Throughput time reduction of the Test and Certification Process.

(1)

Throughput time reduction of the Test and

Certification Process.

KEMA Quality- Products Consumer Appliances

Author: Wessel S. Mulder Student Number: 1462431

Study: Technology Management – Discrete Technology Company: KEMA Quality, Arnhem, Netherlands

(2)

II

PREFACE

After six months of intensive work at KEMA Quality B.V., I proudly present this final piece of research rounding off the inspiring master program of „Technology Management‟. The gratifying working environment of KEMA seemed to be a perfect place to experience the resemblances and divergences between practice and theory. This definitely aroused my awareness to further develop my practical skills and knowledge.

As a matter of course, I want to thank all former colleagues for their willingness to contribute to this master thesis. I sincerely hope that the experienced problem will be grappled with the suggestions in this research.

Furthermore, I am grateful for the inspiring and congenial conversations I had with my direct supervisor Hen van de Water. And, I also owe thanks to my co-assessor dr. G.C. (Gwenny) Ruël.

(3)

III

MANAGEMENT SUMMARY

This research took place at KEMA Quality which is a global company in the field of inspection, testing and certification of electrical, electronic and medical devices, as well as auditing and certifying management systems. Due to a recent merger, KEMA Quality became part of DEKRA, an international safety consultancy company. Management of KEMA Quality expected that the standardised test- and certification process within the business line „products‟ faced efficiency problems. Initial investigation pointed towards long and uncertain throughput time as one of the main causes for efficiency concerns. Therefore the following research purpose was formulated:

“to decrease the average throughput time of the standardised test process”.

Moreover, the research scope has been extended towards the development of a theoretical model which guides the improvement of workload balance in companies carrying the unique characteristics such as KEMA Quality.

At the beginning of the research it was validated that long throughput time is a real problem KEMA Quality is facing. During an extensive problem diagnosis phase consisting of multiple research methods, several causes for long throughput time were found, such as „substantial customer dependency‟, „insufficient ICT facility‟, and „inadequate availability of test equipment‟. In addition, lopsided workload balance and excessive amounts of work in progress (WIP) seemed to be significant causes for long and uncertain throughput time. Also during the cluster meetings the unbalanced workload became a delicate topic. Hence, it was expected significant improvements in throughput time could be made by intervening in the balance of workload and amounts of WIP.

The unique characteristics of KEMA forced me to use both production planning and control as cybernetic literature. Because of the reduced dependency between shop floor workers on lower task levels and the difference in predictability between on one hand the engineer and on the other hand the project managers and support office, the intervention in workload and WIP is decomposed into „control of the engineers‟ and „control of the support office and project management‟. This decomposition complies with the concept of nearly decomposable systems of Simon (1962) which preserves intra-homogeneity.

First several general control interventions were recommended:

- The introduction of a pre-shop pool which absorbs all kinds of fluctuations at the work floor and reduces the influence of the customer (defined as the task environment).

- The adoption of a maximum on-hold period in order to reduce the influence of the customer. - Allowance time should be calculated based on a combination of individual process time

(4)

IV

A second category of interventions concerns the workload control of support office and project managers. Point of focus is exclusively on the improvements of workload balance since the unpredictable nature of work does not allow control of WIP. The aim is to search for a satisficing solution (Simon, 1945) due to the complex nature of the design task and the many uncertainties. It is suggested to control the workload of support office and project manager by means of intrinsic management. Extrinsic meta-management is proposed to measure the effectiveness of the management body. Intrinsic management utilizes the capacities of individual elements by which it matches circumstances with significant uncertainties analogous to KEMA Quality. To increase the effectiveness of intrinsic management of KEMA two additional interventions are recommended. First, each member of the support office should be located at the office of the project managers, responsible for the business teams. Secondly, the exchangeability of work between shop floor workers should be increased to improve the allocation of workload.

The predictability of work for the engineers is significantly higher and the intensity of disturbance is lower. Hence, the analyzability of the problem is of more mechanistic nature according to the System of System Methodology (Flood and Jackson, 1991). Therefore, the relatively „hard‟ production planning and control literature can be applied to control the workload of the engineers. It turned out that there is no integrated PPC concept available in literature that fits the unique characteristics of KEMA. A customized framework for the improvement of workload balance and reduction of WIP was designed. A more detailed planning provides more insight in the activities executed on the work floor, decreases the friction between project managers and engineers about declarable hours, enhances controllability, prevents preferential behavior, and increases the possibilities to exchange activities between employees. Furthermore, WIP should be reduced by imposing upper norms to the quantity of released work. Considering release, it is advocated to fulfill the sequencing of jobs by the calculation of release dates. Such a date can be defined as the due date minus the expected waiting time on the shop floor and the sum of all processing times including some slack to compensate estimation deficiencies. Dispatching has to be done analogous to release. A simple flow conserving priority rule such as First-Come-First- Served should guide the selection of activities.

