The design of an integral hierarchical concept of capacity planning for Philips Glass Roosendaal
Master thesis Industrial Engineering & Management Science University of Groningen, Faculty of Management and Organisation December 2003
Author: Eelko Huizing
Primary supervisor: Prof. dr. J. Wijngaard Secondary supervisor: Dr. Ir. D.J. van der Zee Company supervisors: Drs. G. von Morgen
B. Smet
Table of contents
Management summary 5
1 Introduction to the problem 7
1.1 Introduction 7
1.2 Philips Glass Roosendaal 7
1.3 Origin of the research project 8
1.4 Research framework 9
1.5 Capacity planning 10
2 Capacity availability 11
2.1 Introduction 11
2.2 Production process 11
2.2.1 Lead-time constraints 12
2.2.2 Volume constraints 12
2.2.3 Mix-constraints 19
2.3 Conclusion on operational characteristics 21
3 Capacity requirements 23
3.1 Introduction 23
3.2 Demand characteristics 23
3.2.1 Introduction 23
3.2.2 Push-Pull Point 24
3.2.3 Variety of products 25
3.2.4 Customer variety 25
3.2.5 Predictability of demand 26
3.2.6 Modifiability of demand 30
3.2.7 Agreements on lead-time and reliability of agreed lead-time 30
3.3 Production yields 31
3.3.1 Introduction 31
3.3.2 Production yield 31
3.4 Conclusion on capacity requirements 32
4 Manufacturing planning & control 33
4.1 Introduction 33
4.2 Current way of production control 33
4.2.1 Strategic Reviewing 34
4.2.2 Annual Operational Planning 36
4.2.3 Medium-term planning 37
4.2.4 Operational planning 38
5 Diagnostic conclusion and problem formulation 41
5.1 Introduction 41
5.2 Diagnostic conclusion 42
5.3 Objective 44
5.4 List of demands 44
5.4.1 Functional demands 45
5.4.2 Operational demands 45
5.4.3 Design constraints 46
5.5 Conclusion 46
6 MPC framework design 47
6.1 Introduction 47
6.2 Strategic Reviewing 49
6.3 Annual Operational Plan 51
6.4 Master Planning 54
6.5 Operational Planning 57
6.6 Conclusion 60
7 Information system design 61
7.1 Introduction 61
7.2 Memory functionality 62
7.3 EMPACT step-by-step 63
7.3.1 Step 1: Setting the calendar 64
7.3.2 Step 2: Forecasting demand 64
7.3.3 Step 3: Planning capacity 67
7.3.4 Step 4: Finding a feasible plan 67
7.3.5 Step 5: Generating the output 70
7.4 Formalisation 73
7.4.1 Downloading the plan 74
7.4.2 Assessing realisation 74
7.5 Concluding remarks 75
8 Organisational design 77
8.1 Introduction 77
8.2 Strategic Review 82
8.2.1 Inputs 82
8.2.2 Agenda 83
8.2.3 Output 84
8.2.4 Responsibilities and authorities 84
8.3 Annual Operational Planning 85
8.3.1 Inputs 85
8.3.2 Agenda 86
8.3.3 Outputs 87
8.3.4 Responsibilities and authorities 87
8.4 Master Planning 89
8.4.1 Input 89
8.4.2 Agenda 90
8.4.3 Output 90
8.4.4 Responsibilities and authorities 91
8.5 Concluding remarks 92
9 Implementation 93
9.1 Introduction 93
9.2 Knowledge transfer 93
9.3 IT infrastructure 94
9.4 Initial use 94
9.5 Reviewed use 95
9.6 Concluding remarks 95
Epilogue 96 Appendix 1 Glossary of terms & abbreviations 97
Appendix 2 References 98
Appendix 3 Organisational structure Error! Bookmark not defined.
Appendix 4 Products and product-machine combinationsError! Bookmark
not defined.
Management summary
Philips Glass Roosendaal sees itself confronted with growing competition on the market for glass tubes, for example from countries as China. This increased competition results in pressure on market prices and, as a consequence, profit. Simultaneously, demand for glass tubes is changing, both in volume and mix: demand is growing for the miniature tubes, whereas demand for the classical TL tubes is decreasing rapidly. The management of the glass factory is concerned about the capabilities of the current installed capacity base and the way this capacity is utilised.
However, the management indicates that it lacks insight in the effect of various investment projects or demand trends on the capacity base. This lack of insight has resulted in the initiation of this research project.
The diagnostic phase yielded insight in the characteristics of the production resources (1), the structure of demand for capacity (2) and the current way of manufacturing planning & control (3).
This phase concluded that the current manufacturing planning & control framework was insufficiently matched to the characteristics of the bottleneck resources and demand.
The demand displays a certain seasonal effect with slow demand during summers and peak demand during the first months of the year. Furthermore, production at the customer site is planned according to a certain cyclic schedule. Both aspects, however, are not incorporated in the current manufacturing planning & control framework. Moreover, analyses showed that production yields vary heavily. Not only over time, but large differences are also found between different product-machine combinations. Nonetheless, currently, one overall average yield is used in all planning & control tasks. The same accounts for the average availability of the finishing lines. The observed practical capacity (full capacity corrected for maintenance and changeovers) is not used on any of the planning levels; instead a targeted uptime is used.
With regards to the manufacturing planning & control framework, it was concluded that the strategic planning was insufficiently focused on capacity planning. The only resource incorporated in the strategic process was the furnace, whereas also the finishing lines were appointed bottleneck resources. Furthermore, the seasonal effect was ignored, not only on the strategic level, but also on the next level of annual operational planning. The next level, medium-term planning, was the most underdeveloped level at Philips Glass Roosendaal. The current ad-hoc method of medium-term planning involves a lot of information collection, rumour checking, spreadsheet design and gut feeling. The operational planning functioned fairly well, however, no way of assessing the performance of the operational planning was at hand. Lastly, the lack of formalisation on each level resulted in unclear, badly communicated and preserved plans. Plans were in no way related to each other and the realisation of plans was never assessed.
The redesign consists of three parts: a conceptual manufacturing planning & control framework, an information system and a design of the organisational aspects related to executing the manufacturing planning & control tasks.
