Production of the Product x
for
Company x Electric General Supplies
at
Company x
‘Improving production performance’
Groningen, Monday, December 01, 2008
Written by: Paul van de Ven
Supervised by: prof. dr. ir. J. Slomp (University of Groningen)
dr. J.A.C. Bockhorst (University of Groningen) H. Busschers (Company x)
Production of the Product x
for
Company x Electric General Supplies
at
Company x
‘Improving production performance’
Masters Technology Management Faculty Economics & Business
University of Groningen
Preface
Groningen, Thursday, 25 September 2008
This report is written for the Master thesis research of the course Technology Management at the University of Groningen. It is a report that I have conducted in order of Company x located in Hengelo, the Netherlands. The documentation of this research consist out of this report and appendices, which can be found at the end of this report (see content at p. IV)
The research and writing of this report took about 5 months, an internal project within Company x. This project has been the first assignment that I was responsible for the total outcome. What I liked the most at the final outcome was that it was very surprising to me. This means when I started this project I would never have expected to have this outcome. This makes me as a person a surprising experience richer. Not only the outcome made me an experience richer, but also dealing with different departments and persons made sure that my knowledge has gained, theoretical, but maybe more important practical.
I would like to thank the persons who have assisted me or helped me during my research. As first I would like to thank my superior Han Busschers, who always stood by, in good times and bad times. Without his backup this research could not have been applied in this form. At second I would like to thank Ernst van Raalte, he also always stood by as much as possible and made sure that I got the information I needed for my research. Also I would like to thank Wouter Weersink, another graduate student of my studies at Company x. His Microsoft Excel skills saved me a lot of time during my research. Of course I forget a lot of people at Company x who have helped me at this research, but of course I am also grateful to them. At last I would like to thank my supervisors at the University of Groningen prof. dr. ir. J. Slomp and dr. J.A.C. Bockhorst for their adequate, fast and valuable feed back to my report.
I hope that the results of this report will be applied in the practice.
Summary of the report
This research has been conducted at the department Company x Electric General Supplies (EEGS), which is a department of Company x located in Hengelo the Netherlands. At EEGS the production line ‘Product Market Combination 1’ (PMC1)is responsible for the production of the Product x frame which is supplied to Middle Voltage Systems (MVS) an internal customer. At this moment production management has its doubts if the production of the Product x frame reaches its limits and if this can be ascribed to a welding robot. This welding robot assembles metal sheet parts to each other and is seen as a bottleneck machine because the following two problems are present at the welding robot:
1. the production quantity of the welding robot; reaches already its limits for unknown reasons and maybe it cannot meet (growing) demand.
2. the quality of the Product x; the internal customer has lots of rework because the quality of the Product x frame cannot be assured at the robot welding process.
These problems make the following objective of this research: Increasing product quality and
production quantity at the production process of the Product x. To reach this objective insight
has to be found in the situation in which the welding robot process is embedded. This makes the following research question: Which factors are of significant importance to the product
quality and production quantity of the Product x and how can these factors be influenced within the framework of PMC 1?
To be able to analyse the production of the Product x production process has been divided in three sub processes. The first sub process has been from punching till stock 614, which is responsible for supplying the welding robot. The second sub process is from the welding robot to manual welding and is responsible for assembly of the metal parts into a Product x frame and a quality check. The third sub process represents the production process after manual welding till the internal customer MVS and is responsible for powder coating. The focus of this research has been on the first two sub processes.
means of a conveyor, based on pull production. From the welding robot an order goes to punching and following complete article sets are put into a conveyor. These are transported together through the production process and this results in equal lead times for all articles of a nested set. The conveyor also should trigger that after STOPA all product, the easier and harder ones to produce are taken out for production. In the case of the Product x this results in tremendous work in progress (WIP) reductions and lead time reductions, with as result that stock 614 can supply the welding robot in a more effective way. This way of kitting can also be applicable for other production processes which face the same kind of logistical problems. Sub process 2 also faced high (WIP) levels with long lead times as a consequence. Lacking planning targets caused less production output than input, which make WIP levels raise and the production floor ‘full’. This process goes on until a big ‘clean’ occurs’, and following the same cycles will be gone through. The lack of a system which determined the sequence of output makes that no usable feedback system can be applied in order to improve quality. The solution for these problems has been searched in the application of a FIFO line, based on a pull system. By applying pull with a FIFO line, one will have maximum WIP and no more input than output. The consequence is that the WIP levels and lead times of the Product x frames at the buffer can be lowered tremendously. Because products have to go ‘first in first out’(FIFO) through the process, the line gives the possibility for a usable feedback system Another factor which appeared to be of great importance on the logistical control has been the human factor. Operators at the second sub process have been frustrated which resulted in less commitment to the production performance. This is caused by lack of knowledge which made that they cannot always deal with unforeseen technical problems of the welding robot, which causes production delay. Also the fact that suggestions at management to improve the situation are not always heart, and a lacking reward system for the operators to improve performance, make that the operators can not always act in order of a optimal production performance. Lead times become longer, WIP levels rise and quality can be lacking. To regain some commitment of the operators they have been incorporated into the redesign of sub process 2. This appeared of great worth because the redesign of the FIFO-line principle is mainly based on their suggestions.