This report ends with a theoretical model which guides the process of workload improvements in companies similar to KEMA Quality, according to the systemic perspective.

(5)

V

VI

4.2.2. Customer enquiry stage ... 50

4.3 Workload improvement, project managers and support office ... 52

4.3.1 Extrinsic management ... 53

5. Design Phase Engineers: Workload improvement and WIP reduction ... 55

5.1 Shop floor characteristics standardised test process ... 55

5.2 PPC criteria ... 57

5.3 PPC concept selection ... 58

5.3.1. Preliminary Study & Evaluation ... 58

5.3.2 Detailed Investigation & Final selection... 64

5.3.3. Elaboration workload improvement and WIP reduction ... 68

Reflection.... ... 77

Further research ... 81

Literature…. ... 82

Appendices ... Appendix 1: Preliminary cause-and-effect diagram ... 86

Appendix 2: Current Value Stream ... 87

Appendix 3: Allocation of functions ... 88

Appendix 4: Elaborated Quality Management System ... 93

Appendix 5: Open coding ... 98

Appendix 6: Limit Matrix Analytical Hierarchic Process ... 100

Appendix 7: Analysis of project to obtain an estimation of throughput time ... 101

Appendix 8: Backlog activities ... 102

Appendix 9: Data concerning capacity ... 103

Appendix 10 M/M/1 approach ... 104

Appendix 11: Planning schedules ... 105

(7)

VII

ABBREVIATIONS

List of abbreviations Abbreviation Meaning AT BOB CB CERES CM CONWIP CRM CSA DDC- model DFTB BPS EA EN (Q)Eng. ET ETO IEC IR JPF KK KPI LOMS LUMS MB MPS MRP Ι MRP ΙΙ MS MTO OMS PB PE PM POLCA PPC QMS QR QUA Allowance Time

Beproevings Objecten Beheer Certification Body

Certification Registration System Certification Manager

Constant Work in Progress Customer Relation Management Canadian Standard Association Diagnose Design Change model

Design-Focused and Theory-Based Business Problem Solving methodology External adaptive

European Norm Qualified Engineer Entry Time Engineer-to-Order

International Electrotechnical Commission Internal Routine

Job Preparation Form KEMA KEUR

Key Performance Indicator

Load Oriented Manufacturing Control Lancaster University Management School Management Body

Master Production Schedule Material Requirement Planning Manufacturing Resource Planning Managed System

Make-To-Order

Opportunity Management System Planboard

Product Expert Project Manager

Paired-Cell Overlapping Loops of Cards with Authorization Production Planning and Control

Quality Management System Qualified Reviewer

(8)

VIII

List of abbreviations Abbreviation Meaning RBC SDT SO TPT TOC TRF VMC VSM WB WIP WLC WOWOS

Repeat Business Customers Sample Delivery Time Support Office Throughput time

Theory of Constraints Test Report Form

Versatile Manufacturing Companies Value Stream Mapping

Workload Balance Work in Progress Workload Control

(9)

1 | P a g e

1. GENERAL INTRODUCTION

This research aims to improve the efficiency of the standardised test- and certification process within the business clusters Appliances and Tools. First a brief introduction of the company will be provided, followed by a short overview of the value stream. Subsequently, the shop floor characteristics, the Quality Management System (QMS), and several Information Technology systems will be described. At the end of the chapter a problem introduction will be given.

1.1 Company introduction

As the Dutch electricity industry‟s Arnhem-based test house, KEMA originated in 1927. Nowadays KEMA is a global company, with more than 1800 employees, operating from 20 countries around the world, specialized in energy consulting, operational support, measurements & inspection, and testing & certification. In 2008 the company realized a turnover of 226,7 million euros of which 16,6 million euros comprised the net result1. Historically, the company was divided into two major divisions „Business and Technical Consulting & Services‟ and „Testing and Certification‟. Due to recent reorganisations this will change in short term. Dutch energy companies such as Essent Netherlands B.V. (25,3%), Eneco B.V. (31%), and N.V. Nuon (24,9%) are the major shareholders of KEMA.

