Cycle time control Developing a monitoring and control tool for the material delivery process of SCA Hoogezand

(1)

Cycle time control

Developing a monitoring and control tool for the material delivery process of

SCA Hoogezand

University of Groningen

Faculty of Economics and Business

(2)

I Preface

This thesis is my final project as student of the University of Groningen. I started my student career with the BSc Technology Management, as the operations management part grabbed my attention. After finishing my propaedeutic year I switched to the BSc Business Administration and passed the specialization course of Operations & Supply Chains, returning to my initial field of interest. By finishing the MSc Operations & Supply Chains my time as student ends and working life will start. Writing my thesis at SCA Hoogezand was a very useful experience. In addition to performing research in a real life environment, it is possible to gain some experience in working for a company. I experienced SCA Hoogezand as a very open and informal company with friendly and approachable employees, who made me feel welcome.

During the process of writing my thesis there are several persons who I would like to thank. First I would like to thank my supervisors at the University of Groningen: Erik Soepenberg and Tudor Bodea. By giving critical feedback and having good conversations (sometimes up to three hours), Erik helped me to significantly improve my thesis and enlarge the added value for SCA Hoogezand. In addition, my supervisors at SCA Hoogezand were very important for me in realizing this result: Siebe de Meer and Hans Top helped me a lot with the initial struggle of formulating a proper research question, setting the right direction of my thesis and always willing to answer all my questions. I also want to thank Peter Wentink for helping me with filling in the last details of my thesis. Besides, I always had a lot fun at the office with my ‘roommates’ Peter and Hans.

Finally, a special word of thank to my mother for always supporting my decisions during my six years of studying at University of Groningen. She gave me the opportunity to study a bit longer so I could develop myself by performing several extracurricular activities.

I would also like to thank my friends Jaap and Brenda for giving critical comments on my research and reviewing my English.

(3)

II Abstract

The goal of this study is to improve the delivery performance of the material delivery process of SCA Hoogezand, by developing a monitoring and control tool. By means of a pre-study the most important causes of deviations in delivery performance were diagnosed. In general, the lack of a regular flow throughout the process is the most important reason why delivery performance is not always met. The irregular flow is caused by variability in the number and pattern of arriving orders, variability in processing times and variability in the amount of available capacity. Since the process is operating under constantly changing circumstances, a dynamic solution is required to improve the delivery performance. By developing a monitoring and control tool, process operators will gain insight in the real time performance of the process and will be alerted when taking action is necessary. The tool is developed by first describing the functional, user and contextual requirements and subsequently formulating the content of the tool. The content (parameters) of the tool is derived from the most important variability causes influencing delivery performance. For each parameter, a norm value is calculated indicating when on time delivery is endangered. Whenever action is necessary is indicated by crossing a norm value, implying an out of control situation. For each out of control situation one or more possible actions are described. In addition, a most optimal way of working is described to prevent crossing the norm values.

(4)

1. Introduction

In this chapter, the background of SCA is described. In addition, a description of the Top Gear project is given.

1.1 SCA background

“In 1849 Mölnlycke Väfverie AB, was founded in Råda, near Göteborg in Sweden. The German G.F. Hennig opened a small factory with a weaving machine and a spinning mill. After some time the railway which passed the factory was called Mölnlycke and very soon the whole village built around the factory was called Mölnlycke.

The plant in Hoogezand was part of Mölnlycke until 1975, when SCA merged with the company. SCA is a part of the Swedish SCA Group (Svenska Cellulosa Aktiebolaget). In Europe this is one of the market leaders in manufacturing hygiene products like diapers, tissues, and incontinence products. In 1996 the SCA Group decided to add the name SCA to the name of all the factories for commercial reasons, to show that they all were a part of the SCA Group. For the plant in Hoogezand this meant the name was changed to SCA Mölnlycke.

In 1998 SCA Mölnlycke merged with SCA Hygiene Paper and became SCA Hygiene Products. In 2003 the unit Hygiene Products was dissolved and the Unit Personal Care was established. At this moment SCA/UcM is still part of that business unit. Within SCA, besides the business unit Personal Care, a Global Hygiene, Forest Products, and Packaging unit exists” (Osinga, 2010: 7). Currently, about 37,000 people are working for SCA in 60 different countries.

Production is conducted at 29 facilities in 24 countries. Products are sold in more than 100 countries throughout the world. In 1963 the factory in Hoogezand was opened, the first factory for Mölnlycke in the Netherlands. At SCA Hoogezand, diapers and incontinence products are produced.

1.2 Top Gear project

The process of delivering raw materials from warehouse to machine has been completely re-engineered in 2010/2011. This transformation is part of the Top Gear project, which was initiated in 2008 because SCA Hoogezand needed to cut costs.

Goal of the Top Gear project: design and implement an optimal logistical system for treating the raw materials, from the suppliers’ delivery to the raw material warehouse, to the delivery of the materials at the manufacturing department.

Several sub-projects were launched, where this research focuses on one of these: optimization of the internal transport flow between the Raw Material Warehouse (RMW) and production facility.

(8)

2. Glossary

Osinga (2010: 5) made a glossary, explaining the most important terms used in this research. I will use and expand his list (see Table 2.1).

Term Explanation

Abramskade 4 The address of the central warehouse for the raw materials.

Abramskade 6 The address of the plant on which the manufacturing machines are located.

AGV Automated Guided Vehicle, a vehicle which receives orders from the Warehouse Management System (WMS). It navigates itself by constantly checking its position at reflectors on the wall.

Conveyor The transport bridge which connects Abramskade 4 and 6. On this bridge the materials are transported from warehouse to factory.

RMW Raw Material Warehouse, the warehouse where all raw materials are stored.

Map-XR The WMS which is used by the machine operators to order the materials. Besides, the conveyor loaders process the orders with the help of the system and the AGVs receive their orders from the WMS.

WR-assignment Warehouse Rearrange-assignment, the material order from the machine which is processed by a fork truck driver at the warehouse; the driver loads the order on the conveyor.

AP-assignment Apollo-assignment, the material order which is transported on the conveyor. Apollo is the supplier of the conveyor.

AG-assignment AGV-assignment, the material order which is transported from the conveyor to the machine by an AGV.

AS-assignment The collection of machine waste by the AGVs (waste assignment). AR-assignment The collection of leftover raw materials by the AGVs (return assignment).

Pick location The location where the AGV picks the material. Either at the last position of the conveyor (for normal assignments), or at a location near the machine (for waste/ return assignments).

Drop location The location where the AGV drops the material near the machine. Every machine has locations for the materials used at that machine and almost every material has a separate drop location. Only a few materials are sharing one position.