Content
Preface... I Summary of the report ... II Content ...IV
1 Introduction research ...1
1.1 Introduction...1
1.2 Company x Corporation ...1
1.3 Company x B.V...2
1.4 Company xElectric General Supplies (EEGS)...2
1.5 Application of Product x...2
1.6 Motivation and the assignment ...4
1.6.1 Research focus, restrictions and assumptions...5
1.7 Explanation production process: ...6
1.8 Restrictions & assumptions ...7
2 Theoretical background and conceptual model ...8
2.1 Introduction...8
2.2 Conceptual model...8
2.3 Sub questions: ...9
2.3.1 Logistical control and its relationship to the performance level ...9
2.3.2 Relationship between logistical control and process control...11
2.3.3 Process control and relationship at the performance level...11
2.3.4 Human factor and its influence on the logistical control ...13
2.4 Research methodology ...14
2.5 Research outline ...18
3 Diagnosis of production process of the Product x ...19
3.1 Introduction...19
3.2 Logistical control sub process 1; from punching to stock 614...19
3.2.1 Overall Equipment Efficiency (OEE)...19
3.2.2 WIP level at stock 614 and lead times...22
3.2.3 Utilization of capacity...28
3.2.4 Conclusion of logistical control at sub process 1 ...30
3.3 Logistical control at sub process 2; from the welding robot till manual welding ...30
3.3.1 Production output / order release...30
3.3.2 Work in progress (WIP) at sub process 1...31
3.3.3 No quality usable quality feedback system ...35
3.3.4 Conclusion of logistical control at sub process 2 ...35
3.4 Human factor ...37
3.4.2 Training and education at welding process ...39
3.4.3 Vision culture at welding process ...41
3.4.4 Conclusion human factor ...42
3.5 Summarizing conclusion about current state of sub process 1 & 2...43
4 Logistical design for the production process of the Product x ...45
4.1 Introduction...45
4.2 Design of sub process 1; from punching to stock 614...45
4.2.1 Nesting at punching ...45
4.2.2 Conveyor and kitting...47
4.2.3 Features of the transport device and the consequences for the transport costs ...51
4.2.4 Missing of parts because of failure. ...53
4.2.5 Setup + production cost for the Product x related to the batch size and conveyor ...53
4.2.6 Consequences for stock 614 ...54
4.2.7 Handling as a consequence of the conveyor...58
4.2.8 STOPA and relation to punching...58
4.2.9 Conclusion of logistical control at sub process 1 ...59
4.3 Design for sub process 2; from the welding robot to manual welding ...60
4.3.1 Reducing WIP at sub process 2 with FIFO-line ...60
4.3.2 Planning to manual welding...65
4.3.3 Conclusion of logistical control at sub process 2 ...66
4.4 Summarizing conclusion about future state of sub process 1 & 2 ...66
5 Conclusions & recommendations ...69
5.1 Conclusion ...69
5.2 Recommendations ...70
References ...72
Appendix I: Nesting configuration ...74
Appendix II: Forms which had to be filled in by the operators ...76
Appendix III: Counts of stock 614 and its consequences ...78
Appendix IV: different routings to stock 614 of the articles in a nest for the 3f (p. 81-86) and 4f (p. 86-90). ...81
Appendix V: waste calculations ...91
V.I Fixed article nesting ...91
1 Introduction research
1.1 IntroductionThis chapter includes the introduction of the research. First paragraph 1.2 to 1.5 will explain the problem context top down within Company x Corporation, this means from the corporation/strategic level to finally the production line/operational level. From this the problem context, the research questions and the objective of this research are formulated, this is done in paragraph 1.6. The last paragraph explains the research focus, restrictions and assumptions of this research.
1.2 Corporation
Corporation is a worldwide company which employs 79000 people in 125 countries. Company x Corporation produces for: automotive, fluid power, truck and electrical. In total the turnover of Company x Corporation is $ 13 billion in 2007. Figure 1.1 shows the deviation in percentages of the different markets. Turnover of the electrical part is increasing the most compared to the other markets in the period from 2000 to 2007. It is even expected that electrical and fluid power together will take about 75% of total revenues in 2008 (Annual report, 2007). This research is embedded at an electrical division of Company x , namely within Company x B.V at the Netherlands, this division will be explained at the next paragraph.
Figure 1.1: revenues of their four market segments in percentages
$ 4.16 Billion $ 4.81 Billion $ 1.69 Billion $ 2.08 Billion
1.3 Company x B.V.
Company x B.V. produces electrical components for the distribution of electrical power. The components can deal with voltages between 1000 V (low voltage) and 36000 V (middle voltage). Company x B.V. also provides services like maintaining and implementing their products. Company x B.V. can be divided in sub departments. These are Middle Voltage Systems (MVS), Low Voltage Systems (LVS), Low Voltage Components (LVC) and Company x Electric General Supplies (EEGS). These sub departments produce for the electrical market with about 1100 employees and has total revenue of €153-million yearly. In EEGS this research is conducted. This is why only EEGS is explained further.
1.4 Company x Electric General Supplies (EEGS)
EEGS is an internal supplier of Company x and employs about 130 people. In general the production of EEGS can be classified as a job shop environment. This means a high variety of products with a low volume. This variation makes the control of the material flow very complex. Within EEGS there are four functional production lines. These lines can by classified as ‘product market combination’ (PMC) 1 to 4. PMC 1 is known as ‘sheet metal’, PMC 2 as ‘copper bar’, PMC 3 as ‘turning & milling’ and PMC 4 as ‘punching’. This research is focusing on the Product x, a product which is part of PMC1. In paragraph 1.5 the context and background of the Product x is explained.
1.5 Application of Product x
A Product x is a distribution station and is made for applications till 24 KV (= middle voltage) (see Figure 1.2). It includes a generation of ring main units within Company x Electric B.V. A Product x-installation can be delivered in a three of four way version see (see Figure 1.2, right side), from now on shortened as 3f and 4f respectively. The applications are shown in Figure 1.3. For instance; a Product x is installed in a shopping mall to switch or separate the power, it is also functions as a cutout, the number of ‘ways’ play an important role at the amount of tasks a Product x can perform. So with adding on way, the function of cutout or switch can be added.
Figure 1.2: Product x; three/four way and site view
One of the differences of the Product x to its competitors is that the product is very compact and is known for its outstanding reliability. The installation and the mechanism of the product are covered in a sealed cover an important feature of the reliability. A corrosion guarantee up to 15 years in sensitive areas (like salty sea air) is given to its customers.
Unlike its competitors the Product x also does not contain damaging gasses which protect the installation inside. Instead the volume around the installation is made vacuum, which does not harm the environment. This makes the Product x distinctive for the environmental aware customer.
1.6 Motivation and the assignment
Strategic management (Marketing & Sales) of Company x predicts a growth in the demand of the Product x product range. EEGS produces metal sheet parts and welds them together as a frame in which other parts are assembled by MVS. So with the growth the internal production has also to increase within EEGS. Figure 1.4 shows the previously reached production records per period (1 period stands for 4 weeks). From Figure 1.4 can be concluded that the production is already rising to an average amount of about 40 Product x’s a week (see: 2008 gem.). The Product x frame is produced by a welding robot (see Figure 1.5). At the year 2008 this robot should already take care of extra demand namely, sales are predicted at 60 Product x’s a week at 2009 and about 80 in 2010, this would mean a production growth of 50% and 100%. At this moment the robot is already reaching its limits with production. That is the reason management has doubts if the desired output can be reached.
If the desired output of the welding robot can not be produced, the production management is thinking about buying a new welding robot system with prices from €400.000- to €1.000.000-, a costly matter.
Figure 1.4: Production figures of Product x within EEGS, PMC1 (Busschers, 2008)
Not only the quantity of production has to rise but also the quality has to rise. The problem is that PMC1 can not deliver Product x’s which do meet the customer quality demands.