The research will take place at KEMA quality, which is among others responsible for the test and certification activities. KEMA quality assesses the safety standards and functionality of numerous different electrical products, components, and systems. Although the product range is relatively diverse, they undergo a rather standardized process to test and subsequently certify products and systems. KEMA Quality is quite familiar in the Netherlands due to the issue of its own KEMA-KEUR mark. This quality mark reassures that electrical products are safe. Dating back to 1924, the system is intended as a way of showing that components and end products have passed appropriate safety tests based on international standards. Besides the KEMA Keur (KK), KEMA Quality issues numerous other safety- and functionality-certificates such as Canadian Standard Association certificate, ENEC, CB, GS, EMC, ATEX, IECEx etc.

As of the beginning of November 2009, KEMA Quality (the testing and certification division) will become part of Dekra (a German multinational). This hive off matches the overall strategy to focus it self more on its core businesses. However, at least for the upcoming two years nothing substantial will change in the way of doing business, since Dekra and KEMA quality intend to work relatively independent. The merger of KEMA and Dekra therefore have no substantial implications for this research assignment. KEMA quality Netherlands can be categorised into four major business lines, as shown in the organisation chart below.

(10)

2 | P a g e

Kema Quality Netherlands

Management system

certification Aspect certification Product certification Personnel certfication

Lights Tools

Appliances

Figure 1: Organisation chart KEMA Quality Netherlands including research scope

Three different business clusters termed Appliances, Tools, and Lights make up the business line „Products‟. Due to limitations in time primary focus will be on the business clusters “Appliances and Tools‟. The new insights and recommendation can subsequently be employed in the other cluster „Lights‟. Hence, the prime research scope consists of two business clusters termed Tools, and Appliances. Presently, the business line products encompasses 87 employees dispersed among the different clusters. The business cluster appliances is the largest cluster with a workforce of 29 people, in contrast to the small number of people working within Tools, only three. Total sales in 2007 amounts up to €11,493 million of which more than 50% is generated by the Dutch branch2. The business clusters tools and appliances are accountable for €1.887.000 and €3.254.000, respectively.

Product range within the business clusters Appliances is quite significant such as electric toothbrushes, coffee machines and toasters. Not surprisingly, the business cluster Lights provides testing facilities and certification procedures for all kinds of lighting applications.

1.2 General Value stream and relevant shop floor functions

Intending to get a familiar with the standardised test- certification process a general picture of the main processing steps is drawn (see figure 2). The process normally starts with a customer request for quotation. The iterative communication between customer and KEMA regarding quotation, clarification of order, invoice etc. is contained within the order preparation phase. This phase clarifies the kind of tests needed to fulfil the customer needs. Several different certificates can be awarded depending on the kind of product. KEMA is also authorised to perform the tests required for several foreign safety certificates3, for example a CSA certificate of the Canadian Standard Association. On the European market, CSA International is represented by KEMA. This implies that the company is able to execute

2

Presentation “Strategy Business Line products- Town Hall Meeting The Netherlands”

3 _{To get an overview of all certificates issued by KEMA the author refers to}

(11)

3 | P a g e

tests confirming the standards for the North American market and once the products display the CSA mark, they can be sold throughout the U.S. and Canada.

Each certificate is restricted to a diverse set of norms, imposed by several external bodies. The executed test-activities are exceedingly dependent of the desired certificate. For example, the restrictions for a CSA-certificate are relatively complex and time consuming, since houses in the U.S. are made of wood requiring extra fire precautions.

Customer Customer Order preparation Verification samples, information and prepayments Setup test equipment configuration Testing Tidy test equipment configuration Reporting

Sign and send certificate and

report Samples, info,and

prepayment

Figure 2: General Production Value Stream of test- and certification process

After the order preparation phase the received information, samples and payments are verified. A complete test sample with additional product specification and a manual is needed to enable performing the tests. Verification frequently results in the request of supplementary information, which delays the test process. Thus, the throughput time of the test process is reliant on the effort of the customer. Wrong information or poor quality of the samples complicates and impedes the test process. Moreover, if wrong test samples are delivered, the test process should not start at all.