The conceptual design is embodied by reshaping the following aspects on each of the hierarchical levels, from strategic planning to operational planning
• Objective
• Planning horizon
• Time bucket
• Planning frequency
• Product aggregation
• Incorporated resources
The upper planning level, Strategic Review, is directed towards the long-term (1-4 years) capacity planning. Investment projects are formulated here and it constrains lower level plans by setting the capacity base for the coming years. Strategic Reviewing is done on a yearly basis by the management team, given the strategic impact of the plan. The next level, the Annual Operational Planning, supports the planning of capacity for seasonality. On this level finishing line downtime and furnace outputs are planned for the coming twelve months. The plan thus constrains the monthly finishing line capacity and sets (seasonal) inventory targets for lower planning levels. The PU manager, the Material manager and the Line team manager perform the process of AOP planning four times per year. The next level is the Master Planning. It foresees the coming quarter and schedules finishing line stoppage. It also forms the basis for the process of order promising at the glass factory. Master Planning is done on a monthly basis by the Line team manager and the Material manager. The fourth level, the Operational Planning, schedules production orders for the coming two weeks, taking customer manufacturing-schedules into account.
The second part of the redesign entails the development of an information system supporting the various manufacturing planning & control tasks on the upper three levels: Strategic Reviewing, Annual Operational Planning and Master Planning. The information system, denominated EMPACT, consists of five Excel workbooks. An important characteristic of EMPACT is that enables the user to perform so-called scenario- or what-if analyses and make trade offs between, for instance, reducing seasonal inventory and decreasing planned finishing downtime. On the other hand, issues such as the impact of an annual increase of 15% of the demand for T5 products on the capacity utilisation can easily be evaluated. The difference does not lie in the tools necessary to answer questions as such, but rather in the kind of analyses done by the user.
The third part of the redesign consists of the design of the organisation. An important part of this phase was the development of a so-called Term-of-Reference sheet for each hierarchical level. It was concluded that having a structured meeting helps formalising the required tasks, responsibilities and authorities of each individual involved. Each of the four Term-of-Reference sheets states the practicalities of the meeting, such as time, place, frequency and attendants.
Secondly, it states the objective of the meeting, the required inputs and the expected outputs. The meeting gets structured by determining the last part of the meeting, the agenda. Every agenda is roughly the same. The meeting starts out with looking back to the previous plan and assessing the realisation of this plan. Secondly, it discusses the demand forecast, the expected production yields and the availability of the finishing lines. Lastly, on the basis of the inputs and the previous two agenda items, decisions are taken and actions are formulated.
Currently, the project is in the phase of implementation. User manuals are being written and distributed and various training sessions are held to train the user both in the understanding of the concepts and the use of the developed tools. It is advised to “just” start with planning the meetings and using the EMPACT information system. This will greatly improve the understanding of and trust in the set of tools. It is to be expected that the concepts and tools developed will evolve over time and be adapted to changes either within the organisation or its environment. However, it is advised to schedule an official evaluation to identify most urgent needs for adjustment after one year.
The starting point of design should always be that the MPC framework accounts for the most
dominant characteristics of both the production structure and the demand structure, possibly
leading to a hierarchical approach in which multiple levels of MPC tasks are distinguished. It is
1 Introduction to the problem
1.1 Introduction
The focus of the first chapter is the exploration of the problem area. First, a general introduction will be given on Philips Glass Roosendaal and its position within Philips. The origin of the research project will be discussed in paragraph 1.3. Since the management formulated a rather vague idea of the problem area, a diagnosis was performed to sharpen the problem definition.
The last part of this chapter will discuss the way the diagnosis is carried out.
1.2 Philips Glass Roosendaal
Royal Philips Electronics is divided into five business units: Consumer Electronics, Domestic Appliances and Personal Care (DAP), Lighting, Medical Systems and Semiconductors. The Lighting division is subdivided into four business groups: Automotive & Special Lighting, Lamps, Lighting Electronics and Luminaires. The business group Lamps is further divided into four regional business units, Components and GTD (Global Technology Development).
Philips Glass Roosendaal is a subdivision of Lighting Components, which consists of nine factories in total, of which three TL glass factories, Roosendaal Glass, Pila Glass (Poland) and Chalon Glass (France). An overview of the organisational structure can be found in Error!
Reference source not found..
The Roosendaal production site accommodates, besides Philips Glass Roosendaal, the supply- group TL/CFL-ni, which is a part of Philips Lighting Lamps Europe and Special Lighting. The supply-group is made up of four factories in total: Roosendaal, Terneuzen, Pila (Poland) and Chalon (France). The Roosendaal production site thus accommodates parts of Lighting Components, Lamps Europe and Special Lighting.
Philips Glass Roosendaal produces various glass tubes that are used in the production of TL lamps (Figure 1-1). The factory finds itself at the outset of the supply chain, where raw materials are turned into semi-finished products, which are further processed at the customer site, located on the same production plant to save costs of transportation. The final products are transported to the final customer through regional warehouses.
W D
L
Figure 1-1 Basic shape of a glass tube
Tubes with smaller diameter are the so-called TLd tubes, which are the most widely used TL
lamps. TL tubes have greater diameters and can be found in, for instance, sun tanning products,
such as sun beds. Tubes without flanges (Dutch: kraag) are so-called straight tubes. Lastly, the
T5 tubes are miniature tubes used in, for instance, signs for emergency exits. Apart from this, the
Roosendaal Glass factory produces “starter-glass” as well, a component used in the starters of
TL-lamps.
Production of glass tubes is performed in roughly four steps. First, a set of raw materials is mixed and fed into the furnace, where it is fused at a temperature of approximately 1500°C. The furnace is connected to three glass-drawing tracks. On these tracks a long, hollow glass tube is formed and split at the end into discrete tubes. The drawing tracks are connected to the so-called finishing lines. Here, the tubes are rounded and, if necessary, a flange is created. Finally, the tubes are automatically packed into metallic holders and transferred to the warehouse by fork-lift- trucks.