Because PMC1 cannot meet quality demands, MVS the current internal customer, has lots of rework as a consequence. Rework within MVS to compensate the quality problems is more costly, because extra actions for instance recovering the powder coating layer after extra stud welding have to be taken. In the worst case some quality problems of the product are only discovered in the latest phase of assembly at MVS. In that case the product has to be reassembled and reworked totally, a costly matter. From this can be concluded that two major problems are present:
1. the production quantity of the welding robot
2. the quality of the Product x which does not always meets customer demands.
The eventual assignment is to research the factors which do explain these two problems which are apparent within the production line of the Product x at EEGS. From this the following objective and research question is defined:
The objective of this research:
Increasing product quality and production quantity at the production process of the Product x.
Research question:
Which factors are of significant importance to the product quality and production quantity of the Product x and how can these factors be influenced within the framework of PMC 1?
1.6.1 Research focus, restrictions and assumptions
Figure 1.6: The production process divided in three sub processes at PMC1 within EEGS
1.7 Explanation production process:
the side walls, these first have to be manually welded. By manually welding strips have to be placed on these side walls. When welded together also these articles go to stock 614, which supplies the process ‘robotic welding frame’. When arrived at this production process the products enter sub process 2 of production. The robotic welding frame exists out of welding the frame and stud welding. It can be categorized as an assembly process. In this process the welding of the frame and the stud welding is carried out. When the product leaves the robot one makes a quality check. When the quality specifications are not met the product gets corrective actions which exist of reposition the bold and/or recovering leaks of the welds. Following some bolds are manually welded because the robot can not meet the quality specifications in case of positioning and tightness of these bolds. Also scouring has to be done over here because the mig/mag welding does damage the surface on the outside of the frame which can cause corrosion. This surface has to be scoured in such a way that the powder coating does attach and the product can be guaranteed to be anticorrosive for 15 years outside. Also some weld drips (side effect) which do fall on some bolds do have to be removed before powder coating. If not, the bold can break of as a consequence of friction during assembly. So when the corrective action, manual stud welding and scouring are finished the products will go to sub process 3. This sub process only contains powder coating. When finished at this sub process the products leave to the internal customer MVS.
1.8 Restrictions & assumptions
The research has the following restrictions:
The research has been carried out form the 14th of April 2008 till the 14th of August in
2008.
This research involves the diagnosis and design phase of De Leeuw (2003, p. 181), which will take care that problems will be revealed and solution are designed for these problems. The eventual implementation, the ‘change’ phase of De Leeuw (2003, p. 181) will be carried out in a follow up research.
As mentioned in paragraph 1.6.1 this research includes sub process 1 and 2 of the Product x production process. Sub process 3 will not be included.
The following assumptions have been made at this research:
2 Theoretical background and conceptual model
2.1 IntroductionAt this section different variables are explained which can have an influence on the production results of the Product x at EEGS. Paragraph 2.2 will link these variables in a conceptual model. From the conceptual model three sub questions have been derived. Following this section contains the theoretical background which forms the basis of conceptual model. At first in paragraph 2.3.1 the logistical control will be related to the performance level and following in paragraph 2.3.2 to process control. At third a framework for process control and its influence on the performance level will be explained in paragraph 2.3.3. At fourth the influence of the Human factor on the logistical control and eventually on the performance will be explained in paragraph 2.3.4. Paragraph 2.4 explains the methodologies which have been used to answer the sub questions which were derived from the conceptual model in paragraph 2.2. The last paragraph contains the research outline, which makes the structure of the research clear, and how this has been processed at the report.
2.2 Conceptual model Logistical control Performance level Process control Human factor = out of scope = in scope
Figure 2.1: Conceptual model of research
this research. This relationship will be embedded by operations excellence a department within Company x. Eventually from this conceptual model, the following sub questions have been derived, which will be answered during the research.
2.3 Sub questions:
The following sub questions have been formulated:
1) In what way does logistical control influence the performance level? 2) How does the logistical control influences process control?
3) What is the current influence of the human factor on logistical control?
4) In what way can the current logistical control be enhanced in order to improve the performance of the production process?
2.3.1 Logistical control and its relationship to the performance level
Figure 2.2: six operational decisions of Bertrand et all (1998, p. 113)
Figure 2.2 gives six operational decisions which do form the basics for the control of a production department, these are:
1) Capacity planning; this function makes sure that the future flow of production orders for the production department is in accordance with the available production capacity and vice versa.
2) Order acceptation; a production department is responsible for the carrying out of accepted work. A consequent order acceptation is decisive for the realization of short and reliable lead times.
3) Production order release; is pointed at operational controlling of the work in process in relation to the available capacity and material, in such a way that the released orders are under control.
4) Production order detail planning; determines in which predefined times or period a operation has to be carried out.
5) Capacity assignment/ capacity variation; this decision goes along with a possible extension of capacity. By temporary extension of capacity it is sometimes possible to anticipate on a changing order flow with a changing volume.
During the analysis in the following section will be watched at direct effect of these decision functions on the performance level.
2.3.2 Relationship between logistical control and process control
These decision functions from logistical control mentioned before do not only have effect directly on the performance level, but they also can influence the way in which process control is embedded. Following Bertrand et all (1998, p. 183) the relationship between the process control and the logistical control is very narrow. This because one cannot control mistakes fast, when the logistical control is not apparent adequate or very slow. Hopp and Spearmann (2000, p. 293) also claim that the production performance is a composite of the inventory, lead-time and quality efficiency. Following it is stated that the higher the quality, the lower the Work in Progress (WIP) levels can be (Hopp and Spearman, 2000, p. 347) because less outburst. This means when looking the other way around, when reducing WIP, continual efforts to improve quality are necessary to assure an adequate production performance. From these theoretical arguments it is clear that logistical control can have a strong relationship with the process control. These relationships will be applied to the Product x case, to operationalise them. So after having explained the logistical control variable, and the relationship with process control, now the actual process control variable will be explained.