Once the delivered samples and information are complete, an engineer builds up the test configuration and subsequently conducts the appropriate test. Frequently used test configurations are fixed. Others should be constructed by the engineer, consisting of different test devices. Directly after the testing, the non-fixed measuring configurations must be cleared unless it has to be rebuilt within several days and the used measuring devices are not reserved for other tests.

(12)

4 | P a g e

Apart from the stamp letters, the certificate manager checks the followed procedure and ultimately signs the certificate so the project can be finalized and sent to the customer. Although the kind of test and report activities depend on the kind of applied certificates, the overall process is rather standardized. Above and beyond, this process should be standardised so the different facilities around the world work in a similar way preserving the quality standards.

On the work floor there are five functions involved in carrying out the test processes. Those are the project manager, the engineer, support office, the certificate manager, and the product expert. A project manager should possess considerable management skills, because he guides and bears responsibility for the projects. However, since his job responsibilities include the provision of quotations, some testing knowledge should be obtained as well. The engineer conducts the test processes and has specific knowledge about the actual test processes and the products. The engineer should write the report as well, which is reviewed by a qualified colleague. Subsequently, the certification manager evaluates if the appropriate procedures are executed for issuing a certain certificate. Support office supports the test process, by printing activities, modifying lay-outs, planning etcetera. The final relevant shop floor function is the product expert, who has extensive knowledge of all different products and norms. Besides, he participates in the so called norm commissions which evaluate present norms and create new ones.

1.3 IT and Quality management System (QMS)

The different clusters at KEMA are supposed to work with a Quality Management System (QMS) which describes all quality standards and work procedures, formulated by KEMA and external authorities. Not surprisingly, in order to be authorized for issuing safety certificates a company should comply with several external norms. These companies are supervised by independent external bodies. For example the conformity to the NEN is supervised by the Dutch government, the EN norms are controlled by the International Electrotechnical Commission, and the CSA marks by among others The American National Standard Institute. The contents of these norms limit the possibilities in production and should be guarded during this research.

KEMA has introduced the QMS to make sure that the pursued procedures are according to the norms. This QMS sets a detailed prescription for all different processes and activities. Furthermore, the system contains important documents and all types of information. The QMS will be described in more detail in paragraph 3.3.2.

(13)

5 | P a g e

this connection differs between the systems. As can be seen a substantial number of systems are used.

Table 1: Index model Information Systems of KEMA

System Purpose

Opportunity Management System Quotation registration

Job Info Customer assessment

Axiant Customer Registration

WOWOS Work order information system and job registration

Planboard Planning

Persis Registration of employee qualifications

Ceres, CB/CCA numbering, Certificate registration Certification registration

InFocus Measuring business intelligence

Arnos Notification receivable accounts

Inspection Trips Inspection planning

BOB+ Sample registration

Logic LTI Calibration registration

Table 1. Information System of KEMA

1.4 Problem mess and problem selection

Based upon several meetings with management and employees a number of unsatisfactory state of affairs seems to be under discussion. Within the business line products there exists a strong feeling that the efficiency is not optimal. The result of an extensive internal audit declared that just a few employees strictly work according to QMS. Management suspects there is a rather big gap between the practical and normative execution of activities. In addition, it is expected that several time related problems occur:

bad due date performance

long and uncertain throughput time

substantial time is devoted to wasteful activities such as the search for materials and test equipment.

(14)

6 | P a g e

The above situation can be described as a problem mess which is a large and complex set of

interacting, dynamic systems of problems (Ackoff 1981). Since problem messes can not be solved, but

(15)

7 | P a g e

2. RESEARCH APPROACH

This chapter starts with the formulation of the research purpose, central research question and the research sub questions. Subsequently the preliminary cause-and-effect diagram will be described. To avoid any confusion about the suppositions regarding this research, several assumptions will be discussed too. Paragraph 2.3 will finally elucidate the research methodology.

2.1 Research Purpose, Central Research Question and Sub Questions

Based upon several meetings with management it turned out that efficiency problems caused by substantial devotion of time to wasteful activities is a significant unsatisfactory state of affairs. As a result the throughput time is rather long and uncertain. In functional problem terms, this results in among others, bad due date performance. Out of the problem mess, long throughput time is selected as the business problem.

A preliminary cause-and-effect diagram (Appendix 1) is used to identify potential causes for long throughput time. During the problem diagnosis stage the actual causes will be explored and validated, according to Van Aken et al. (2007).