Blocking The time where vehicles are unable to move due to other vehicles (Beamon, 2010: 384). Deadlock Two or more AGVs are blocking each other so no movement is possible.

Run time The time it takes to consume one piece of raw material.

Cycle time The time an individual job takes to traverse a routing (Hopp & Spearman, 2008: 331). Lead time1 Maximum allowable cycle time for a job (Hopp & Spearman, 2008: 331).

Table 2.1: Glossary.

1

(9)

3. Description of the process

In this section the process being studied will be described; the material delivery process. Machines need raw materials for production, which need to be transported from the Raw Material Warehouse (RMW) to the right location. The process can be divided in several steps, depicted in Figure 3.1.

WR-assignment AP-assignment AG-assignment

Total cycle time Figure 3.1: Material delivery process.

Step 1: Ordering material

The process always starts by ordering raw materials by the machine-operator. Ordering is triggered by a Kanban process (pull system): each machine has a use location and a buffer location for each material. As soon as a machine runs out of a particular raw material at its use location, the buffer is used. The usage of the buffer material should trigger a machine-operator to order new raw materials to replenish the buffer. The order is entered into Map-XR, the warehouse management system. Step 2: Loading material on the conveyor

As soon as the order is entered into Map-XR, the two fork truck drivers – responsible for loading the conveyor in the raw material warehouse - see a new order on their screen. Each unfulfilled order is displayed according to the oldest at the top of their screen and the newest at the bottom. The fork truck drivers use a first in first out (FIFO) strategy for picking the orders. Orders are picked and loaded on the conveyor. The time between ordering and loading the material on the conveyor is the WR-assignment time.

Step 3: Picking material by the AGVs

At the end of the conveyor, the orders are picked by the AGVs. As soon as a pallet or reel is located on a pick location, an idle AGV is triggered to pick the raw material. The time needed to transport the material from the first to the last position of the conveyor is the AP-assignment time. To get a better understanding of the conveyor, I will describe it in some more detail.

To bridge the distance between Abramskade 4 (RMW) and Abramskade 6 (production facility) two parallel conveyors are used. One conveyor is used for the delivery flow, while the other one is used for the return flow. The conveyors have several sections, where they each have room for one raw material unit. The conveyor is displayed in Figure 3.2.

As depicted in Figure 3.2, both conveyors have 17 parallel sections. After these sections, the pallets and reels are distributed to one of the four zones. Reels are moved to zone 11 or 14 and pallets to zone 12 or 13. De zones have unequal lengths, causing variability in cycle times. Also, on average, pallets have a shorter conveyor cycle time compared to reels. The reason is that reels have to be tilted before AGVs can transport them. In addition, reels require more AGV handling time. The last positions of zones 11, 12, 13 and 14 are pick locations for the AGVs; indicated by AGV01 – AGV04. Step 4: Delivery of the material

After the AGV picks the material, it drives to the right machine to drop its load. The time needed to drive from the conveyor to the machine is the AG-assignment time. After dropping, the AGV is

(10)

available to receive new orders. The routings AGVs travel are fixed and they adjust their speed to the specific environment (obstacles, other AGVs, etc.).

Total process

Summarizing, the process starts with a material-request by a machine and ends with the delivery of the requested material at that particular machine. The sum of the WR/AP/AG-assignment times gives the cycle time of the entire process. This operation is the process of main interest; however, two additional processes influence this operation.

Waste (AS-assignment)

Besides finished goods, machines also produce waste. Waste can have several forms, for example: finished products not satisfying the quality norms, wrappings and raw material left overs (e.g. left over after cutting paper).

Waste has to be collected by AGVs and moved to a specific location in the production facility. Waste can be moved with or without a specific waste cage. In case a cage is used, the AGV first collects a full cage at the machine. After dropping this cage at the central waste location near the conveyor, the same AGV picks an empty cage which is brought back to the machine.

Returning unused materials (AR-assignment)

When a machine switches to producing the next order, there can be some leftovers of order specific materials. In that case, the materials should be returned to the RMW. AGVs are responsible for collecting these returns and dropping them at a central location near the conveyor. Then, a fork truck driver collects these materials and loads them on the conveyor moving the materials back to Abramskade 4. Other reasons for returning raw materials are: not satisfying the quality standards or materials ordered too late; which means that the raw materials cannot be used since the machine is already producing the next order. Sections A-21 until A-25 (zone 14) depict the first steps of the return flow and via section 27 the raw material is returned to Abramskade 4 on one of the parallel conveyors (see Figure 3.2).

(11)

Figure 3.2: Schematic representation of the conveyor. 18-B 19-B 20-B 27 moving 24 26A 26B 17-B 16-B 15 elevator

(12)

4. Problem statement

In this section I will describe the initial management question, followed by the research objective and research questions. Finally, a general outline of the structure of this research is presented.

As described in the introduction, the process of delivering raw materials from warehouse to machine has been completely re-engineered in 2010/2011. To prevent material shortages, the delivery of those raw materials must occur within a certain timeframe. Material shortages result in machine stoppage implying additional costs. Based on calculations covering market demand, salaries, available capacity, etc. a penalty of € 400 per lost machine hour is determined. It is clear that machine stoppage is very undesirable and should be prevented.

Since the machine efficiency2 target for 2015 is 87%, while the current efficiency is 70%, the utilization of the machines increases in the future. This means raw materials should be delivered more frequently (more raw materials are consumed per unit of time) and faster (faster depletion of the material buffer).

4.1 Management question

Based on the efficiency improvements – implying more stern targets regarding delivery lead time – the inbound logistics manager is under the impression that the current performance of the material delivery process is not sufficient to meet future demands. He is unsure about what the target lead time should be and if the process is capable of achieving this performance. In addition, can the process handle the increase in demand? Finally, in case the current delivery performance is insufficient; how to achieve the required performance?

This research aims at determining a solution to ensure the process is capable of achieving future needs. Because the problem should first be diagnosed, a study is performed. The goal of the pre-study is to determine the need and the direction of a solution.

Firstly, how the process should perform in 2015 will be clarified; based on the run times of the materials – the time it takes before one unit of raw material is depleted – a target lead time is determined. Additionally, based on expected market demand, a required capacity is calculated and checked with the available capacity. Next, it should be checked if the current process is capable of meeting the target delivery lead time. Subsequently the current performance of the process is shown. Moreover, an investigation should be initiated as to why current delivery performance differs from the capable performance. Finally, a solution should be discussed to ensure target performance will be met in the future. This can be translated in the following five pre-study research questions:

1. What should the delivery lead time be in 2015?

2. Is the current process capable of achieving the desired lead time and increased demand? 3. What is the current delivery performance of the material delivery process with respect to the

2015 target lead time?