2.3.3 Process control and relationship at the performance level
In the previous paragraph it is explained what the relationship of logistics can be to the process control. At this paragraph the aspects of an adequate process control system is explained in order to reach quality products. In this case the dimension of quality involves conformance. This is the degree to which the product’s appearance and function conform to preestablished standards (Groover, 2001, p. 633). Groover (2001, p. 69-71) claims that one can work against quality problems with the use of process control systems in order to improve the production performance. The control systems in an automated system can be either closed loop or open loop. A closed loop control system, also known as a feedback control system is one in which the output variable is compared with an input parameter, and any difference between the two is used to drive the output into agreement with the output. A closed loop control system consists of six basic elements:
2. process; is the operation or function being controlled
3. output variable; is the output variable that is being controlled in the loop 4. feedback sensor; measures the output variable
5. controller; compares the output with the input and makes the required adjustment in the process to reduce the difference between them
6. actuator; the adjustment is accomplished using one or more actuators
The open loop system works without the feedback loop, so no comparison is made between the actual value of the output and the desired input parameter. Open loop systems are appropriate when the actions performed are very simple, the output is reliable and any reaction forces opposing the action are small enough to have no effect on the actuations. If these characteristics are not applicable, then a closed loop control system may be more appropriate.
Quality control (QC) further involves the operational and engineering activities and serves to guarantee the quality of the product. The process has to be under control totally; in such a way that minimal control is needed. As shown in Figure 2.3 (Taguchi, 1989) quality control systems do exist out of supportive activities (off-line QC) and inspection activities (on-line QC). Off-line control goes along with designing product and processes and goes before the actual production process. On-line quality control goes along with the inspection of the actual production and the communication or relation with the customer after sending the product. Figure 2.3 gives the relationship between on-line and off-line QC. With Figure 2.3 one can analyze and control its processes, which results in better quality. However the focus of this research will not be on process control and its eventual effect on the performance level. At this moment the process control will be analyzed further by operations excellence, a department within Company x which is embedded in a Six Sigma1 improvement project. For readers still interested in the off-line on-line quality control framework they are referred to Groover (2001, p. 69-71).
1
Figure 2.3: Taguchi's off- en on-line quality control
2.3.4 Human factor and its influence on the logistical control
are assumed to have an impact on the logistical control are: training/education, teamwork and culture.
2.4 Research methodology
In this paragraph the methodology to answer each sub question has been explained. For the first two sub question the same methodologies have been used. The first two sub questions were as follows:
1. In what way does logistical control influences the performance of the production? 2. How does logistical control influences process control?
To answer these questions the methodologies have been appointed to each sub process, in this case the first two sub processes in the production process, as shown in Figure 1.6. The methodologies used for each sub process are as follows:
Sub process 1
Existing Overall Equipment Efficiency (OEE) data: As one could read in paragraph 1.6 the
production problems reveal itself at the robot welding process. From this information the causes of the production problems are searched in a software application which keeps up the OEE at the welding robot located at sub process 2, but which can trace down problems which are caused by sub process 1. The OEE is filled in by the operators for about 1,5 years now and categorizes the following problems:
1. no orders 2. breakdown robot 3. breakdown equipment 4. no material 5. setup production 6. no operator
operators do log in at the same computer. Nevertheless these data are still taken to analyze the unforeseen problems. To get the data more reliable a correction factor has been carried out. It is assumed with correction a more reliable judgment can be made. The actual correction is explained at paragraph 3.2.
Baan: After analyzing the OEE, it seemed that ‘no material’ caused the most production
disturbance at the welding robot. From this Baan is extensively used to analyze logistical data to search for causes of ‘no material’. Baan is an Enterprise Resource Planning (ERP) software package which Company x (Company x ) uses. An ERP package is an integrated system which can plan and control raw materials, components, people, finances, machines etc. At this way it can control the total flow of materials from purchase till delivery to the customer. This research is partly based on the data from Baan. The logistical data for which have been searched for are based on the six operational decisions mentioned at paragraph 2.3.1 and have been applied on the actual components (articles) from which the Product x is welded. From the six decision variables the following variables are derived: stock levels (WIP), routings, lead times and capacity. Following this information extracted from Baan has been validated by the responsible logistical managers during meetings. In case of the actual lead times at sub process 1 they could not confirm the reliability of the data at Baan. An experiment with 2 orders has been put up for that, which was followed daily. With this experiment the actual lead times of BAAN could be validated.
Meetings and working with operators: Because Baan does not always explain the causes of
the data which its presents, a lot of confirmative meetings have been carried out with supervisors, logistical managers, planners and product drawers and operators. Especially in the beginning the operators were cautious and reluctant at these meeting. To overcome this distance another approach has been searched for. Eventually the approach has been carried out in the form of working together with operators for about month. This means in their normal environment on the work floor. Talking and working with employees in their ‘natural’ environment provided helpful information and new perspectives. So the conversations with the people at the work floor could be seen as a valuable data source.
Sub process 2:
Forms for logistical and effect on process control: The operators at sub process 2 have been
process 2, because these data were not available at BAAN. This means that a Product x starts (first production step at sub process 2 of Figure 1.6) it is welded in the welding robot and when it ends it has been manually welded (the last production step at sub process 2 of Figure 1.6). From these forms the lead times, WIP, work order sequences and the planning is analyzed. Also the effect on process control could be determined.
Meetings and working with operators: Also at sub process 2 a lot of conversations have
been carried out on the work floor with the operators, supervisors, logistical managers, and planners. The meetings and working with operators has been seen as necessary to get the practical information from the work floor, in this case how they handle quality and logistical aspects at sub process2.
With the information above the first two sub questions could be answered, following the third sub question needs to be answered:
3. What is the current influence of the human factor on logistics?
To answer this question the human factor of only sub process 2 has been analyzed. Different methodologies of data gathering have been used. Below these different methodologies are explained:
Desk research and reflection by observation: With desk research the theory behind the
factor has not been treaded as a full variable in the conceptual model, but only as a moderator which can ‘help’ the logistical control in order to improve the production performance level. OEE: The OEE has been analyzed to search for problems which can be related to the human factor. See for further explanation of the OEE page 14. At this time the OEE has been used to see how many, how long and with variance disturbances could be discovered, which cause production problems.
Summarizing sub question 1,2 & 3 with value stream mapping (Rother and Shook, 2003
p. 3): The findings of the first three sub questions have been processed in a value stream map, a technique to define the current state of the production flow. The value stream map summarizes visually the value and not value activities at the production flow, which can declare the logistical problems at sub process 1 & 2.
The questions above involved the diagnosis phase of the Leeuw (2003, p. 181) of the research. After the diagnosis the fourth sub question could be answered:
4. In what way can the current logistical control be enhanced in order to improve the performance of the production process?
This question mainly involves the design phase of the Leeuw (2003, p. 181) of the research. At the design phase problems factors which arose from the diagnosis of the logistics and the human factor are tried to incorporate in the solutions. Before coming up with a new logistical design the following steps have been carried out:
Desk research: At the literature which came to the fore during the studies Technology
Management at University Groningen has been searched to possible solutions of the problems at sub process 1 and sub process 2. When the literature did not provide all the answers to the problems prof. dr. ir. J. Slomp of the Lean Operations Research Center (LORC) at the University of Groningen has been consulted. This consult appeared to be of great worth to the design phase, because he came up with an innovative solution to the logistical problems at sub process 1. The solution he came up with was derived from the Master Thesis report of Tim Staudt (2006), a report which partly has been used as a guide during the design phase.