2.1.1 Research Purpose

Given the problem selection, the following research purpose can be formulated:

“to decrease the average throughput time of the standardised test process”.

This will improve the efficiency whenever it is only expressed in time units. After all, one is able to execute the same activities in shorter time periods. However, it is acknowledged that efficiency is a much wider concept. Therefore, it is aimed to attain the research purpose without the need for investing in disproportional resources and money which will eventually condense efficiency.

2.1.2 Central Research Question and sub questions

Derived from the purpose, the central research question is stated as follows:

“how can the average throughput time of the standardized test- and certification process be decreased?”.

(16)

8 | P a g e

indispensable in the identification of causes for long throughput time. Therefore sub question one runs as follows:

1) How are the different activities and processes related to the standardized test- and certification process executed in practice?

KEMA possesses a quality management system which describes the normative way of executing the different processes. It is however not a strange thought that practice deviates from the normative way. Both the practical way and normative way of operating are potential sources for improvements. Hence, the normative way should be explored and a gap analysis should elucidate the differences so „suspecting‟ obstacles can be specified. Subsequently, it should be investigated why these differences arise.

2a) What does the standardized test- and certification process look like as prescribed by the QMS? 2b) What are the main differences between the normative way of testing and the practical way? 2c) What are the main reasons for such divergences?

Sub question three devotes further attention to causes for long throughput time which are experienced by the shop floor workers. Especially these variables are possible sources for improvements since the shop floor workers are most acquainted with the processes (Nicholas, 1998). However, only relying on questionnaires will not provide sufficient validity claims. Therefore, sub question three aims to find as many operational facts as possible so that validity of experienced problems is assured. Additionally, the theoretical analysis clarifies the relations of Work in Progress (WIP), workload balance and throughput time.

3) What are the relevant causes and consequences of the long throughput time?

Once the appropriate causes are identified the preliminary cause-and-effect diagram provides guidance to the propriety processes to approach the validated problem variables. The problem diagnoses phase explains that one of the most significant problems is the lopsided balance of workload. Also large amount of WIP seems to cause long throughput time. Sub question four proposes interventions to improve the throughput time by grappling large WIP amounts and unbalanced workloads. Subdivision is made between „general control intervention‟ „interventions concerning the project managers and support office‟, and interventions in order to control the workloads of the engineers.

(17)

9 | P a g e

Initial aim is to reduce the average throughput time of the test process. However, it is intended to extend the scope towards the development of a theoretical model which guides the improvement of workload balance in companies carrying the unique characteristics such as KEMA. This will be done parallel to the design phase enclosed by research question four. As will become clear during the remaining report, KEMA accommodates differentiating characteristics compared to common production companies. Through the research I observe that a literature gap exists in finding production planning and control concepts for companies facing similar characteristics as KEMA. This directly triggered me to contribute to the scientific literature.

2.1.3. Preliminary cause-and-effect diagram

Appendix 1 describes the preliminary cause-and-effect diagram, which is categorised in the elements T1-Tn that constitutes the Average Throughput Time, variables which affect the throughput time but

can not be controlled directly and processes on which the management do have a hold. The model is mounted prior to the actual field research and thus of conceptual nature. Hence, these relations do not have to be necessarily true. The average throughput time is defined as the time between order confirmation and shipment of the final test report. However, the order preparation is embraced in the research scope due to management desires. Besides, order acceptance (as part of the order preparation phase) influences the average throughput time (see chapter 4). A number of different sub times are distinguished in the preliminary cause-and-effect diagram, starting with the order preparation time in which the customer needs are discussed and translated into a customer order. Some kind of planning is probably following. Additionally, KEMA should take delivery of a test sample, expressed in the external transportation time. Then the measuring configuration should be set up so tests can be executed and the report can be drawn. Between these different stages there is undoubtedly some type of waiting time and internal movement of the employees(indicated by internal transportation time).

The second column consist s of variables which is the result of two brainstorm sessions in thinking of possible causes for the long throughput times. Together with the operational manager and business development manager seven variables were identified. The other session, performed with two shop floor workers, yielded two additional variables: „cooperation of the customer‟ and „availability and quality of test equipment‟. Each participant had to write down at least three troublesome variables independently. Subsequently, all these variables were centrally discussed which generated additional problem variables. Finally, those variables of which we expected that they were most significant were included in the preliminary cause-and-effect diagram. Although the existence of these causes can not be validated, it does serve as a good basis for further analysis.

