual efforts.
After updating the Lumiramics-Intransit workbook the information can be copied into the “3 Intransit”
sheet.
Master thesis project – Roy Hartevelt Page 117
Step 4 Copy the ROPs from the monthly ROP calculation
Every month we review and recalculate the Reorder Points, as the business is continuously changing.
The newly calculated ROPs need to be transferred to the weekly workbook, to these may be used in the
weekly execution to determine the replenishment quantities
Step 5 Check the master data
The master data is limited to the GGI batch sizes, the GGI/Saber yield, the lumiramics batch sizes, the
platelet distribution and the product family. Typically this data will not change much, but sometimes
adjustments are required.
Master thesis project – Roy Hartevelt Page 118
Step 6 Import SnOP and review distribution
The replenishment requirements of the platelets will closely follow the GGI starts. Also, in order to
anticipate on potential volume and/or mix changes the monthly S&OP at the product family is added. In
order to support the weekly process at bin level, we need to transform the monthly product family
quantities in to weekly bin-level quantities. We do this by dividing by the number of weeks in the month
according to the Philips calendar and by using the bin-distribution.
Typically the bin-distribution should be the same as used for the monthly ROP calculation. It can be
derived from the EPI bin distribution but it should also include the availability of pump-bins for the
product families.
Master thesis project – Roy Hartevelt Page 119
Step 7 Load GGI starts
The planned GGI starts for week N + 1 will be the best forecast for the platelet consumption in week
N+3, under the assumption that it takes GGI/Saber (in Singapore) 2 weeks to arrive at Platelet Attach in
Penang. This time window gives the backend of Maarheeze to respond to the replenishment quantities
Master thesis project – Roy Hartevelt Page 120
Step 7a Load the backlog
In case of backlog, we need to capture this separately, otherwise it is assumed to be netted in the stock
take. The backlog will also be calculated in sheet “8 Replmts”, but the actual backlog will override the
result of the calculation.
Step 8 Replenishment calculation
In sheet “8 replmts” the replenishment calculation is done. Also, all the inputs are summarized per
product family / bin in order to make the calculation result transparent.
Master thesis project – Roy Hartevelt Page 121
Let’s assume we are at the end of week 1028 and planning for week 1029. The pipeline stock can be
found in T13:T15 (taken from “1 Stock take”, “2 WIP”, “3 InTransit”). The monthly calculated ROP can be
found in T10 (from “4 ROPs”). The GGI build plan can be found in T7 (from “7 GGI Starts”) and the
Backlog in T6 (from “7a Backlog”). If the Backlog is specified in “7a Backlog” then this value is taken,
otherwise the value is calculated. The simplified version of the formula is: MAX(0; R7-T15-T14), so if the
GGI starts in week 1027 (R7) are larger than the current stock and intransit, then the remainder is
considered as backlog.
In T8 we take the GGI starts (from T7) if specified, otherwise we take the S&OP value (from “6 SnOP”).
Now we have all the inputs for the replenishment calculation: the current pipeline (T13:T15), the
expected demand (T8), the backlog (T6) and the ROP (T10). The formula is: T10-[SUM(T13:T15))-T8-T6],
so first we determine the remaining economical stock: current pipeline minus expected demand minus
backlog. This result we compare with the ROP (T10) and we reorder the difference (to order up to the
reorder level. The actual formula also does the rounding to the nearest production batch. So, the result
in T11 is the rounded replenishment request.
In T12 the confirmed quantity will be shown, this is facilitated by sheet “9c Confirmed Orders MHZ” and
explained below.
In order to understand how well the ROP replenishment is working we need to calculate to KPIs:
calculate the number of weeks economical stock in the pipeline and the quantity stock on-hand. The
first will indicate whether the ROP is still at the correct level, the second KPI will indicate whether we
have enough inventory to avoid line stoppages.
In T16 we calculate the economical stock at the end of the week: = SUM(T13:T15) + T12 - T8 - T6. So,
basically the economical stock at the beginning of the week + the confirmed supply – GGI starts –
backlog. In case the confirmed supply is not available, the calculated replenishment quantity is given
(T11). Also, if the actual information is not available (so SUM(T13;T15) = 0) then the end stock of the
previous week will be used (S16).
The quantities of row 16 are translated into week of future demand. We are using the numbers of row 8
(GGI Starts, S&OP) as input. The average demand of the first 4 weeks will be used to calculate the
number of weeks stock. The results can be found in row 17 (T17).
The on-hand projection of the stock can be found in row 18. The formula in T18 is SUM(T14:T15) - R8 -
T6, so the current on-hand plus the in-transit minus the GGI Starts of 2 weeks earlier, minus the backlog.
This on-hand projection (T18) should be greater than 0, otherwise (part of) the platelet/attach
production steps may come to standstill due to platelet shortages.
For all the product families / bins the above information has been compiled in sheet “8 Replmnts”. The
drop-down at the upper left hand corner allows to quickly access the required product family / bin.
Master thesis project – Roy Hartevelt Page 122
The details of the replenishment are given in sheets “9a Calc Replenishment”, “9b Production Orders
MHZ”, and “9c Confirmed Orders MHZ”. This will be explained in the next section.