Meetings two times a week: To validate the findings from the previous step the practice has
and operators were involved to judge the possible solutions during meetings. The logistical managers assisted with defining the most realistic logistical design for sub process 1 and 2.
Summarizing sub question 4 with value stream mapping (Rother and Shook, 2003 p. 3):
The findings of the third sub question have been processed in a value stream map, this time the technique has been used to define the future state of the article flow of the Product x at EEGS. The value stream map summarizes visually the value and not value activities of the production flow, which can show the differences with the current state of the diagnosis phase.
2.5 Research outline
In Figure 2.4 the outline of this report has been presented, which involves the chapters of the report, the research stages of de Leeuw (2003, p.181) and the related subjects. The overview given over here is given for each of the 4 chapters. Chapter 1 contains the problem statement. Chapter 2 contains the theoretical background of the research and the research methodologies. Only for logistics and the human factor a diagnosis has been carried out (sub question 1, 2 & 3). These objects of research are carried out concurrently. Findings or suggestions from the human factors diagnosis will also be used during the design of a new logistical system which represents the small striped arrow at the figure. Chapter 4 does contain the conclusion and recommendations of this research.
3 Diagnosis of production process of the Product x
3.1 IntroductionAt this section the first two sub questions will be answered as defined at paragraph 2.3. and will be carried out respectively at paragraph 3.2 and 3.3. Following the third sub question has been answered at paragraph 3.4. At the end of the section a general conclusion will be given about the first three sub questions.
3.2 Logistical control sub process 1; from punching to stock 614
This paragraph gives answer at the first two sub questions:
1. In what way does logistical control influences the performance of the production? 2. How does logistical control influences process control?
First this paragraph will analyze the logistical control of the production process of the Product x at sub process 1 as schematically shown in Figure 3.1.
Figure 3.1: Focus on Sub process 1
3.2.1 Overall Equipment Efficiency (OEE)
Deviation production time robot
0% 10% 6% 14% 6% 60% 4% 0% 10% 20% 30% 40% 50% 60% 70% No orde rs Bre akdo wn robo t Bre akdo wn equ ipm ent No mat eria l Set up Pro duct ion No Ope rato r 'activity' % o f p ro d u c ti o n t im e
Figure 3.2: OEE data, kept by operators, representing the period 12-03-2008 to 12-6-2008. Explanation of the activities at Figure 3.2:
no orders; the buffer after the robot is full (is only the case when pull will be applied)
breakdown robot; there is a failure at the robot itself
breakdown equipment; there is a failure is the welding head, supply of welding wire or studs
no material; there is no sheet metal, welding wire or studs setup; reprogramming of robot
production; robot is operational
no operator; the operator is not present or busy with other tasks
However, a side mark has to be made about the OEE data. As mentioned before it became clear that not all operators are consistent in keeping up to date the OEE. Sometimes the personal computer is turned off, when it should keep up OEE. Also a practical problem seems apparent by tracking the OEE. This problem is that OEE data get lost of a particular
production Production output number of weeks average unit production a week practical production time a unit production time a week available hours a week % of actual production time 3f 577 27 21,4 0,75 16,0 75,5 21,2% 4f 319 27 11,8 0,92 10,9 75,5 14,4% back wall 286 27 10,6 0,28 3,0 75,5 3,9% total % 43,8 39,6%
production day when different operators do log in at the same computer. Nevertheless these data are still taken to analyze the unforeseen problems. To get the data more reliable a correction factor has been carried out. It is assumed with correction a more reliable or valid judgment can be made about the magnitude of the problems.
The correction has been done from the real production output of the year 2008, from week 1 to 27. The production output is shown in the second column of Table 3.1. The real practical production time of one unit 3f or 4f at the welding robot are derived from a previous research at sub process 2 of Rientjes T. (2007). The results of his research were that that one 3f takes 0,75 hours and one 4f 0,92 hours of production. Table 3.1 shows that 75,5 production hours (during two shifts) are available each week. With the actual production output of 27 weeks, the available production time and the time to produce one Product x it is possible to calculate the actual OEE. The formulas belonging to Table 3.1 are:
production output / number of weeks = average unit production a week
average unit production a week * practical production time a unit = production time a week production time a week / available hours a week = % of actual production time
total % = 3f + 4f + back walls of % of actual production time
From these formulas the actual OEE results in only 40% (39,6%) of production time in stead of the measured 60% in Figure 3.2.
%actual production time - % measured disturbance time = % difference between actual and disturbance time
So the corrected value gives a difference of 20% (60% - 40%). So, the 20% difference in production time has to be added to the non production time, which will become 60% now. This 60% of non production time will be divided in the same ratio as before the correction, because is has been assumed that the data are measured consequently wrong and not that they are manipulated by the operators on purpose. So when these are measured consequently wrong the deviation of the non production time is most likely the same as before. This has been processed in Table 3.2 with the following formula:
current ratio of disturbance time * % non production time = corrected value %
15%. The rest of the results are calculated the same like this and these are shown in Table 3.2. In this paragraph the focus is at no material which causes 21% of down time of the welding robot. No material can be a consequence of the logistical control, which will be analyzed from now on.
3.2.2 WIP level at stock 614 and lead times
When analyzing Table 3.2, ‘no material’ is with 21% consumption of available production time the biggest disturber of the production of the welding robot. This can occur because the welding at the robot can be seen as assembly and for this a complete set of parts/articles is needed. So when just one single article is not available production stops. Figure 3.3 is the first of three different counts of the WIP level at stock 614. The other two counts with their relating graphs are shown in Appendix III (Table III.I and related Figures III.I & III.II). At the graphs the article names which are needed in stock 614 are shown. When in the graph behind the article name 3f or 4f is present then the article can only be used in that type of Product x, if not present it can be used in both. When it can be used in both the different articles have to be shared by the 3f and 4f in a ratio 2:1.