4. Which factors cause the actual performance to differ from the capable delivery performance?

5. What is the desired solution to achieve the target delivery performance?

The pre-study (see Appendix A) resulted in the following answers to the questions stated above: in 2015, ~99% of all the materials should be delivered within 60 minutes. Implying that the delivery lead time should be 60 minutes with an on time delivery performance of 99%. Based on available capacity and current cycle times, the process is capable of achieving this performance. The average cycle time is currently 31:37 minutes and based on expected required demand, the average bottleneck

2

(13)

utilization would be ~58%. However, at the moment only 90.27% of all the orders are delivered within 60 minutes indicating an insufficient delivery performance. This implies that a solution is necessary to improve the delivery performance, answering part one of the pre-study goal. The next step is determining the direction of the solution.

A number of variability factors are the reason why the delivery target is not always met. All causes of deviations in delivery performance can either be characterized as interarrival time variability, processing time variability or variability in available capacity. In Figure 4.1 a diagnostic model is presented based on Hopp & Spearman (2008) and Anupindi et al. (2006), indicating all factors influencing on time delivery. By dividing the cycle time in processing time and waiting time, then calculating both of them for the WR, AP and AG assignments, the most important causes of deviations in on time delivery were found (marked grey). A large part of the cycle time is waiting (47.55% of total cycle time) and mainly caused by “Waiting for fork truck” and “Waiting for AGV”. Regarding processing time, the malfunction and blocking time in the AG process have the most important impact (24.82% of total processing time). In Table 4.1 the most important causes are indicated and linked to one of the variability factors.

Influencing factor Type of variability

AGV malfunction time (E9/ G10) Processing time variability/ variability in available capacity Blocking time (E10/ G9) Processing time variability/ variability in available capacity Arrival pattern (F2/ G8) Interarrival time variability

Destination mix of loading (F3/ H3) Processing time variability/ variability in available capacity Number of operational machines (G1) Interarrival time variability

Ordering behavior (G2) Interarrival time variability Conveyor loader breaks (G4) Variability in available capacity Loading pattern (G5) Interarrival time variability Reels/ pallets mix of loading (G7) Variability in available capacity Waste/ return flow (G11) Variability in available capacity Table 4.1: Linking influencing factors and variability types.

(14)

On time delivery Target lead time Cycle time A B1 Z Processing time B2 Y Waiting time Picking time Conveyor time AGV time Waiting for AGV Waiting for room conveyor Waiting for fork truck Distance Speed Distance Speed Setup time Distance Speed Destination mix of loading AGV malfunction time Blocking time Queue Queue Availability of drop location Orders p/h > capacity p/h Conveyor malfunction time Conveyor strategy Run time Machine efficiency Arrival pattern Loadings p/h > capacity p/h Arrivals p/h > capacity p/h Queue Number of operational machines Ordering behavior Loading pattern Conveyor malfunction time Reels/ pallets mix of loading AGV flexibility Waste/ return flow AGV malfunction time Blocking time Arrival pattern Destination mix of loading C1 D1 D2 D3 D6 E1 E2 E3 E4 E5 E6 E7 E8 E10 E11 E9 E12 F1 F2 F3 F4.1 F4.2 F5.1 F5.2 F6.1 F6.2 G1 G2 G3 Conveyor loader breaks G4 G8 G9 G10 G11 G12 H3 Picking time H1 Conveyor time H2 E13 E14 D4 D5 G5 G6 G7 C2 Fork truck flexibility AGV time H4

(15)

4.2 Research objective and research questions

While looking for a dynamic solution I have come to the conclusion that some kind of measurement tool should be developed to give real time insight in the current situation and to provide control instruments to optimize the current way of working. This also gives the possibility to work more pro-actively: being able to see potential problems before they actually occur. This results in the following research objective.

Research objective: Improve delivery performance by developing a monitoring and control tool for the material delivery process.

Improving delivery performance means being able to deliver all orders within 60 minutes. The current average cycle time is satisfactory; however, the standard deviation of the cycle times should be reduced to guarantee a 60 minutes lead time.

In order to reach the goal of this study, some vital questions must be answered:

1. Which requirements should the tool satisfy and which assumptions should be made? 2. What is the structure and design of the tool?

3. How should the tool be used in the operation?

Ad 1) Before the tool can actually be designed, the requirements and assumptions should first be defined. Both can be divided into functional, user and contextual requirements or assumptions. This is necessary to find out what the tool should and should not do, and be able to evaluate the tool in the end. Does it meet the requirements and assumptions specified before (Verschuren & Hartog, 2005)? This first design stage can be characterized as determining the function of the tool. During this stage it should be clarified how the tool should perform to achieve the goal of the research (Siers, 2004).

Ad 2) After specifying the requirements and assumptions (= function), the structure of the tool can be developed. Which characteristics, aspects and parts should be included in the tool, to satisfy the whole set of requirements and assumptions (Verschuren & Hartog, 2005)? Thus, how is the function determined in the first stage accomplished? This results in the actual tool design.

Ad 3) After the tool is designed, it should be explained how it is used in the operation (on a day-to-day basis). How can data being presented by the tool be used to control the process, thus making the right decisions. The right action(s), at the right time for the right situation should be defined.

(16)

5. Methodology tool design

Just like the pre-study, the design of the tool also requires the collection of data. The most common data gathering methods are:

1. Archiving data (literature) 2. Observation

3. Interviews 4. Questionnaire

5. Quantitative data (company data)

By using multiple sources of data, data triangulation is made possible. This provides multiple measures of the same phenomenon, hereby safeguarding construct validity (Yin, 2003).

5.1 Data collection

Since each question requires different data collection methods, I will discuss them separately. Which requirements should the tool satisfy and which assumptions should be made?

• Interviews: to determine the requirements of the tool, open interviews are used to find out the requirements of the different users. All employees involved in the process are interviewed: direct users of the tool (process operators), conveyor loaders and managers3. • Archiving data: literature describes several guidelines which should be taken into account by

designing a tool. These guidelines are used to formulate requirements suitable for this research.

What is the structure and design of the tool?

• Quantitative data: to determine appropriate norm values (when is taking action necessary?) the pre-study data files are used again. This includes all order data of February 2012 (e.g. cycle times and interarrival times).

• Archiving data: literature describes several graphical methods to display statistics and control variability. These methods are input for design choices of the tool.

• Interviews: to check the suitability of several performance indicators and norm values; open interviews with direct users and managers are held.

How should the tool be used in the operation?

• Interviews: to get an overview of possible control possibilities, open interviews are held with direct users and managers. Based on knowledge of the process and experience, the employees can indicate which actions are possible to get the process back on track.