1.3 Origin of the research project
Philips Glass Roosendaal sees itself confronted with growing competition on the market for glass tubes, for example from countries as China. This increased competition results in pressure on market prices and, as a consequence, profit. Simultaneously, demand for glass tubes is changing, both in volume and mix: demand is growing for the miniature tubes (T5), whereas demand for the classical TL tubes is decreasing rapidly. The management of the glass factory is concerned about the capabilities of the current installed capacity base and the way this capacity is utilised. On the one hand, capacity must be used up to the maximum of its capabilities to meet the requirement of cost price reduction but, on the other hand, capacity might be not suited for the developments in demand, both volume and mix.
The management of the glass factory has formulated several possible courses of action to overcome these problems. In the case of under capacity, an option is to expand the current capacity by investing in additional machinery or by increasing the speed of the current machines.
Another possibility is to allow for more inter-factory deliveries between all three Philips glass factories in Europe (Roosendaal, Chalon and Pila), which could lead to a more optimal situation.
In the case of overcapacity, sales to third parties can increase utilisation or capacity can be
slowed down during slow demand. This leads to either cost reduction or increase of turnover and
thus strengthens the position of Philips Glass Roosendaal. However, the management indicates
that it lacks insight in the effect of various investment projects or demand trends on the capacity
base. This lack of insight has resulted in the initiation of this research project.
1.4 Research framework
The thesis will be structured according to the framework proposed in Figure 1-2. Roughly, the thesis is divided into four parts: start – diagnosis – redesign – end. The start covers the motivation for the start of the project, which has been covered in the preceding discussion on the origin. The diagnostic phase is aimed at finding out why the management team of the glass factory lacks insight in the effects of various changes and projects on the capacity base. This phase is structured by the concept of capacity planning, which is made up out of three elements: capacity requirements, capacity availability and manufacturing planning and control. The diagnosis ends with a conclusion on the main problems, which serves as the starting point for the third part, the redesign.
Origin
Manufacturing planning
& control
Diagnostic conclusion
Implementation StartDiagnosisRedesignEnd
Capacity
requirements Capacity availability
Organisational structure IT
structure
MPC framework
Figure 1-2 Research framework
The redesign entails three integrated elements, a manufacturing planning & control framework
together with an information system and organisational design. The thesis is concluded with a
overview of the implementation issues. The next section will introduce the concept of capacity
planning, by which the diagnostic phase will be structured.
1.5 Capacity planning
The issues the management of the glass factory sees itself confronted with, are all capacity related. Investing in additional machinery or increasing machine-speeds could expand capacity;
on the other hand, capacity utilisation depends on volume and mix developments and/or the acceptance of orders from the external spot-market. Capacity planning will therefore be the central theme of the diagnostic part of the thesis. The managerial objective in planning capacity is to ensure the match between capacity available in specific work centres and capacity needed to achieve planned production. (Vollmann et al., 1997, 120) When looking more closely at the definition, three components can be distinguished; capacity availability in specific work centres, capacity needed (or capacity requirements) and planned production (or manufacturing planning &
control) (Figure 1-3).
Manufacturing planning & control Capacity
requirements
Capacity availability in specific work
centres match
match
Figure 1-3 Capacity planning
Capacity availability in specific work centres relates to the capabilities of the production set-up, referred to as the operational characteristics. Specific work centres should be interpreted as critical resources or bottlenecks. Capacity availability relates to issues such as the number of production days and shifts per year, the possibility to work overtime and/or the time devoted to maintenance or changeovers. Attention will also be drawn upon the technical restrictions of the production set-up, for instance, the maximum machine speeds from a technical perspective or the highest achievable furnace output.
The second aspect, capacity requirements, relates to all aspects occupying the available capacity.
Naturally, product demand is of major interest here. Insight will thus be developed in the number of customers and the characteristics of each customer. Furthermore, issues such as seasonality and the distribution of demand over the different products will be covered as well. The second component, which generates a demand for capacity, is the production yield. Quality problems lead to decreasing yields, thereby increasing the capacity requirements.
The last aspect relates to production planning and control. Bertrand et al. (1990, 17) define production control as the coordination of supply and production activities in manufacturing systems to achieve specific delivery flexibility and delivery reliability at minimum costs. Thus, attention will be drawn upon items such as the way production orders are scheduled, the way customer orders are accepted or the hierarchical layering of the production planning & control concept.
The three aspects from the capacity planning definition will be discussed in the subsequent
chapters, starting with the operational characteristics in chapter 2. After this, demand
characteristics and production planning and control will be covered.
2 Capacity availability
2.1 Introduction
Let us first recall the definition of capacity planning. Capacity planning was defined as the assurance that capacity available in specific work centres matches capacity needed to achieve planned production. This chapter will cover the first part of this definition: capacity available in specific work centres. When we take a closer look at the definition, two remarks can be made.
First, specificity is defined as relevant to capacity planning; work centres as such are often referred to as bottleneck resources. Secondly, capacity availability can be defined in various ways, for instance either as the theoretically achievable output or some fraction of it? This chapter will thus answer the following two questions:
• What are specific work centres at Philips Glass Roosendaal?
• What capacity is available in the specific work centres?
These questions will be answered by discussing three aspects that characterise a production set- up (Bertrand et al., 1998, 331):
• Lead-time constraints
• Volume constraints
• Mix constraints
Lead-time consists of the sum of non value-added time (waiting, inspection) and value-added time (actual processing time). Constraints on volume are typically the result of limited available capacity on the bottlenecks (specific) resources. Mix-constraints, in conclusion, are usually the result of long set-up times between batches that limit efficiency. These set-up times can in turn be sequence dependent. First, a short introduction is given on the production process. This is further detailed in the subsequent discussion of lead-time, volume- and mix-constraints.
2.2 Production process
Production of glass tubes is performed in several steps. A set of raw materials is mixed and fed into the furnace, where it is fused at a temperature of approximately 1500°C. The furnace is connected to three glass-drawing tracks. On these tracks a long, hollow glass tube is formed and split at the end into discrete tubes. The drawing tracks are connected to end-forming machines.
One of the drawing tracks has a split at the end of the track, therefore, in total four end-forming
machines (1A, 1B, 2B and 3B) are connected to the tracks. On the end-forming machines the
tubes are rounded and, if necessary, a flange is created. Finally, the tubes are automatically
packed into metallic holders and transferred to the warehouse by fork-lift-trucks. Figure 2-1
summarises the production process and will be used in the analysis of the volume, mix and lead-
time constraints in order to study the production set-up.