Number of Xiria's which can be produced with stock 614
0 20 40 60 80 100 120 140 supp ort p late strip Lpro f 3f Lpro f 4f supp ort_ plat e side -wal l lef t side -wal l rig ht back -wal l bol ted 3f prof ile fo r bol ted back wal l Fron t-pla te- b in b elow -3f Fron t-pla te- b in b elow -4f back -wal l bol ted 4f Fron t-pla te-3 f Fron t-pla te-4 f article n u mb e r o f p ro d u c ts w h ic h c a n b e p ro d u c e d 3f 4f
Figure 3.3: Counting 1 of WIP level at stock 614 at 26-6-2008 ‘activity’ measured OEE Correction Corrected value % breakdown robot 10% +5% 15% breakdown equipment 6% +3% 9% no material 14% +7% 21% setup 6% +3% 9% production 60% -20% 40% no operator 4% +2% 6%
This means that for the support plate (up), strip, support plate (down), side wall left, side wall right, and the profile for back wall 2 articles are needed in both types (configurations) of Product x. For these articles the following formula has been used:
(number of articles at stock 614/number of articles in configuration) * demand ratio = number of products which can be produced
So when for instance 6 articles in stock 614 are available needed 2 per Product x, 6/2*2/3= 2 3f can be produced and 6/2*1/3 = 1 4f can be produced. But the articles support plate (up), support_plate (down) and profile for bolted back wall form an exception, which are also divided differently for the 3f and 4f, because of difference in configuration. This means that 2 articles are needed for the assembly of one 3f and 3 articles for one 4f.
So when for example 351 article are available in stock 614 then (351/2) 2/3 = 117 (3f) and (351/3) 1/3 = 39 (4f) can be assembled, which is together 156 Product x’s. When calculating at the same way the articles of stock 614, the graphs shows that the wrong deviation of the article numbers are present at stock 614.
count 1 count 2 count 3
max 3f 0 13 0
max 4f 10 7 4
max 3f + 4f 10 20 4
Table 3.3: Summarized how many Product x's could be welded with the articles in stock 614
at three different point of time
Art name with conveyor
Lead time parts 3f in weeks Lead time parts 4f in weeks support plate 3,7 2,5 strip 3,6 3,6 Lprof 3f 3,9 Lprof 4f 4,0 support_plate 2,8 1,9 side-wall left 1,2 1,2 side-wall right 1,1 1,1 back-wall bolted 3f 0,4
profile for bolted back wall 1,3 0,9 Front-plate- bin below-3f 0,8
Front-plate- bin below-4f 2,2
back-wall bolted 4f 4,6
Front-plate-3f 0,9
Front-plate-4f 1,6
Table 3.4: Lead time at stock 614 per article
The average lead time resulting from the average stock level is presented in Table 3.4. The belonging formula is:
(average number of articles at stock 614/number of articles in configuration) * demand ratio = number of products which can be produced
number of products which can be produced / demand = lead time
What can be concluded from Table 3.4 is that the lead times at stock 614 are different and can be up to 4,6 weeks. The products which have the lowest inventory level on average do have the shortest lead times.
derived from the routing data out of BAAN, shown in Appendix IV. At the appendix the routing data for each article are presented in table IV.I In the column ‘omschrijving’ (in English description), one can see the production process which have to be followed in order by a certain article. These data have been transferred to visual routings in Figure 3.4of the 3f. The 4f nest has exactly the same routings except that its end point is stock 614 (4f). When articles do have the same arrow color in Figure 3.4 they follow the exact same production steps. The arrows between the store 614 (3f) and store 614 (4f) present leveling of articles. This means that when for example the article number 665166 can be used in a 3f or 4f. So sometimes it occurs that when the 3f is without such an article, this article will be taken from the store 614 (4f), which will be as a consequence without any parts later. An exact level of occurrences has not been determined, but it happens according to the operators on a daily basis.
From the software package BAAN it has been analyzed that the total lead times of the articles to stock 614 do differ enormous, which has been presented Figure 3.5. These real lead times recorded by BAAN do differ from about 4 days to about 30 days (the strip is not a part of the fixed nest during punching, that is why it is not displayed) in Figure 3.5.
Average lead time in days per article nr. in nest to stock 614
0 5 10 15 20 25 30 35 supp ort p late strip Lpro f 3f Lpro f 4f supp ort_ plat e side -wal l lef t side -wal l rig ht back -wal l bol ted 3f prof ile fo r bol ted back wal l Fron t-pla te- b in b elow -3f Fron t-pla te- b in b elow -4f back -wal l bol ted 4f Fron t-pla te-3 f Fron t-pla te-4 f article ti m e i n d a y s in 3f nest in 4f nest
Figure 3.5: Real lead times of the articles in a nest to stock/store 614 (data from 28-9-2007 till 4-6-2008)
The longer lead times in Figure 3.5 are not only the consequence of the theoretical lead times, as presented at Figure 3.6.
Theoretical lead time in days to stock 614
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 supp ort p late strip Lpro f 3f Lpro f 4f supp ort_ plat e sid e-wall left side -wal l rig ht back -wal l bol ted 3f prof ile fo r bo lted back wal l Fron t-pla te- bin b elow -3f Fron t-pla te- bin b elow -4f back -wall bol ted 4f Fron t-pla te-3 f Fron t-pla te-4 f article th e o re ti c a l le a d t im e i n d a y s 3f 4f
3.2.3 Utilization of capacity
At meetings it became clear that it takes for one total nest to be taken out of STOPA 1 hour, this has been checked by the operators which have confirmed this information. Even if this 1 hour (of delay) seems long, this time cannot declare excessive lead times up to 30 days, it also cannot declare differences between the lead times of different articles. The next step has been to search for the cause of these excessive and different lead times in the utilization of the different process steps. With a high utilization it could be that long queues are apparent with delays as a consequence in the production process of the different articles. The utilization of the Product x line to stock 614 has been shown in Table 3.5. In general the delay in Table 3.5 as a consequence of utilization can be written in the following formulas:
The conclusion of the data from Table 3.5 is that actually none of the four process steps which have to be carried out are fully utilized. It is expected that the utilization on average causes delays of no significance. This has been calculated with the formula of Erlang (Bertrand et all, 1998, p. 226), which is:
expected waiting time = expected time utilized * utilization degree / (1 – utilization degree)
average utilization capacity in hours Expected waiting time in hours
punching nc Trumpf 5000 5,60% 80 0,27
bending 11,64% 450 6,90
manual welding 26,37% 32 3,02
benchwork klinkm. 24,90% 32 2,64
Table 3.5: Average utilization of the Product x line from week 27 to 35 in 2008
are produced and the easiest are taken, which again do not solve the out of stock problems. When after several times of ordering the articles are still not delivered which were not needed, operators of sub process 2 ask (pull) if some operator of sub process 1 can bench, bend etc the articles which are out of stock. Mostly this can be done directly because the articles needed are already punched ad available at STOPA, which is an automated store system. Following the needed articles in STOPA are not taken out with the ‘first in first out’ (FIFO) principle, which can explain the differences in lead times and what makes the average lead time higher. The same is applicable to products which are just left behind on the work space as WIP, before they reach stock 614. To confirm these statements of the operators the days in STOPA have been analyzed, these are the days that they are not picked up and only the easier parts are taken. So this is the measured lead time (from Figure 3.5) which cannot be declared by utilization, setup and production times. In formula form:
days in STOPA = actual measured lead time - production time – setup time – expected waiting time
Average days in STOPA per article
0,00 5,00 10,00 15,00 20,00 25,00 30,00 35,00 supp ort p late strip Lpro f 3f Lpro f 4f supp ort_ plat e side -wal l lef t side -wal l rig ht back -wal l bol ted 3f prof ile fo r bol ted back wall Fron t-pla te- b in b elow -3f Fron t-pla te- b in b elow -4f back -wal l bol ted 4f Fron t-pla te-3 f Fron t-pla te-4 f days in STOPA a rt ic le 3f articles 4f articles
Figure 3.7: Average days in STOPA of the articles
3.2.4 Conclusion of logistical control at sub process 1
The logistical process at sub process 1 influences the performance level of the production process negatively. This is caused by lead times problems of the articles to stock 614. The lead time problems are caused by two facts. Fact one; the lead times are different of the related articles as a consequence of the different production routings. In practice those articles cannot arrive simultaneously and can cause starving of the welding robot.