• Observations: by observing and analyzing the process it is possible to identify certain actions in case of out of control situations. Observing the process shows the current way of working and possible alternatives for critical situations. Observations also show the limitations of the process; as in which actions are not possible.

5.2 Data analysis

Excel 2010 is used to display data in graphs and statistical software package Minitab 16 is used to depict an example of how parameters can be measured.

3

(17)

6. Tool requirements, assumptions and preconditions

In this chapter the first sub-question will be answered, meaning that the requirements, assumptions and pre-conditions are determined. The sub-question states:

“Which requirements should the tool satisfy, which assumptions should be made and which pre-conditions must be met?”

In order to deliver a tool which is useful for the organization and will actually be used, it is important to first determine the requirements of the tool. There are three types of requirements: the first consists of the functional requirements; the functions that the artifact (tool) should fulfill or perform once it is realized. The second consists of the user requirements; to be fulfilled on behalf of the future users of the artifact. The third type consists of contextual requirements; set by political, economical, juridical and social environment (Verschuren & Hartog, 2005: 735). After the requirements are settled, the actual tool design can be determined.

6.1 Requirements

Functional requirements are derived from the goal of the tool. Properly fulfilling the functional requirements should also fulfill the goal. To determine the functional requirements, I will first formulate the goal of the tool:

“Goal: give insight in the performance of the material delivery process, provide short-term control possibilities in case of performance deviations and indicate long-term improvement possibilities.” As indicated by the goal; the tool should include both a short-term and a long-term orientation. Short-term control is necessary to be able to quickly react in case of performance deviations, while long-term measurements should indicate where major performance improvements are possible. Requirements are formulated both related to short-term control and for long-term performance improvements.

Short-term control:

• Give real time insight in the current performance. Logically derived from the goal of the tool. Real time insight is necessary for a quick response in an out of control situation.

• Give insight in the cause(s) of an out of control situation. To give direction where action is necessary, the causes of a worsened performance should be displayed. The causes are derived from the diagnosis (pre-study).

• Operators should be able to influence parameters. Control is not possible when operators cannot influence a parameter. Only parameters which can be influenced should be included in the tool.

• Decision support: indicate when taking action is necessary and which action to take. Direct users indicate they wish to be notified when taking action is necessary. This also prevents taking action too early or too late. To achieve this, norm values should be calculated.

• Report function of notifications and actions taken during a shift. To gain insight in what happened during a shift and to see if the right action was taken; a report function is necessary. The managers are interested in this information. In addition, because differences exist between shifts/ users, a user identification option is necessary. Users should log in at the start of their shift and log out at the end of their shift.

(18)

Long-term performance improvements:

• Indicate the main areas for improvement. Managers are interested in information regarding long-term areas for improvement. Based on long-term measurements, it should become clear which sub-process will perform below average or have possibilities for major improvements.

“The other two types of requirements to be fulfilled in the process of designing regard the interface between the artifact (tool) to be designed and the world outside” (Verschuren & Hartog, 2005: 735). In this case, the users are the world outside; the employees who have to work with the tool. There are two groups of users for this tool: the process operators directly responsible for monitoring the entire material delivery process and solving breakdowns, and the managers responsible for making long-term process improvements regarding the material delivery process. Because it is important that the users can properly use the tool and that the tool satisfies their needs: requirements per user group are defined. This results in the following user requirements.

Shared requirements:

• Graphic user interface: conveniently arranged (Greenes et al., 1994). Users indicate a clear visual presentation of the content of the tool is important for easy usage. This makes it possible to quickly interpret the graphs and figures.

• A clear work instruction. A work instruction should be available, explaining the functions of the tool and how to use it during the operation. This prevents the tool from needing a lot of support by managers (because questions are asked by the direct users). Thus, both operators and managers require this work instruction.

Process operator requirements:

• Decision support: notification when taking action is necessary and by who. Users want to know when taking action is necessary and who should take this action. This automatically implies it is clear who is accountable for solving the problem.

• It should be possible to access the tool all over the plant. For a short response time, the tool should be accessible all over the plant. E.g. users can directly see where a new breakdown occurred, instead of first returning to a centrally located computer.

Manager requirements:

• In addition to the ‘dynamic’ functional requirement: it should be possible to adjust parameters by key users (without intervention of a programmer). Based on possible future changes of the process, managers want to be able to adjust key parameters quickly. When this is possible without intervention of a programmer, response time is limited and flexibility is increased.

Finally, contextual requirements should be satisfied. Are there any constraints set by the government, labor unions or laws? Applied to the tool to be developed:

• If it is needed to use any external developed software; a legal license should be purchased. This is set by law. Otherwise, SCA is risking juridical action.

• In case employee performance evaluation is linked to properly using the tool, it is important to take confidentiality in mind (Greenes et al., 1994). Set by the law and also SCA policy. • In addition to the second bullet point, only employees who actually need the information

(19)

6.2 Assumptions

In order to meet the requirements, some assumptions should be made regarding the functions, users and context. Not satisfying the assumptions would make a fruitful use of the tool (party) impossible (Verschuren & Hartog, 2005). This means the assumptions should be checked with the final design; this check will be performed in the next chapter.

Functional assumptions:

• It should be possible to collect data automatically. Automatic data collection saves a lot of time and prevents faulty data.

• Reliable data: accessed data represent their actual true current state (Greenes et al., 1994). Because actions are based on displayed data, it needs to be correct. Faulty data results in wrong decisions.

• Availability: 99.9% uptime of the tool. Because actions are based on the information the tool provides and quick action is needed, the tool should (nearly) always be up and running. Crashes, lost connection with the server, etc. makes proper use of the tool impossible. User assumptions:

• Users should have basic understanding of English. Since the tool includes some English terms, users should be able to understand it.

• Users should have time available to use the tool. Users should be able to access the tool at least every 5 minutes. This prevents out of control situations being unnoticed and actions not executed on time.

Contextual assumptions:

• Enough database capacity to store all data for one year. Because data will be used to analyze the process and make long-term improvements, data should be collected for a long period of time. This requires database capacity.

(20)

7. Tool structure

In this chapter the second sub-question will be answered, meaning the requirements, assumptions and pre-conditions are used to formulate an actual tool design. The sub-question states:

“What is the structure and design of the tool?” With the following description:

“After specifying the requirements and assumptions (= function), the structure of the tool can be developed. Which characteristics, aspects and parts should be included in the tool, to satisfy the whole set of requirements and assumptions (Verschuren & Hartog, 2005)? Therefore, how to accomplish the function determined in the first stage? This results in the actual tool design.”