2.2.1 Lead-time constraints
Lead-time consists of both waiting- and processing-time. Waiting-time is the result of a lack of tuning of availability between the subsequent processing stations. (Bertrand et al, 1998, 333) Yet, routings in the Roosendaal Glass factory are limited and simple. Since no in process inventory is kept, no waiting-time occurs and lead-time is totally made up of processing times. Furthermore, after a manufacturing batch has been set-up, lead-time is almost negligible. The major issue is therefore not lead-time but tact time or production output in pieces per minute. This production output is dependent on machine speeds. This matter will be discussed in the next section on volume constraints.
constraints on
Furnace
1A
1B
2B
3B Drawing track
"right"
Drawing track
"left"
Drawing track
"middle"
Splitter
lead-time
volume
mix
Figure 2-1 Production resources
2.2.2 Volume constraints
Bertrand et al. (1998) state that volume constraints are the result of limited available capacity.
Nevertheless, available capacity is not as straightforward as it may seem and it can be defined in
several ways. Vollmann et al. (1997,143) state that the first step in choosing a capacity measure
is to identify resources that are short in supply. This is parallel to the objective of this chapter to
identify bottleneck resources. The second step is to define the unit of measure, for instance
machine-hours, labour hours or physical units. The final step is to estimate available capacity, or
the volume constraint. McNair et al. (1998, 27) define available capacity as the amount of work a
resource can support. However, it is stated that many different views exist about which baseline
or maximum capacity estimate is most logical. Vollmann et al. (1997, 143) conclude that the
primary issue is theory versus practice. An engineer can provide theoretical capacity from the
design specifications of a machine. An issue therefore is whether to use “full” (theoretical)
capacity or some fraction thereof. It is concluded that the measure should be based on what is
achievable, with allowances for maintenance and other necessary activities. To follow the
objective of this section, to detail the volume constraints of the production set-up, two measures of
available capacity will be used: theoretical capacity and practical capacity. Theoretical capacity is
by definition the higher of the two. It allows for no adjustment for, for instance, maintenance,
shutdowns or failure. Dependent on the chosen measure it may be 100% of available production
time or the maximum achievable output. The second measure is practical capacity. Practical
capacity equals theoretical capacity corrected for aspects such as maintenance, failure or
changeovers. All steps involved in the assessment of bottleneck resources and their theoretical
and practical capacities are summarised in Figure 2-2.
The following sections discuss the volume constraints of all resources using the mentioned steps.
Step 1 Potential bottleneck
resource?
Step 2 Define unit of
capacity measure
Step 3 Estimate theoretical
capacity
Step 4 Estimate practical capacity
Exit yes
no
Figure 2-2 Discussion of volume constraints
• Storage of raw materials
Fusing a number of raw materials, such as sand, soda and chemicals, produces glass used in the production of TL tubes. The raw materials are stored in silos and are refilled on a predetermined frequency. This frequency is directly derived from the annual average operating rate of the furnace. The size of the silos, together with the short replenishment lead times, make that the silos can be considered as non-critical resources. Availability of raw materials will not become critical, even when the furnace operates at maximum output.
• Furnace
The furnace is operated around-the-clock all year long and stopped only once in every seven years for overhaul. The furnace is capable of producing a total output of xxxx kg/h, which equals approximately xxxx tons of glass annually. This output is believed to be feasible.
Consequently, the furnace will pose volume constraints when annual output exceeds these xxxx tons of glass and is therefore a potential bottleneck resource. The capacity measure has implicitly already been chosen, glass output in tons per time period. The average downtime of the furnace over 2002 and the first part of 2003 was x%. This leads to the following volume constraints:
− Capacity measure: glass output in tons per time period
− Theoretical capacity: xxxx tons annually (xxxx * 365 * 24)
− Practical capacity: xxxx tons annually (xxxx*365*24)*(1 - xxxx)
Since no in-process inventory is kept, furnace downtime directly results in the unavailability of the finishing lines. Furnace failure will therefore be incorporated in the practical capacity of these finishing lines. Nevertheless, it should not be overlooked that the reduction of furnace downtime results in the relaxation of the accompanying volume constraint.
• Drawing tracks
On the drawing tracks, the glass is gradually cooled and shaped into a hollow tube conforming
diameter and wall thickness specifications of the tube being manufactured. Since the tracks
can keep up with the furnace at maximum output (xxxx kg/h), the drawing tracks pose no
volume constraint on the internal production structure. Hence, concerning volume, drawing
tracks will not be a critical resource. Nevertheless, drawing track downtime directly leads to
finishing line unavailability, for the same reason as mentioned above. As the drawing tracks
are not classified as a critical resource, its failure will be incorporated in finishing line
(un)availability. The drawing tracks will thus be left out of the rest of the analysis.
• Finishing lines
The finishing lines are used to finalise and pack the glass tubes. The ends of the tube are rounded and, if necessary, a flange is created. In total, four finishing lines are present, referred to as 1A, 1B, 2B and 3B. Finishing lines 1A and 1B are connected to the same drawing track by the use of a splitter. The finishing lines are capable of producing a specific set of products.
Finishing line 2B, for instance, is the only line capable of producing TL diameters, whereas the 3B finishing line is unique in its capability of producing T5 products and Starter glass.
Furthermore, the 3B finishing line cannot produce tubes with flanges. The length of the tube, apart from its diameter and flange, is the third aspect determining the finishing line compatibility. However, the finishing lines offer a limited amount of flexibility. A number of products can be produced on several finishing lines, providing flexibility to the production set- up. Examples of these products are given in the subsequent table.
Product 1A 1B 2B 3B TLd 36i ● ● ● TLd 18i ● ● TLd 58i ● ● TLd 30i ● ● TLd 23i ● ● TLd 38i ● ● TLd 16i ● ●
TLd 15r ● ●
TLd 58r ● ● ●
TLd 36r ● ●
Table 2-1 Capacitive flexibility
In practice, though, the so-called straight tubes (TLd 15r, TLd 58r and TLd 36r) are produced on the 3B finishing line. This line is not capable of producing tubes with flanges; therefore, it is dedicated to producing straight tubes. The largest part of the product portfolio can be produced only on a single finishing-line and thus poses constraints on the internal structure. An overview of possible glass tube - finishing lines combination can be found in Error! Reference source not found..