Fact two; the operators like to produce the easiest parts to produces at first and leave the more difficult parts in a buffer, also called STOPA. The consequence is that the variance between lead times of the related articles occur and that the lead times become to long. This makes that the article sets are not complete at stock 614 at the moment of production of the welding robot.
Fact one and fact two deteriorated each other. The consequence is that the difference in lead times of the related articles become too long to supply stock 614 in an optimal way.
The effect of different theoretical lead times is deteriorated by the way of working of the operators.
3.3 Logistical control at sub process 2; from the welding robot till manual welding
This paragraph also likes to answer the first two sub questions:
1. In what way does logistical control influence the performance of the production? 2. How does logistical control influence process control?
This paragraph will analyze the logistical control of the production process of the Product x at sub process 2 as schematically shown in Figure 3.8.
Figure 3.8: Focus on sub process 2
3.3.1 Production output / order release
The results of the production output are shown in Figure 3.9. This figure shows the deviation of the production number (output) and the number of times that they occurred.
Deviation of production Xiria at welding robot
0 10 20 30 40 50 60 70 80 90 100 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 production number ti m e s o c c u re d 3f + 4f 3f 4f
Figure 3.9: Production number (output) and the times they occurred in the period week 1 till 27 in 2008
The most remarkable at Figure 3.9 is that in total 91 times of 264 samples no production occurred from a 3f or 4f at a certain day. This means that in total 35% of the samples no production occurred at all. Only the logistical control at sub process 1 can not clarify these non production days, because these only had 21% impact on production time, which is as nearly as much as 35% of no production days. So the cause of the non production days also has to be searched somewhere else. Further the graphs of the 3f and 4f show similar patterns. However as one can see is that the standard deviation of the 3f is larger in comparison to the standard deviation of the 4f (see Table 3.6), but the output of the welding robot can be called in both cases not regular The total average is about 6,5 Product x’s of production per day. This average is almost mixed up at the market demand of 2:1 (3f:4f) in practice (4,3:2,3).
Table 3.6: Production figures of the 3f and 4f of the welding robot
3.3.2 Work in progress (WIP) at sub process 1
One of the complaints of the operators was that the quality checks are not carried out directly. This results in that the outcome of the quality checks cannot be used anymore to take immediately and adequate action to prevent any more production with non confirming
quality. This problem is caused by the buffer between the robot welding and the quality check. To validate these statements of the operators, actual measurements have been taken.
Lead time of finished orders at welding proces
0% 20% 40% 60% 80% 100% 120% < 1 dag < 2 dag < 3 dag < 4 dag < 5 dag < 6 dag < 7 dag < 8 dag < 9 dag < 10 dag < 11 dag < 12 dag < 13 dag < 14 dag < 15 dag < 16 dag >16 dag time % c u m m . fi n is h e d 3f+ 4f 3f 4f
Figure 3.10: Lead time of finished Product x’s/orders at sub process 2 (measured 13-6-2008 till 14-7-2008)
From these measurements one graph is made which are shown in Figure 3.10. The conclusion from Figure 3.10 is that within the first five days almost 60% of the Product x’s are through the welding process, the other 40% within 16 days. The reason is that the actual space of the buffer is full (see Figure 3.11) and that at some areas are better accessible in the buffer. So the easiest accessible Product x’s are taken out first from the buffer. The other 40% take longer because they are less accessible. Table 3.7 summarizes the measurements of Figure 3.10. From this table can be concluded that the 3f takes 6,5 days and the 4f takes 7,9 days to be finished, the period it takes when quality problems will be revealed. The minimum of these times could be 0,11 days and 0,12 days respectively (production time at robot + quality check with marker + manual welding and scouring).
average days in buffer
3f 4f 3f + 4f
finished 6,5 7,9 7,0
In theory 0.11 0,12 0.11
From the forms also the WIP data at sub process 2 have been determined. Figure 3.12 shows the actual WIP levels belonging to the buffer Figure 3.12 shows that the WIP at the welding process can peak up to 78 Product x’s in total. The average WIP level has been 50 Product x, with a standard deviation of 11 (see Table 3.8). The WIP level is divided by 57% (28/50) for the 3f and 43% (22/50) for the 4f. The first three bars at the graph are not present, these data are erased because the measurements went wrong, as the operators needed to get used to the filling in the forms.
WIP at welding proces
0 10 20 30 40 50 60 70 80 90 13-6 -200 8 15-6 -200 8 17-6 -200 8 19-6 -200 8 21-6 -200 8 23-6 -200 8 25-6 -200 8 27-6 -200 8 29-6 -200 8 1- 7-2008 3- 7-2008 5- 7-2008 7- 7-2008 9- 7-2008 11-7 -200 8 13-7 -200 8 15-7 -200 8 date n u m b e r 3f+ 4f 3f 4f
Figure 3.12: WIP level at the buffer of sub process 2 (measured 13-6-2008 till 14-7-2008) average total WIP 50
stdev. 11
sample size 29
One would expect that the output follows the production pattern of the input. However Figure 3.13 shows that the output does not follow the input and no other pattern can be found between input and output. Table 3.9 summarizes the consequences of the pattern in Figure 3.13. At this table the average output is 5,6 a day, which is 1,4 lower than the input of 7 Product x’s. This makes it likely that this is the cause of the high WIP levels. When the production floor is full a big ‘clean up’ of WIP will be carried out and. when there is almost no WIP anymore the same cycle of less output than input will be gone trough.