Step 9 review the replenishment proposals and align with Maarheeze
The replenishment calculations are given in sheets “9a Calc Replenishment”. This sheet is for
information purposes only.
In sheet “9b Production Orders MHZ” the replenishments are rounded to the production batches in
Maarheeze. Also, highlighting is used to indicate priorities. The highlighting is related to the economical
stock in weeks (row 17 in “8 Replmts”). This sheet is for information purposes only.
Sheet “9c Confirmed Orders MHZ” is essentially a copy of the previous sheet, and it allows the input of
the confirmed quantities. The calculated values should be replaced with the confirmed quanties. The
Master thesis project – Roy Hartevelt Page 123
highlighting will change color when the confirmed quanties are different and the calculated weeks of
economical stock fall into a different category (< 1 wk is red, < 2 wks is orange, < 3wks is yellow, >= wks
is no highlighting, not used is blue)
Step 10 Summary of on-hand inventory projection
The “Summary” sheet shows the on-hand inventory projection. This sheet has not been finalized yet.
Maintain procedure
In this section we will describe the steps necessary to modify the sheet, e.g. when new product families
/bins have to be added to the sheet. In the following sections we will give an overview of the key
assumptions which are important to understand why certain things cannot or should not be done. Also,
we describe the macros so it becomes clear what these will do.
Master thesis project – Roy Hartevelt Page 124
Adding
The key assumptions for modifying the Weekly_Lumiramics.xlsm workbook are listed below.
• All of the sheets refer to “1 Stock take” for the first 4 columns
• New products become automatically visible after running the AddProduct macro
• This creates consistency but also limitations: moving products is prohibited
• This setup is chosen to minimize the maintenance overhead, but limits the flexibility somewhat
• 1 Stock take (continued)
– The first 4 columns (product family, 12nc MHZ, LUMI Penang, bin) are copied to all the
subsequent sheets. So this is the "master" for all item numbers.
– Adding new item numbers should be done here, and only here.
– The macro AddProduct can be used to create the referring links on all other sheets and
copy the relevant blocks of information
– OnActivate: automatically update the stock take pivot and transfer the values to the
stock take list in week N+1 The macro contains some hard-coded exceptions:
• rows 10-17, for Rebel WW also add the LUMI.0079 to the stock per bin
• rows 22-35, for Hikari only count the total stock (not per bin)
– In case the item numbers move to different rows (see general remark: this should be
avoided), the macro has no way of knowing this, as it goes by row number, and
therefore it needs to be changed to refer to the correct rows
• 6 SnOP
– The input is taken from the monthly ROP (Orientation / SnOP)
– The S&OP is at the product family level, when new products are added the division
among the bins should be added by hand.
– The first row of the first product/bin should not be moved
(because the macro assumes the first row is at row 41)
• 8 Replmts
– This sheet summarizes per product / bin all the input and calculates the replenishment
Master thesis project – Roy Hartevelt Page 125
– As it has many references to all the input sheets, we have a separate macro
"Add8Replmts" to update this sheet
– It copies the range of 12 rows and 6 columns and updates the references, then
copy/paste column 6 towards the end of December
• 9a Calc Replenishment
– Details the Replenishment calculation
– Adding new product is part of the AddProduct macro and in a separate macro:
CalcRepl9a
• 9b Production Orders MHZ
– Rounds the Replenishment requests to MHZ batch sizes
– It uses conditional highlighting to indicate urgency.
– The values used are in sheet "8 StockDays" In order for the highlighting to work with
values on a different sheet we need to use named ranges: StDays
– OnActivate: to move the highlighting to week N and N+1
– Adding new product is part of the AddProduct macro and in a separate macro:
ProdOrders9b
• 9c Confirmed Orders MHZ
– This sheet shows the values from the previous sheet (9b Production Orders MHZ) and
allows this values/formulas to be overtyped by the confirmed order quantity
– It uses conditional highlighting to indicate urgency.
– if the formula is replaced by a confirmed value, this value will show in the "8 Replmts"
sheet
– OnActivate: to move the highlighting to week N and N+1
– Adding new product is part of the AddProduct macro and in a separate macro:
ConfirmedProdOrders9c
• 8 StockDays
– This sheet support the highlighting on sheets 9b and 9c.
– The values (economical stock in weeks) are copied from the "8 Replmts" sheet
Master thesis project – Roy Hartevelt Page 126
– Adding new product is part of the AddProduct macro and in a separate macro:
UpdateStockDays8
– Manually copying the rows down will not work, because then the reference to "8
Replmnts" will be off
• Macro’s
– AddProduct
• For each sheet check with “1 Stock take” and add the missing references to the
stock take sheet (if any). Also update the drop-down list (on “8 Replmts”) to
include all the product families / bins.
– Add8Replmts
• Copy the first product / bin range, paste for the missing products, and update all
references
– CalcRepl9a
• Copy the first product / bin range, paste for the missing products, and update all
references.
– ProdOrders9b
• Copy the first product / bin range, paste for the missing products, and update all
references.
– ConfirmedProdOrders9c
• Copy the first product / bin range, paste for the missing products, and update all
references.