The output of the finishing lines (pieces per minute) is calculated using the following formulas, where every formula corresponds to a specific finishing line:
3B: Min
Furnace output drawing track 3B (grams/min) Tube weight + overcut weight (grams)
; maximum machine-speed 3B 2B: Min
Furnace output drawing track 2B (grams/min) Tube weight + overcut weight (grams)
; maximum machine-speed 2B 1A/1B: Min
Furnace output drawing track 1A/1B (grams/min) Tube weight 1A + tube weight 1B + overcut weight (grams)
; maximum machine-speed 1(A/B)
The furnace output used in the formulas is the glass flow from the furnace on the drawing track connected to the respective finishing line. The tube weight corresponds to the required amount of glass for a specific tube or tubes in the case of the combined 1A-1B finishing lines. The overcut weight is (the weight of) a small piece of glass that is cut off from both ends to bring the tube to the correct length. The maximum machine-speed, in conclusion, is a hard constraint on the output rate of the finishing lines. This maximum-speed is not product dependent; it is merely a consequence of the technical characteristics of the finishing line.
A number of conclusions on the machine-speed determination can be drawn when we take a closer look at the presented formulas:
− Furnace output ~ machine speeds
An increase in furnace output increases machine speeds until the technical constraint on machine speed is reached.
− Tube specification ~ machine speeds
Smaller (lighter) tubes can be produced at higher speeds than heavier tubes. This formula implies that for the smaller tubes the maximum machine speed constrains volume. For tubes of greater length, the furnace output can become the bottleneck resource.
− Machine specification ~ machine speeds
The technical constraint of the finishing line functions as an absolute upper bound on machine-speeds.
These three points may be clarified in the following example
From the above, it can be concluded that the finishing lines are potential bottleneck resources and should thus be included in the analysis. The next step is the formulation of a capacity measure. The measure used for the furnace operation, output, is not suitable here because of the product- and furnace dependency of the machine-speeds: the capacity measure “pieces
The graph shows that the lighter TLd36i tube can be produced on the 2B finishing line at higher speeds than the heavier TL100i or TLd70i tubes. Machine-speed increases until the maximum machine-speed is constrained by the technical specifications of the finishing line, in this case 120 pieces per minute.
Machine -speed determination 2B
0 20 40 60 80 100 120 140
400 600 800 1000 1200 1400 1600 Furnace output (kg/h)
Speed (pcs/min)
TLd36i TL100i TLd70i Max speed
per minute” is meaningless since this depends on what product are manufactured. For instance, the annual capacity of finishing line 2B would be either xx million TL40i tubes or xx million TLd36i tubes. A more meaningful capacity measure is available machine hours, or the time the finishing lines are available for production. The notion of theoretical capacity is rather straightforward; the finishing lines are operated in a five-shift system, 365 days a year.
Theoretical capacity thus equals 365 days multiplied by 24 hours a day.
1The quantification of practical capacity involves more clarification. Practical capacity here is defined as the fraction of theoretical capacity available for production. The difference between the two is formed by downtime as a result of the following activities:
− Preventive maintenance
Preventive maintenance is made up out of unavailability as a result of the periodic maintenance activities on the finishing lines.
− Corrective maintenance
Corrective maintenance consists of downtime as a result of failure of the finishing lines.
Furthermore, as stated in the previous section, downtime of both the furnace and the drawing tracks immediately leads to finishing line unavailability since no buffer stock is used. Thus, failure of the preceding resources is included here as well.
− Changeovers
The process of changing over a finishing line to another product can be divided into four phases, illustrated in Figure 2-3. The finishing line is shut down at the end of phase A and the adjustments of all machines are completed in phase B. The finishing line is started up again in phase C and production yields will slowly increase up to the normal level in phase D. The finishing line is completely unavailable for production only during phase B, whereas some output is generated in phase C. Changeover unavailability is thus defined as the time between shutting down the finishing line for set-ups and the moment the lines are restarted.
The impact of varying yields (phase C) is included in the production yields discussed in the next chapter.
Phase A:
production batch one
Phase B:
shutdown for changeover
Phase C:
start-up yields
Phase D:
production batch two Yields
Time
The following results on finishing line availability were obtained from an analysis of 18 weeks of production data, measured on a weekly level.
Standard deviation Finishing line Average practical
capacity Absolute Relative
1A xxxx xxxx xxxx %
1B xxxx xxxx xxxx %
2B xxxx xxxx xxxx %
3B xxxx xxxx xxxx %
Table 2-2 Average and variance of finishing line availbility
It can be concluded that the availability is the lowest on finishing line 3B, scoring xxxx %.
However, the variance is over xxxx % on three out of four finishing lines, which equals almost xxxx hours on week level.
• Warehouse storage
Philips Glass Roosendaal uses a climate-controlled warehouse for storage of finished goods.
Normally, tubes are stored in metallic holders (Figure 2-4, left and middle) of which a limited number is available. This implies that not the actual space in the warehouse is constraining but the number of available metallic holders. The facility in Roosendaal has approximately 2850 metallic holders at its disposal. Since the holders are used to carry work-in-progress in all production facilities, a fraction of the total amount is available to the glass factory. Current estimations are 65%. Additionally, the number of tubes per metallic holder depends on the characteristics of the tube.
2The number of tubes per holder is larger for short tubes with small diameters than for, for instance, long TL tubes. This implies that the actual constraint is dependent on product mix, as was the case with the finishing lines.
When tubes are produced for external customers, tubes are manually repacked into carton boxes supported by a pallet board (Figure 2-4, right). In the case of unavailability of metallic holders, these boxes can be used as a temporary solution as well. Naturally, in this case, available storage space will be the constraining factor. Since, storage in carton boxes requires extensive labour hours for repacking, this alternative should be used with caution.
The chosen capacity measure here is the number of metallic holder available to the glass
factory. Theoretical capacity is therefore 2850 holders and practical capacity is set at 1850
holders
3.
2.2.3 Mix-constraints
Mix-constraints are usually the result of long set-up times between batches that limit efficiency.
Therefore, the analysis of mix-constraints will focus on the issue of changeovers.