I will use the following outline to translate the requirements into an actual design: first, the short-term and long-short-term requirements are translated in more detailed instruments for operational use. Subsequently, for each instrument a visual representation is proposed. Finally, each instrument is checked with the earlier determined assumptions. This is done to check the feasibility of the proposed design. In each step the user requirements are also taken into account.

For clarity, in Figure 7.1 the complete design structure is depicted:

Figure 7.1: From goal to design.

7.1 Short-term performance measurement

As indicated before, the short-term parameters particularly concern the process operators. The operators should monitor a number of parameters and take action in case of performance deviations. The functional requirements are taken into account while developing the parameters. As indicated before, the causes of performance deviations should be included in the tool. The most important causes – as found in the pre-study – are depicted in Figure 4.1 and are input for the design. However, as formulated in the requirements it is required that operators can influence a parameter. Because the “Number of operational machines” (G1) and the “Waste/ return flow” (G11) cannot be influenced by the operators, these causes should be excluded from the design.

The selected parameters are marked in Figure 7.2. The dark grey parameters can be measured in an easy way. The light grey parameters cannot be measured individually, but it is possible to introduce a combined parameter measuring them together. The arrival pattern (F2 = G8) is not measured, since it is almost fully determined by the loading pattern (G5) and has little added value (refer to pre-study). Finally, all these parameters ultimately lead to a certain cycle time (via processing time and waiting time). Therefore, the cycle times for all the sub-processes (WR, AP and AG) are also measured. Since the waiting time and processing time parameters cannot be measured individually for all sub-processes, some measures are combined:

WR-assignment time = “Waiting for fork truck” + “Picking time” + “Waiting for room conveyor” 4 AP-assignment time = “Waiting for AGV” + “Conveyor time” 5

4

Only data regarding the time between ordering and loading the conveyor can be recorded, making it impossible to make a distinction between the processing time and waiting times.

Functional requirements

Tool goal Visual representation

(21)

AG-assignment time = “AGV time”

Cycle time = WR assignment time + AP assignment time + AG assignment time

Since it is not clear if the process is performing well without a target value, each parameter also has a norm value. Exceeding the norm triggers an out of control situation, leading to a certain action (see next chapter). Besides a norm value, an optimal value is presented as well. Performing close to this value indicates a minimal cycle time. The parameters, norm values and optimal values are based on the diagnosis presented in the pre-study. All parameters relating to the causes and performance measures will be explained in the next sections. The WR, AP and AG division is used again.

7.1.1 Cycle time targets

Before parameters and norms can be formulated for all different processes it should first be discussed how the individual processes should perform to meet the target of delivering within 60 minutes. As seen in Table 7.1, the average cycle times of the WR, AP and AG assignments are very close to 10 minutes. Since the total lead time target is 60 minutes, an equal distribution across the assignments seems suitable: 20 minutes lead time target per assignment (60 minutes divided by 3 sub-processes). This implies that each order may stay in a sub-process for 20 minutes at most. Exceeding the 20 minute norm triggers an out of control situation.

Assignment type Minimum cycle time (minutes) Average cycle time (minutes)

WR 0:37 11:56

AP 5:10 9:57

AG 2:49 9:44

Total 10:04 31:37

Table 7.1: Minimal and average cycle times material delivery process (repetition of Table A.5).

5

(22)

On time delivery Target lead time Cycle time A B1 Z Processing time B2 Y Waiting time Picking time Conveyor time AGV time Waiting for AGV Waiting for room conveyor Waiting for fork truck Distance Speed Distance Speed Setup time Distance Speed Destination mix of loading AGV malfunction time Blocking time Queue Queue Availability of drop location Orders p/h > capacity p/h Conveyor malfunction time Conveyor strategy Run time Machine efficiency Arrival pattern Loadings p/h > capacity p/h Arrivals p/h > capacity p/h Queue Number of operational machines Ordering behavior Loading pattern Conveyor malfunction time Reels/ pallets mix of loading AGV flexibility Waste/ return flow AGV malfunction time Blocking time Arrival pattern Destination mix of loading C1 D1 D2 D3 D6 E1 E2 E3 E4 E5 E6 E7 E8 E10 E11 E9 E12 F1 F2 F3 F4.1 F4.2 F5.1 F5.2 F6.1 F6.2 G1 G2 G3 Conveyor loader breaks G4 G8 G9 G10 G11 G12 H3 Picking time H1 Conveyor time H2 E13 E14 D4 D5 G5 G6 G7 C2 Fork truck flexibility AGV time H4

(23)

7.1.2 WR-assignment parameters and norm values

First, it should be reasoned how the WR parameters can be measured. In addition, optimal values (implying minimal cycle time) should be calculated and suitable norm values (indicating when on time delivery is endangered) should be determined. In the diagnosis two causes of performance deviations are identified – regarding the WR sub-process – which can be influenced by the involved employees: 1) the non-optimal ordering behavior of the machine operators (increasing short-term requested capacity) and 2) the disrupting breaks of the conveyor loaders (reducing short-term available capacity). Both causes have influence on the WR-assignment time. Measuring the causes could trigger a potential out of control situation in advance. In Table 7.2 the parameters are translated in measures with suitable norm values and optimal values. All parameters will be explained in some more detail.

Parameter How to measure Optimal value(s) Norm value(s)

Ordering behavior (G2) Actual number of orders per hour (sum of all orders placed during 1 hour).

1. Constant number of orders per hour (based on the amount of operational machines).

1. Upper norm value: expected amount of orders per hour divided by 70%. 2. Lower norm value:

(expected amount of orders per hour divided by 70%) * 40%.6

Conveyor loader breaks (G4) 1. Interarrival time of loading the conveyor (time between loading two separate orders). 2. Queue length of unfulfilled orders. 1. Interarrival time of loading equal to interarrival time of ordering. 2. Queue of 0 unfulfilled orders. 1. Maximum interarrival time in minutes = 60 minutes divided by (expected amount of orders per hour – 11).7 2. Maximum queue of 11 unfulfilled orders. WR-assignment time (D1 + D4 + D5)

Time between ordering and loading.

1. 1:30 minute. 2. 20 minutes. Table 7.2: WR measures and norm values.

7.1.2.1 Ordering behavior

The non-optimal ordering behavior can be measured by observing the number of orders placed per hour. Large differences between orders placed per hour indicate ordering does not occur at the right point in time. It is only relevant to compare this number within one shift, because the amount of operational machines differs from shift to shift (but not within a shift). The optimal value is an equal number of orders per hour during an entire shift. This value is dependent upon the amount of machines running during a shift. For all shifts in February 2012, the relation between the amount of operational machines and amount of orders is calculated (see Figure 7.3; 70% line). The expected amount of orders per hour = number of operational machines * 1.55 + 10.55. This relation is based

6

Example: amount of operational machines = 10, implying 26.05 orders per hour (= 10 * 1.55 + 10.55). The upper limit would be 37.21 (= 26.05 divided by 70%) and the lower limit would be 14.89 (= 26.05 divided by 70% * 40%). Crossing one of these norm values would trigger an out of control situation.