– UpdateStockDays8
• Copy the first product / bin range, paste for the missing products, and update all
references.
– UpdateSheet1
• Copy the product / bins from sheet 1 Stock take and redefine the named range
“ProdFam”
Deleting
Master thesis project – Roy Hartevelt Page 127
It is not recommended to delete rows / columns, because this may prevent the macros from working
correctly. We propose to work with a new file every year, and before starting the operational use, we
can clean up the sheets. We need to make sure that the data in the input sheets is cleaned up correctly.
Master thesis project – Roy Hartevelt Page 128
Kanban calculation front end Maarheeze
Model setup
Number of kanban tickets is based on;
MAT (moving average total – copy from monthly ROP) per product
Lead-time frontend as measured in the MES of Maarheeze
Service level (Z) = 99%
Standard deviation = MAT * Coefficient of variance (copy from monthly ROP)
Production batch size Maarheeze
Formula for calculating the number of kanban tickets per product:
((MAT*lead-time) + (Z * Standard deviation))/ Production batch size
Back-end MhzFront-end Mhz
FIFO
OXOX
FIFO
S&OP Lumiramics
Lumiramics
production Penang
Monthly ROP Weekly
replenishment
Master thesis project – Roy Hartevelt Page 129
Screenshot of the calculation model:
Yellow colored fields are adjustable.
The kanban calculation activities are based on the monthly ROP output. Every month the kanban tickets
of the front-end Maarheeze will be updated.
Last updated; June 2010
Product Family Lumi nmbr Png 12nc Maarheeze Bin D CV LT Q Z STD D*LT Safety stock Totaal
Luxeon Flash5 LL60.0163 3222 023 62010 330 228,538 0.34 2.94 110,000 99% 77,703 7.0 2.0 9.0
Luxeon Flash5 LL60.0163 3222 023 63010 440 533,256 0.34 2.53 110,000 99% 181,307 13.0 4.0 17.0
Luxeon Flash6 LL60.0134 3222 023 63310 222 116,751 0.71 2.34 117,000 99% 82,893 3.0 2.0 5.0
Luxeon Flash6 LL60.0134 3222 023 63410 333 817,259 0.71 2.10 117,000 99% 580,254 15.0 12.0 27.0
Luxeon Flash6 LL60.0134 3222 023 63610 444 1,050,762 0.71 2.14 117,000 99% 746,041 20.0 15.0 35.0
Luxeon Flash6 LL60.0134 3222 023 63510 555 350,254 0.71 2.17 117,000 99% 248,680 7.0 5.0 12.0
Luxeon Rebel WW LL60.0187 3222 023 61200 98,129 1.36 3.43 90,000 99% 133,455 4.0 4.0 8.0
Luxeon Rebel Hikari 2700K 80CRI T&R LL60.0170 91,847 1.00 3.47 60,000 99% 91,847 6.0 4.0 10.0
Luxeon Rebel Hikari 3000K 80CRI T&R LL60.0174 3299 999 65100 147,404 1.00 3.56 60,000 99% 147,404 9.0 6.0 15.0
Luxeon Altilon C2 LL60-0156-018 3222 023 60100 1820 11,886 0.79 2.93 40,000 99% 9,390 1.0 1.0 2.0
Luxeon Altilon C2 - 3222 023 60400 1813 62,401 0.79 2.61 40,000 99% 49,297 5.0 3.0 8.0
Luxeon Altilon C2 - 3222 023 60300 1806 72,801 0.79 2.36 40,000 99% 57,513 5.0 4.0 9.0
Luxeon Rebel PC Amber LL60.125 3222 023 64200 159,863 0.84 6.00 95,000 99% 134,285 11.0 4.0 15.0
Total # Kanban 106 66 172
# Kanban tickets
Master thesis project – Roy Hartevelt Page 130
Appendix L Glossary of terms
Term Description
APAC Asia Pacific Region
CLIP Confirmed Line Item Performance
CV Coefficient of variation (=standard deviation (σ) / average (μ))
CVP Confirmed Volume Performance
DC Distribution Centre
DP Demand Pattern
ERP Enterprise resource planning (an integrated software application)
GGI Gold to gold interconnect, production step at a Lumiled component supplier.
IDEF0 Integration DEFinition for Function
Kanban Also spelled kamban and literally meaning "signboard" or "billboard", is a concept
related to lean and just-in-time production.
LCD A liquid crystal display (LCD) is a thin, flat electronic visual display that uses the
light modulating properties of liquid crystals.
LCS Logistic department at the Component Supplier
LED A Light-Emitting Diode is a semiconductor light source.
LT Lead-time
MES Manufacturing Excellence System
MRP Master Requirement Plan
MPS Master Production Schedule
PP Production Plan
ROP Re-Order-Point
SADT Structured Analysis and Design Technique is a software engineering methodology
for describing systems as a hierarchy of functions.
Master thesis project – Roy Hartevelt Page 131
SC Supply Chain
SS Safety Stock
S&OP Sales & Operations Planning
TIP Technology Institutions Process
WOW Way Of Working
WIP Work In Process
WW World Wide
Z Service Factor