• Storage of raw materials
Glass is a mixture of several components, such as sand, soda, chemicals such as iron oxide and complemented with cullet, either bought externally or scrapped products from the production process. Although the components of glass are fixed, the composition of the mixture may vary over time; the amount of external cullet may be increased during high internal production yields to stabilise the total amount of cullet in the mixture. Another possibility is that the amount of iron oxide is modified in order to influence the UV transmission.
An example of this will be given below. However, changing the composition of the mixture is a relatively straightforward process, which can be accomplished virtually immediately. Storage of raw materials therefore does not result in any loss of capacity in the event of mixture changes
• Furnace
The glass composition is equal for all products in the current product portfolio. This means that the furnace operation does not constrain the mix of the internal production set-up.
Nevertheless, it is imaginable that products demanding a different glass composition are added to the product portfolio. In that case, the furnace dramatically constrains mix flexibility.
Changing over the furnace requires changing the mixture of ingredients, washing out the furnace and producing cullet with the required mixture. Total time for such a changeover is estimated at one week. Historically, furnace changeovers were not considered since no demand for other glass compositions existed. However, the production unit Special Lighting recently inquired into the possibility of production of Open UV Glass, a type of glass used for tanning products. This glass requires the removal of iron oxide from the mixture. In order to do this, the furnace has to be washed, which takes approximately 12 days. Currently, the furnace does not mix constrain the product portfolio; yet this may change in the future.
• Drawing tracks
The diameter and wall thickness are formed on the drawing track. Hence, when a change in
either of these occurs the drawing track has to be changed over. The process operators
perform this changeover. However, in comparison to the changeover of the finishing lines, the
duration of the changeover can be neglected. Thus, concerning mix, the drawing tracks are not
a critical resource.
• Finishing lines (1A/B; 2B; 3B)
The first part of the loss due to set-ups, the actual standstill, was covered in the paragraph 2.2.2. The remaining part will be discussed here. Set-up times between batches of dissimilar tubes can be extensive and are also sequence dependent on diameter and length.
Furthermore, for certain types, set-ups are structurally higher than for other types. Set-up times range from ten minutes to an hour and the ‘warming-up’ period can last up to several days. As a consequence, set-up duration results in a loss of available capacity (as covered in the section on volume constraints) and the duration of warming-up periods results in varying yields. The duration of these warming-up periods are currently not measured, which blurs the available data on production yields. Changeovers can be measured on a three-point scale of discomfort (Table 2-3). In spite of the qualitative nature of this data, it is illustrative of the sequence dependency of the set-ups. The issue of production yields will be covered in the chapter on needed capacity, the main issue here is that both set-up duration and yields during warming-up periods constrain the mix flexibility of the production set-up.
Type of set-up Discomfort Diameter & length - - -
Diameter - -
Length - Table 2-3 Discomfort of set-up
• Warehouse storage
The use of metallic holders or carton boxes does not constrain the mix of the internal
production structure. All holders can be used for all different types and changeover times are
thus not of relevance.
2.3 Conclusion on operational characteristics
The preceding section covered the discussing on available capacity and was aimed to answer two questions:
• What are specific work centres in the production set-up of Philips Glass Roosendaal?
• What capacity is available in the specific work centres?
The discussion was structured by using the aspects of lead-time, volume constraints and mix- constraints. Lead-time was concluded to be almost negligible, whereas output was the main issue in the production set-up. A capacity measure was defined for all bottleneck (specific) resources and the theoretical and practical capacity were quantified in the analyses on volume constraints.
Mix constraints, in conclusion, were analysed by looking at changeovers. This is summarised in the subsequent table.
Constraint Resource
Volume Mix
Capacity measure unit Theoretical capacity Practical capacity Storage of raw materials
Furnace ●4 о Output (tons/year) xxxx xxxx
Drawing tracks
Finishing line 1A ● ● Total production time 100% xxxx %
Finishing line 1B ● ● " " xxxx %
Finishing line 2B ● ● " " xxxx %
Finishing line 3B ● ● " " xxxx %
Warehouse storage ● Number of metallic holders 2850 1850
Table 2-4 Operational characteristics overview
It can be concluded from the table that all resources, except for the storage of raw materials and the drawing tracks, are potential bottlenecks. These specific resources are therefore relevant in capacity planning. The practical capacity rates from Table 2-4 are defined to be the capacity available in these resources, in answer to the second question. However, the capacity measure chosen for both the finishing lines and the warehouse storage deserves some extra attention. An alternative capacity measure for the finishing lines would be the output in pieces per minute. This would relate the demand for tubes directly to the needed production time. This capacity measure is not used because of the interdependency between production speed, product type and furnace output. This interdependency is also observed in the warehouse storage, since different glass tubes consume a dissimilar metallic holder space. The main point here is to illustrate the interdependency between product mix and capacity usage and the choice for a capacity measure unit derived from this interdependency.
When discussing mix constraints, the divergent nature of the production process is important.
Because of the commonality in the glass composition of all tubes in the current product portfolio, both the storage of raw materials and the furnace pose not constraints on mix. However, when tubes of a different glass compound (e.g. Open UV Glass used for tanning products) are considered, the impact of furnace on mix constraints is considerable.
4 ● = potential bottleneck resource; о = potential bottleneck in case of open UV glass
Furnace
1A
1B
2B
3B
Figure 2-5 Resources relevant to capacity planning
The resources identified as potential bottlenecks have to be included in the process of capacity
planning and are classified as specific resources. The non-bottleneck resources can be omitted
from the production set-up used for capacity planning (Figure 2-5).
3 Capacity requirements
3.1 Introduction
Let us first return to the definition of capacity planning: capacity planning ensures capacity available in specific work centres matches capacity needed to achieve planned production. The first part of this definition was discussed in 2. This chapter will relate to the second part of this definition. This part entails needed capacity. Two aspects are the determinants of total required capacity (Figure 3-1):
tube demand
(pcs)
÷
machinespeeds=
of minutesgross no.÷
productionyields=
net no. ofminutesgross no.
of minutes
Figure 3-1 Origin of capacity requirements
• Demand for glass tubes
Ultimately, the customer consumes capacity by ordering glass tubes; in the case of the absence of a customer, the need for capacity would be zero.
• Production yields
The yields of the production process also have a considerable impact on needed capacity.