7

(24)

on a machine efficiency of 70% (on average 28.89 orders per hour) and will change as soon as the machine efficiency increases.

The norm values are based on the same formula and the possible short-term utilization of the machines; because this also determines the amount of orders. For ~3-4 successive hours, a machine can produce with 100% efficiency; indicating more orders are placed than expected beforehand. In addition, the machine can also produce with ~40% efficiency (to reach the 70% average). This means norm values are determined as 40% and 100% of the maximum amount of expected orders (40% and 100% lines in Figure 7.3). Crossing these norms implies the ordering behavior of the machine operators is not right. Important remark: crossing the upper norm is more problematic because this increases the short-term utilization. The reason for also setting a lower norm is the following: crossing the lower norm implies fewer orders are placed than expected. Because the amount of orders placed per hour will move towards the average, an order peak is expected after crossing the lower norm.

Since the 70% efficiency sample values do not cross the 40% and the 100% regression lines; it is plausible that, based on the determined norm values, correct decisions will be made.

Figure 7.3: Relation between the number of operational machines and the expected amount of orders per hour.

7.1.2.2 Conveyor loader breaks

The breaks of the conveyor loaders can be shown by looking at the time between two successive load actions and the queue length of unfulfilled orders. A long time between two load actions implies the conveyor loaders took a break. In addition, a queue will build up increasing waiting time. Therefore, a norm value for the maximum interarrival time is calculated and a maximum queue length is determined. Since the available capacity of the conveyor loaders exceed the maximum requested capacity; ideally each order is picked and loaded as soon as it is placed. This strategy

y = 2,21x + 15,07 y = 1,55x + 10,55 y = 0,89x + 6,03 0 10 20 30 40 50 60 70 0 2 4 6 8 10 12 14 16 18 20 O rd e rs p e r h o u r Number of machines

Fitted line plot

(25)

would imply an equal interarrival time of ordering and loading and a queue of 0 orders (optimal values). However, there are some limitations making this infeasible:

• Interarrival time variability: several orders can be placed within a very short time interval making quick delivery more difficult (requested capacity > available capacity).

• Unbalanced reels/ pallets mix: since one conveyor loaders is responsible for reels, while the other picks pallets an unbalanced reels/ pallets mix makes a quick delivery difficult.

• Conveyor loaders have the right to take a break: indicating fluctuating available capacity. Therefore, a norm should be determined in addition to the optimal value. Since the maximum WR-assignment waiting time is 18:30 minutes (20 minutes – 1:30 processing time) and the average required capacity will be 36.73 orders per hour, the maximum average allowed queue – as explained by Little’s Law8 – has 11 unfulfilled orders (36.73 orders per hour * 18:30 minutes). This means the first order is, on average, open for 17:58 minutes ((11 orders / 36.73 orders per hour) * 60 minutes) and can still be loaded on time. The minimum amount of orders to be fulfilled within one hour (A) can be calculated as: the expected amount of orders per hour subtracted by the maximum queue length. This can be translated in a maximum interarrival time (first norm value), by dividing 60 minutes by (A). An important remark: because this required working pace implies the queue will build up in time, conveyor loaders may always work faster. It is important the queue will not exceed 11 unfulfilled orders (second norm value).

The reason why a maximum interarrival time is determined instead of just a maximum queue of unfulfilled orders is important: in this case, conveyor loaders wait until the queue builds up (take a break) and then quickly load all orders. This implies an irregular flow, seriously disrupting the flow (as indicated in the pre-study diagnosis).

7.1.2.3 WR-assignment time

As indicated before, both causes have influence on the WR-assignment time. This performance measure has an optimal value of 1:30 minute (average processing time), implying 0 minutes waiting time. The norm value is 20 minutes, as explained in section 7.1.1. This includes the “Waiting for fork truck”, the “Picking time” and the “Waiting for room conveyor”.

7.1.3 AP-assignment parameters and norm values

Like the WR parameters, it should be reasoned how the AP parameters can be measured. In addition, optimal values should be calculated and suitable norm values should be determined. In the diagnosis two causes of performance deviations are identified – regarding the AP sub-process – which can be influenced by the involved employees: 1) the irregular pattern of loading the conveyor (increasing short-term requested capacity) and 2) the non-optimal reels/ pallets mix of loading the conveyor (reducing short-term available capacity). Both causes have influence on the AP-assignment time. Measuring these causes could trigger a potential out of control situation in advance. In Table 7.3 the parameters are translated in measures with suitable norm values and optimal values. All parameters will be explained in some more detail.

8

(26)

Parameter How to measure Optimal value(s) Norm value(s) Loading pattern (G5) Interarrival time of

loading the conveyor (time between loading two separate orders).

1. Constant number of orders per hour (based on the amount of operational machines). 1. Maximum interarrival time in minutes = 60 minutes divided by (expected amount of orders per hour – 11). 2. Maximum queue of 11 unfulfilled orders.

Reels/ pallets mix of loading (G7)

Reels/ pallets mix (number of successive reel/ pallet load actions).

1. A maximum of 5 successive reel or 6 successive pallet load actions9. 2. Out of the last 10

load actions a maximum of 8 may be a reel and 9 may be a pallet10. 1. A maximum of 10 successive reel or 22 successive pallet load actions. AP-assignment time (D2 + D6)

Time between loading and picking by the AGV.

1. 5:22 minutes. 1. 20 minutes. Table 7.3: AP measures and norm values.

7.1.3.1 Loading pattern

The loading pattern can again be measured by the interarrival time between two successive load actions. Since the “Waiting for room conveyor” is actually WR-assignment time, the same norm values can be used.

7.1.3.2 Reels/ pallets mix of loading

The reels/ pallets mix can be measured by the amount of successive reel or pallet load actions. Loading too many reels/ pallets within a short time interval hinders the flow to the four pick locations at the end of the conveyor (because of limited space). The first optimal value is based on the amount of reel/ pallet specific sections available at the conveyor. There are five reel specific sections, indicating a sixth successive reel could not flow to a pick location because of limited space. Besides, a pallet arriving after the sixth successive reel also has to wait because the reel is blocking the way. In addition, there are six pallet specific sections, indicating a seventh successive pallet could disrupt the flow as well. The second optimal value is an assumption, based on:

• Data analysis of the loading pattern: when does the mix increases cycle time?