Number of Xiria's processed at phase 2
0 2 4 6 8 10 12 14 16 18 13-6 -200 8 15-6 -200 8 17-6 -200 8 19-6 -200 8 21-6 -200 8 23-6 -200 8 25-6 -200 8 27-6 -200 8 29-6 -200 8 1- 7-2008 3- 7-2008 5- 7-2008 7- 7-2008 9- 7-2008 11-7 -200 8 13-7 -200 8 15-7 -200 8 date N u m b e r o f X ir ia 's p ro c e s s e d
Started Xiria's at phase 2 Finished Xiria's at phase 2
Figure 3.13: Started (input) and finished (output) Product x’s processed at sub process 2 started Product x's Finished Product x's average 7 5,6 stdev. 4,3 5,5 sample size 23 23
Table 3.9: summarized data of Figure 3.12
3.3.3 No quality usable quality feedback system
Following production management FIFO has to be carried out. From the forms which had to be filled in also a judgement could be made about the performance in case of FIFO. Table 3.10 shows an example of one production day which shows the degree in which FIFO has been carried out. This table presents the exact sequence in which Product x’s have been finished at that day, this means when they leave sub process 2. As one can see in the table at the red boxes the finished Product x’s are not selected on the date that they are started or on the Product x number which are given by the date of start in chronological sequence. Actually in this example the Product x’s which have been started first at the welding robot, are finished as last, and vice versa. Because the input which goes into the production process is not put in chronological sequence (or Product x number) into the production process a structured and useful quality feedback can not always be given.
Product x
number date of start time of start date of finish time of finish
828202 8-jul 6:15 15-jul 6:30 828402 10-jul 7:15 15-jul 8:00 829103 14-jul 8:20 15-jul 7:00 829105 14-jul 10:30 15-jul 7:30 826103 23-jun 19:25 15-jul 9:00 826208 24-jun 21:45 15-jul 10:00 826203 24-jun 13:00 15-jul 11:00 826202 24-jun 11:15 15-jul 13:00
3.3.4 Conclusion of logistical control at sub process 2
The conclusion of the logistical process at sub process 2 has been presented in Figure 3.14 and explained further over here. The influence on the performance level can be ascribed to the long lead times, which in theory should be 0,11 days for the 3f and 0,12 days for the 4f, but are in practice respectively 6,5 days and 7,9 days. The long lead times are caused by the high WIP levels of the buffer. The high WIP levels at the buffer are caused by no real production planning targets, which have to be met by the operators. This makes that certain periods can have less production output than input at the production process. The consequence is that WIP levels rise. These WIP level cause two major process control problems. At first the high WIP levels delay quality feedback. At second the WIP is not taken for production with the FIFO principle or any other structured way. The results of the quality check cannot always be used anymore. In this situation structured quality feedback can not always be given in time to interfere in the production process of the welding robot.
3.4 Human factor
This paragraph contains an analysis of the human factor at sub process 2 which have been schematically displayed at Figure 3.15 and gives answer at the second sub question:
2) What is the current influence of the human factor on logistical control?
Figure 3.15: Focus on sub process 2
First has been looked in what kind of way the team work looks like, and following how much training and education the operators do have. As last the culture has been analyzed.
3.4.1 Team work at welding process
At sub process 2 the operators work in teams. The way in which team work is embedded can influence the production results. The production results are influenced by team members which cannot always deal with unforeseen problems and Bacdayan (2001) claims that this dramatically influences the production output. Table 3.2 showed the OEE with the production disturbers ‘breakdown equipment’ and ‘breakdown robot’ which can be categorized as ‘unforeseen problems’. These two factors together consume 24% (15% + 9%) of available production time.
Table 3.11 categorizes the OEE in absolute numbers. The mean time of happening, the mean time between happenings, stdev time of happening and std dev. of time between happening
Mean time of happening Mean time between happening Stdev time of happening Stdev. of time between happening No material 2:13:54 63:22:43 2:09:53 242:09:15 No operator 0:52:06 23:10:55 1:24:37 168:55:30 No orders 0:18:52 23:10:55 0:23:19 19:23:23 Production 2:30:03 14:44:19 4:33:53 31:06:31 Set up 1:55:49 110:42:07 4:14:41 142:10:39 Tool breakdown 1:21:10 27:17:30 42:21:03 82:20:35 Robotic breakd. 1:12:36 28:46:23 1:47:12 84:00:32
are not corrected (as in paragraph 3.2.1) because no realistic correction can be made about these. It is likely that these data are actually worse than they show in the table. Still the data are taken as a guideline to analyze how the operators can deal with unforeseen problems. From this it appears that the tool breakdown has a mean time of 1:21:10 hours, the standard deviation is 42:21:03 hours and this happens on average every 27:17:30 hours of production time. From these data can be stated that enormous variation in the process is apparent, which has as a consequence that one can not predict precisely how long the breakdown will take. The robotic breakdown has a mean time of 1:12:36. The mean time between breakdowns of the robot are on average 28:46:23 hours. The robotic breakdown has a standard deviation of 1:47:12 hours which is still considerable.
From these figures can be concluded that there is no control in case of predicting breakdowns. This appears from the enormous variation, 82:20:35 hours standard deviation between happening by tool breakdown and 84:00:32 hours standard deviation by robotic breakdown. Because of this uncertainty in control team members can get frustrated and blame the production process for the wrong reasons (Van der Vegt en Van de Vliert, 2002). One way of preventing this is making sure that the team is composed on result in stead of tasks( Van der Vegt en Van de Vliert, 2002). Task related team work stands for the degree in which team members do have to share materials, information and expertise before the preferred output. Result related teamwork stands for the degree of performance which has to be reached by the team. At the welding robot, a degree of performance could be to reduce the tool breakdown (reactive) and robotic breakdown with a certain degree which can be defined later. Also targets for production could be set. These two factors do meet each other, the less break downs the more production and vice versa. Within the welding process the teams are not judged by result on a daily basis, this appears from statements like this one: “I stop producing for today, pffff yesterday I was tired when I came home!” (this operator had still 7 working hours to go). This decision was based on his individual targets and he was not worried that he would influence the performance of the welding robot negatively.