When, for instance, a customer orders 90 tubes and machine speed would be one piece per minute, gross capacity requirement would equal 90 production minutes. However, when the production yield turns out to be 90%, the net capacity requirement would be 100 minutes. In that case, the customer consumes 90% of the needed capacity and the internal yield consumes 10% of the needed capacity.
This chapter thus deals with both the characteristics of demand and the production yields realised in the process. Demand will be covered in the following section and yields will be discussed and the successive section.
3.2 Demand characteristics 3.2.1 Introduction
Bertrand et al. (1998, 320, 324) state that the first step in modelling demand characteristics is to set the position of the Push-Pull Point (Dutch: KOOP). The Push-Pull Point (PPP) is the point where production switches from push (unconnected to actual customer orders) to pull (connected to actual customer orders). Subsequently, the characteristics of demand will be analysed using the demand aspects derived from Bertrand et al. (1998, 325-326):
• Variety of products on the Push-Pull Point
• Predictability of demand
• Modifiability of demand
• Agreements on lead time and delivery reliability
As stated, the position of the Push-Pull Point must be determined first and consequently, the aspects characterising demand downstream of the Push-Pull Point will be integrated in the presented model of the bottleneck resources. The analysis of these aspects will be based on that model.
3.2.2 Push-Pull Point
The identification of the Push-Pull Point is not as clear as one would expect from theory. At Philips Glass Roosendaal, products are characterised as either fast-movers or slow-movers. Fast-movers are interpreted as products with frequent demand. On the other hand, slow movers are products with only sporadic demand. Before discussing the Push-Pull Point of both fast- and slow movers, the way customer orders are generated will be covered first.
The customer enters their production planning for the coming week into the SAP-system. This production planning is stated in lamps as an output measure. This planning is translated into a glass requirement by incorporating a certain yield loss on the customer production lines. These glass plans could be interpreted as being customer orders (Figure 3-2).
Customer pr oduction
planning (lamps)
Glass planning
(tubes)
customer order customer
yield loss
Figure 3-2 Customer orders
For the fast-movers, the customer order is translated into a production order for the coming week.
Batch-size is calculated by comparing stock levels to the customer order. In some occasions, the customer order is delivered directly from inventory. Batch sizes are increased in other instances to built up some stock to satisfy expected demand in the coming weeks. The production of these fast-movers thus entails features of both make-to-order and make-to-stock. The production of slow-movers is similar to make-to-stock. When the inventory levels drop below a reorder point, the MRP system proposes a production order to refill stock. However, these production orders may or may not be released to the factory by the material manager, depending on information on upcoming customer orders. Nevertheless, slow-mover products are mainly controlled on stock level.
Regarding the PPP of the production process, an upstream upper bound is set on the position of
the PPP by the features of the furnace resource. Resulting from the fact that the furnace cannot
be stopped it is thus by definition unconnected to customer orders. All remaining other operations
can be done on customer order. The first PPP is therefore the furnace operations and covers the
products with make-to-order features or the fast-movers. Since a PPP is by definition a stock
point, the furnace is thus perceived as a stock of fused glass. The second PPP is the end stock of
finished goods and covers the products with make-to-stock features or slow-movers. Nonetheless,
the remark must be made that these PPP are not as black-and-white as suggested here. Figure
3-3 summarises the position of the PPP and matches them with the demand characteristics.
Furnace
1A
1B
2B
3B
activities connected to customer orders activities unconnected
to customer orders PPP1
PPP2
activities unconnected to customer orders activities connected to customer or ders
y variety of products y variety of customer s y predictability of demand y modifiability of demand
y agreements on lead-time & - reliability
Figure 3-3 Demand characteristics
3.2.3 Variety of products
The current product portfolio is made up out of 32 different glass tubes. An overview of the total product portfolio can be found in Error! Reference source not found.. Although the product portfolio is relatively large, a limited number of products represents a large number of total annual production. This is shown in Table 3-1.
Product Production (Pcs/yr)
Relative (%)
Cumulative (%)
Production (Tons/year)
Relative (%)
Cumulative (%) TLd 36i xxxx 36.94 36.94 xxxx 39.27 39.27 TLd 18i xxxx 21.55 58.48 xxxx 11.14 50.41 TLd 58i xxxx 14.83 73.31 xxxx 19.76 70.17 TL 100i xxxx 3.52 76.82 xxxx 9.60 79.77 T5 28W HOR xxxx 3.48 80.30 xxxx 2.64 82.41 TLd 30i xxxx 3.47 83.77 xxxx 2.74 85.15 Starter glass xxxx 3.35 87.12 xxxx 2.97 88.12
Table 3-1 Product portfolio (initial estimations for 2003)
This table shows that approximately 7 (out of 32) products represent 88% of the annual production in tons or approximately 87% of the annual production in pieces.
5Or stated differently, 22% of all products represent 88% (tons) or 87% (pieces) of total production.
3.2.4 Customer variety
As mentioned in the introduction, the total customers base is very small. The glass factory mainly serves the next-door (internal) customers. These customers can be subdivided into two production units (PU):
• PU Tube Lighting (TL)
The PU-TL produces TLd and TL5 lamps in very large quantities and is, by far, the largest customer of Philips Glass Roosendaal. The PU-TL accounts for approximately 89% of annual
5 Estimations for 2003/2004
demand stated in pieces. The PU-TL manufactures lamps on several finishing lines: the so- called HOR lines (6000/1; 6000/6; 3000/1; A1), referring to the horizontal orientation of the lamps during the process and the VTL3 line, where lamps are produced vertically.
• PU Special Lighting (SL)
The PU-SL produces smaller series of TL, TLd-r and T5 lamps. The lamps produced here are specialty products, for instance used in photocopiers, sun beds and disinfection-lamps. The PU-SL accounts for approximately 7% of annual demand. The PU SL has three production lines at its disposal, the HOR3000/3, VTL5 and the MPSTL (Multi Purpose Standard Lighting).
Apart from the internal customers, the Lommel lamp factory accounts for the remaining demand (approximately 4%), which is fully made up of the starter glass.