• Workable solution: since at the moment a FIFO policy is used and simplicity should be kept in mind.

The norm value is based on the shortest possible time between two pallet or two reel pick actions by the AGVs. As indicated in Appendix E, AGVs can pick a reel every 1:39 minutes and a pallet every 0:44 minute. Since the minimal processing time for a reel is 5:30 minutes, the maximum allowable waiting time is 14:30 minutes (= 20 minutes – 5:30 minutes). This means, at most 10 successive reel load

9

Example: after 5 successive reels a pallet must be loaded first before a reel may be loaded again.

10

(27)

actions are allowed before crossing the maximum waiting time (= 14:30 minutes divided by 1:39 minutes + 2 11). The same calculations12 result in a maximum of 22 successive pallet load actions.

7.1.3.3 AP-assignment time

As indicated before, both causes have influence on the AP-assignment time. This performance measure has an optimal value of 5:22 minutes (average processing speed), implying 0 minutes waiting time. The norm value is 20 minutes, as explained in section 7.1.1. This includes the “Waiting for AGV and the “Conveyor time”.

7.1.4 AG-assignment performance measures and norm values

Like the WR and AP parameters, it should be reasoned how the AG parameters can be measured. In addition, the optimal values should be calculated and the suitable norm values should be determined. In the diagnosis three causes of performance deviations are identified – regarding the AG sub-process – which can be influenced by the involved employees: 1) the AGV malfunction time (both increasing processing time and reducing short-term available capacity), 2) the AGV blocking time (both increasing processing time and reducing short-term available capacity), and 3) the non-optimal destination mix of loading the conveyor (both increasing processing time and reducing short-term available capacity). The blocking time is also influenced by the destination mix and loading pattern (see section A.7.2.3). All causes have influence on the AP-assignment time. Measuring the causes could trigger a potential out of control situation in advance. In Table 7.4 the parameters are translated in measures with suitable norm values. All parameters will be explained in some more detail.

Parameter How to measure Optimal value Norm value

AGV malfunction time (E9 = G10)

Additional time to reach a drop location on top of the minimal AGV time. 1. Moving average AG-assignment time of 5:37 minutes (moving average of 14 orders). 1. Moving average AG-assignment time of 11:26 minutes (moving average of 14 orders). Blocking time (E10 = G9)

Destination mix of loading (F3 = H3)

Interarrival time of loading the conveyor (time between loading two separate orders regarding the same destination).

1. A minimum of 3:30 minutes between two load actions with the same destination.

1. A maximum of two orders with the same destination may be loaded per 7 minutes. AG-assignment time (D3)

Time between picking by the AGV and delivery at the machine.

1. 5:37 minutes. 1. 20 minutes.

Table 7.4: AG measures and norm values.

7.1.4.1 AGV malfunction and blocking time

The blocking time can be measured by the additional AGV time to reach a drop location above the minimal AGV time. However, the AGV malfunction time also increases the AGV time13. Because it is not possible to measure the individual contribution of both factors, the parameters are measured together and only one optimal and one norm value is determined. To determine the maximum

11

+ 2 since the first two orders can be picked by an AGV immediately (waiting time of 0 minutes).

12

Maximum waiting time = 14:50 (= 20 minutes maximum cycle time – 5:10 minimal processing time). Maximum successive pallet load actions = 22 (= 14:50 minutes divided by 0:44 minute + 2).

13

(28)

allowable time above the minimal transport time, each routing (conveyor to drop location) needs a different norm value. Because it is very time consuming to calculate this, a general measure will be used instead to calculate a single norm value. To be able to satisfy the required capacity each AGV should deliver 2.62 orders per hour (average required capacity of 36.73 orders per hour divided by 14 AGVs). Since the time needed to drive to the pick location roughly equals the time needed to return to the conveyor; the maximum allowable average AGV time for an order is 11:26 minutes (60 minutes divided by 2.62 orders per hour divided by 2). Because one AGV crossing the 11:26 minutes norm value does not directly endangers the required available capacity (since other AGVs possibly deliver faster than 11:26min), a moving average should be chosen. Because there are 14 AGVs, a moving average of the last 14 successive orders seems suitable. When the moving average of 14 successive orders crosses 11:26 minutes, the available capacity is too low to satisfy the requested capacity and “Waiting for AGV” (D6) will increase.

The optimal value is 0 minutes AGV malfunction/ blocking time, implying a weighted average AGV time of 5:37 is expected. A moving average of 14 is used again, to level out differences in distance and use the same measure.

7.1.4.2 Destination mix of loading

The destination mix can be measured by the interarrival time of loading the conveyor, regarding orders with a destination in the same area. Since the loading pattern is closely related with the arrival pattern for the AGVs, the interarrival time of loading the conveyor is a suitable measure. Because there are two bottleneck areas where blocking is most likely to occur, this measure will focus on these areas. The two bottleneck areas creating the most blocking time are the three Heracles and two Pegasus machines. These five machines account for 78.79% of all the orders that have an AG-assignment time of more than 20 minutes. This implies that 25% of all machines cause almost 80% of all late AG orders.

The optimal value would be an interarrival time – between loading two orders with the same destination – that results in 0 minutes blocking time. To reach this goal, it should first be explained how blocking time is created. In the following example, the Heracles area will be used (see Figure 7.4). The following rule holds for AGVs entering the Heracles area: the first AGV accessing the Heracles area can drive to its drop location, drop its load and finally drive back empty without any waiting time. The second AGV which needs to access the Heracles area has to wait at location “A” until the first AGV finished its drop and passed location “B”. The time needed to drive from “A” to the machine, drop its load and return to “B” takes on average 3:00 minutes. The same “A” – drop – “B” cycle holds for the Pegasus machines and takes 3:30 minutes. This means an interarrival time of 3:30 minutes is always sufficient to prevent any blocking time (optimal value).

In addition, also a norm value is determined. As explained in section 7.1.4.1, each AGV should drop its order within 11:26 minutes to prevent capacity shortages. In addition, the average minimum AGV time for the Heracles machines is 7:00 minutes and the average minimum AGV time for the Pegasus machines is 5:52 minutes. Based on this information, minimal cycle times can be calculated regarding AGVs arriving at the same time at Heracles or Pegasus (see Table 7.5). As can be seen in the table, the third successive AGV entering the area exceeds the maximum allowable cycle time. This means loading three orders with the same destination within a short time interval should be prevented. To ensure that the waiting time at “A”, and thus the AGV time, are not able to grow (as calculated in Table 7.6), a maximum of two orders with the same destination may be loaded within 7 minutes14. As

14

(29)

can be seen in Table 7.7, the 7 minutes interarrival time between order 1 and 3 ensures that the 2nd AGV has just passed “B”, while the 3rd AGV arrives at “A”. As soon as the “A” – drop – “B” cycle will take longer than 3:30 minutes, waiting time at “A” will increase.