3.2.5 Predictability of demand
Considering the number of customers one would expect a rather predictable demand since a small number of professional customers is being served. (Bertrand et al, 1998, 47) Furthermore, using the lifetime reliability of TL-tubes, replacement investments could be used to calculate derived demand. In practice, of course, no absolute certainty about demand exists. Nevertheless, future demand is forecasted in a number of ways. This section examines the different sources of demand information, which thereby increase the predictability of demand.
• 3-Monthly annual schedules
In this so-called ABC-planning, the customers issue demand estimation broken down to individual products. A forecast is given of expected sales for the coming twelve months at the beginning of each quarter. The ABC-plan thus is a so-called rolling plan.
• Seasonal effects
The demand for glass products shows certain seasonal effects. Demand drops during the standard factory holidays, since production lines at the customer site are stopped or the number of shifts is reduced during these days. Decreased demand is experienced during the Easter holidays, the Whitsuntide (Dutch: Pinksteren), Ascension (Dutch: Hemelvaart), two weeks and three weekends during the summer holidays and during Christmas and New Years Eve. Apart from the seasonal effect due to factory holidays, a general seasonal effect can be distinguished for most products. An analysis was done on monthly aggregated data of shipments. To assess seasonal effects, a seasonality index was used and defined as the ratio of the actual demand in the corresponding period (month) of the last cycle (year) to the average demand of the cycle (Greene et al. 1970, 8-13). The seasonality indices and cumulative demand in pieces of the both the TLd36i tube and the aggregated demand data of TLd-i are presented as an example. The bars in the first chart illustrate the monthly seasonal indices of TLd36i in 2002 and 2003 and the lines present these numbers in a cumulative fashion. The second chart states the same information summarised for all TLd-i tubes.
Quantitative data on the seasonal effect of all products, including the average and standard
deviation of the seasonal indices of all product types can be found in Error! Reference source
not found..
S e a so n a l e ffe ct T L d 36 i
0 % 2 % 4 % 6 % 8 % 1 0 % 1 2 % 1 4 %
jan f eb mar a pr may ju ne july a u g s e p oc t no v d ec
M o n th Seasonality indices
0 1 0 .0 00 2 0 .0 00 3 0 .0 00 4 0 .0 00 5 0 .0 00 6 0 .0 00
Thousands
2 00 2 2 00 3 2 00 2 2 00 3
Figure 3-4 Seasonal effect TLd36i
TLdi
0,0%
2,0%
4,0%
6,0%
8,0%
10,0%
12,0%
14,0%
1 2 3 4 5 6 7 8 9 10 11 12 month
seasonal index
0,0%
20,0%
40,0%
60,0%
80,0%
100,0%
120,0%
2002 2003 2003 2002
Figure 3-5 Seasonal effect TLd-i
• Weekly schedules
The next level of demand information is the customer production planning, which is transformed into glass tubes requirements (see Figure 3-2). The customer normally finalises his schedule for the coming week (Saturday – Friday) on Wednesday leading to demand visibility of nine days. However, the customer production planning is changed occasionally during the first days of the planning horizon, possibly leading to rush orders in the glass factory.
• Informal circuit
Apart from the above-mentioned formal ways of gathering information on demand, the planner
regularly discusses with customer production planning. The planner collects information on, for
instance, upcoming boosts of demand for certain products or expected downtime at the
customer plant. However, collecting information through the informal channels is very time-
consuming and is performed in an unstructured manner.
• Production cycles at the customer (PU TL)
The coating of the glass tubes is a very capacity intensive process and its changeovers are highly sequence dependent. To reduce capacity loss resulting from set-ups, the coating operation is performed from light to dark colours. Hence, production at the customer site is scheduled in fixed sequences dependent on the colour of the coating. The colours are not relevant to the glass factory since only blank tubes are produced, but the sequence is also dependent on tube length. The planning cycles therefore increase the predictability of demand.
The first planning cycle is the HOR 6000/6 production wheel.
41
2/7 % 14
2/7 %
41
2/7 %
41
2/7 %
41
2/7 %
41
2/7 %
41
2/7 %
TLd 36i TLd 58i
TLd 36i
TLd 58i
TLd 58i
TLd 36i
W eek 1 W eek 2
W eek 3 TLd 30i
Figure 3-6 Planning cycle HOR 6000/6
Figure 3-6 shows that both TLd 36i and TLd 58i are produced on a weekly basis, but the TLd 30i tube is produced only once every three weeks. Production frequencies are one week and three weeks for respectively TLd 36i - TLd 58i and TLd30i. The total cycle time is therefore three weeks and all three mini-cycles last one week each. The total proportion of each product within the cycle is unknown, but all cycle-times (including the mini-cycles) are kept at one week.
The second planning cycle is the HOR 6000/1 production wheel (Figure 3-7); this cycle only consists of two products, which are produced alternately. Again, the portions within the product cycles may vary depending on customer orders, but the total cycle time is fixed on one week.
Production frequencies of both TLd 36i and TLd18i are thus kept at one week.
% 35
% 65 TLd 36i
TLd 18i
The final planning cycle is the one used for the HOR3000/1 (Figure 3-8). This cycle consists of four weeks in total, but within the cycle four min-cycles can be distinguished. Total cycle time is fixed at four weeks and each of the mini cycle times is fixed at one week.
TLd36i TLd70i TLd18i
TLd30i
TLd70i TLd58i
TLd30i
TLd36i
TLd23i
TLd36i
TLd38i
W eek 1
W eek 2
W eek 3 W eek 4
Figure 3-8 Planning cycle HOR 3000/1
However, some tubes are produced on multiple lamp lines and the three customer planning cycles need to be combined in order to get a full outline of the resulting glass tube cycle. Since the cycle times of the HOR 3000/1 and HOR 6000/6 respectively are four weeks and three weeks, the glass cycle time equals twelve weeks.
Week HOR 6000/6 HOR 6000/1 HOR 3000/1
1 36 58 36 18 70 18 2 36 58 36 18 30 36 3 36 58 30 36 18 70 36 38 23
4 36 58 36 18 30 36 58 5 36 58 36 18 70 18 6 36 58 30 36 18 30 36 7 36 58 36 18 70 36 38 23
8 36 58 36 18 30 36 58 9 36 58 30 36 18 70 18 10 36 58 36 18 30 36 11 36 58 36 18 70 36 38 23
12 36 58 30 36 18 30 36 58 Table 3-2 Glass cycle