Figure 7.4: Heracles machines and drop locations.

AGV number Cycle time Heracles (minutes) Cycle time Pegasus (minutes)

1 7:00 5:52

2 10:00 (= 7:00 + 3:00) 9:22 (= 5:52 + 3:30)

3 13:00 (= 7:00 + 3:00 + 3:00) 12:52 (= 5:52 + 3:30 + 3:30) Table 7.5: Cycles times bottleneck areas.

(30)

Order number Departure from pick location Arrival at “A” Departure from “A” Arrival at machine Arrival at “B” AGV time

1 0:00 min 4:07 min 4:07 min 5:52 min 7:37 min 5:52 min

2 0:00 min 4:07 min 7:37 min 9:22 min 11:07 min 9:22 min 3 6:00 min 10:07 min 11:07 min 12:52 min 14:37 min 6:52 min 4 6:00 min 10:07 min 14:37 min 16:22 min 18:07 min 10:22 min 5 11:00 min 16:07 min 18:07 min 19:52 min 21:37 min 8:52 min 6 11:00 min 16:07 min 21:37 min 23:22 min 25:07 min 12:22 min Table 7.6: AGV times of order deliveries at Pegasus machines.

Order number Departure from pick location Arrival at “A” Departure from “A” Arrival at machine Arrival at “B” AGV time

1 0:00 min 4:07 min 4:07 min 5:52 min 7:37 min 5:52 min

2 0:00 min 4:07 min 7:37 min 9:22 min 11:07 min 9:22 min 3 7:00 min 11:07 min 11:07 min 12:52 min 14:37 min 5:52 min 4 7:00 min 11:07 min 14:37 min 16:22 min 18:07 min 9:52 min 5 14:00 min 18:07 min 18:07 min 19:52 min 21:37 min 5:52 min 6 14:00 min 18:07 min 21:37 min 23:22 min 25:07 min 9:22 min Table 7.7: AGV times of order deliveries at Pegasus machines.

Explanation of Table 7.6 and 7.7:

• Departure pick location: the point in time the order is picked by the AGV.

• Arrival at “A”: the AGV needs 4:07 minutes to travel from the conveyor to “A” (5:52 minutes to reach the machine – (3:30 minutes divided by 2 15)).

• Departure from “A”: the AGV can depart at “A” as soon as the last AGV which entered the Pegasus/ Heracles area reached “B”.

• Arrival at machine16: 1:45 minutes after the AGV departs from “A” (3:30 minutes divided by 2).

• Arrival at “B”: 1:45 minutes after the AGV departs from the machine (3:30 minutes divided by 2).

• AGV time: the time between the moment the AGV picks the order and the moment the AGV arrives at the machine.

7.1.4.3 AG-assignment time

As indicated before, all causes have influence on the AG-assignment time. This performance measure has an optimal value of 5:37 minutes (weighted average minimal processing speed), implying 0 minutes additional (non-value added) processing time. The norm value is 20 minutes, as explained in section 7.1.1. This includes the “AGV time”.

7.1.5 Total process

Up till now, several performance indicators are formulated for the three sub-processes. However, the most important target is not measured: the cycle time of the entire process. Thus, the last parameter is the cycle time of the entire process with an optimal value of 12:29 minutes (1:30 WR + 5:22 AP + 5:37 AG processing times). The norm value is 60 minutes, as explained in section 7.1.1. Comparing the cycle time with the target lead time automatically results in an on time delivery percentage.

15

Assumption: time“A” – machine = timemachine – “B” = 1:45 minutes (3:30 minutes divide by 2). 16

(31)

7.2 Long-term process improvements

The long-term requirements relate to making process improvements: how to improve the current process. The long-term requirements particularly concern the managers.

Earlier in this chapter I mentioned the division between short-term and long-term. Up till now I covered short-term parameters, optimal values and norm values. However, it could be possible that a certain breakdown occurs significantly more often compared to other breakdowns. Or a certain machine has a very irregular order pattern disrupting the entire flow. These irregularities can only be discovered and analyzed after term data collection. Managers can use this data to make long-term improvements to optimize and stabilize the process. This can result in achieved performance closer to the optimal values, and eventually more tight optimal values. Therefore, I describe a number of performance measures which should indicate main areas for improvement in the long run. These measures are based on the short-term measures mentioned in the previous sections. Actions based on the measures are presented in the next chapter.

7.2.1 WR process improvements

1. Order pattern per machine when operational (orders per hour/ shift and per machine): improves “ordering behavior” (G2). To discover if a certain machine is disrupting the flow and ignoring the rules regarding order moments, the order pattern of all machines can be analyzed for a period of time. This could also indicate that a certain shift or day of the week differs from the average.

7.2.2 AP process improvements

1. Reels/ pallets mix (mix per hour/ shift and per machine): improves “reels/ pallets mix of loading” (G7). Since the norm values relating to the reels/ pallets mix are partly based on assumptions it is important to review these norms once in a while. This can be done by observing the mix and check if differences exist between shifts/ days/ machines. The loading mix can also be compared to the AP performance to see the influence of different loading strategies.

7.2.3 AG process improvements

1. Blocking time per routing (total amount of blocking time in a routing divided by total amount of times routing traveled): improves “destination mix of loading” (F3 = H3) and “blocking time” (E10 = G6). Indicates the (bottleneck) routings which frequently increase the AG-assignment time.

2. Breakdowns per routing (number of breakdowns in a routing divided by total amount of times routing traveled): improves “AGV malfunction time” (E9 = G10). Indicates the routings where breakdowns frequently occur. By carefully analyzing and observing this routing, the root causes of the breakdowns can be found.

3. Breakdowns per node17

(average number of breakdowns per week): improves “AGV malfunction time” (E9 = G10). Gives an even more detailed overview of where in the plant breakdowns frequently occur.

4. Breakdowns per AGV (average number of breakdowns per week): improves “AGV malfunction time” (E9 = G10). Indicates which AGVs are (almost) worn out/ malfunctioning and need to be repaired or replaced. Quick action prevents breakdown time.

5. Available AGV capacity (= total capacity – blocking time – breakdown time – used waste/ return flow capacity) versus performance: improves “waiting for AGV” (D6) and “waste/ return flow” (G11). Should give a better understanding of what the process is capable of: what is the minimal required available capacity to be able to meet the delivery target. Based on this number it can be seen how much capacity should be available for the delivery flow.

17

Cycle time control Developing a monitoring and control tool for the material delivery process of SCA Hoogezand