(18)

10 | P a g e

Besides, it will lead to longer waiting times, since downstream employees might wait for upstream workers. Analogously, this will cause an unbalance of workload. The potential of rework increases if employee skills are low, the complexity of the tests are high or if the quality of test equipment is low. Furthermore, rework entails additional planning and longer testing time. Complicated tests (commonly allied with sample complexity) enhance the likelihood of longer order preparations, setup times, testing times and reporting times respectively. Sample complexity increases with technical sophistication and amount of components. In case of unavailable or low quality test equipment waiting times and setup time are generally longer and internal transport increases due to search for the right material and equipment. Moreover an engineer should reckon with the low quality of test equipment.

ICT facility refers to the presence of information systems which support the different activities related to the test process. The absence of proper information systems might increase the order preparation time, planning time, reporting time and transportation time. After all, these sub times will benefit from proper communication- and registration tools. Reporting time is reduced when information systems generate appropriate reporting formats and internal transportation is reduced when the exchange of hard copy documents is omitted. An unbalanced load of work among the employees will increase the amount of WIP (Work in Progress) as a result of queuing phenomena (Hopp en Spearman, 2000). Furthermore, unbalance of workload will cause waiting time as indicated. Besides, the planning tasks are complicated because it is more difficult to plan a particular order when workload is unbalanced or the amount of work in progress is high. According to the analogy of the river of inventory (Bertrand, Wortmann, and Wijngaard, 1998) rework and waiting times can be reduced if the amount of Work in Progress is limited. This metaphor explains that an abundance of water in a river prevents the visibility of damaging rocks. In those circumstances a boat can unflustered sail down the water, without any knowledge of the underlying dangers. On the work floor, problems can be covered in a similar way by releasing vast quantities of work in progress. As long as enough work is released to the work floor and time buffers are incorporated, potential production problems will not be discovered. Thus, problems will not be eliminated causing longer waiting times and rework.

More complex tests demand additional capacity so WIP is increased. Sample complexity (which increases with technical sophistication and amount of components) causes external transport, because physical transport is harder. After all, increased technical sophistication and a higher amount of components generally enhances the fragility. Therefore it is more difficult to transport the sample and possibly additional steps for protecting the sample are required. Cooperation of the customer determines to some extent the external transportation time and waiting time. Low effort resulting in among others longer external transport time will force KEMA to wait.

(19)

11 | P a g e

improved as it is still blurred how to control the relevant variables. Therefore, eight important processes or policies were selected by means of literature, so that detrimental variables can be managed.

Human resource management can be defined as the strategic and coherent approach to management

an organization's most valued assets - the people working there who individually and collectively contribute to the achievement of the objectives of the business (Armstrong, 2006). Evidently, this

management philosophy consists of numerous different processes. Special attention is provided to the „recruitment and training process‟ and the „quality programme‟ since these are essential in the assurance of propriety qualifications within KEMA. The recruitment and training programme should ensure that engineers gain enough experience so that qualifications are acquired conforming the standards of accreditations. Needless to say, an insufficient range and level of qualifications will strongly restrict KEMA in the scope of work. The quality programme guards that an adequate quality level is maintained. For example, Total Quality Management refers to an integrated approach by

management to focus all functions and levels of an organization on quality and continuous improvement (Nicholas, 1998). The customer perspectives, both internal and external, should take a

central position in determining the appropriate quality level. This requires total participation and commitment company wide (Nicholos, 1998). Thus, the quality programme determines to some extent the internal work procedures and the amount of satisfying qualifications. The interrelations between HRM, recruitment and training, and the quality programme is illustrated by the bilateral arrows. It is assumed that the numerous other processes related to human resources (administration, reward policy, time management, etc.) are covered by human resource management.

The quantity of rework is mainly influenced by other variables. Logically, inappropriate employee skills and tests that are more complex will increase the potential of rework. Also low quality test equipment will enhance the chance of rework, due to more defects and equipment brake downs. A non optimal work prescription by the quality programme can cause rework as well.

The available range of qualification, the sample complexity and the sales policy influences test complexity. The sales process is responsible for the selling of projects in order to attain one‟s commercial ends. The level of quality is part of the offered value proposition, so the quality programme influences the sales policy. During the sales process, the test and sample complexity is determined.

These days, the purchasing activity is defined as selecting those suppliers and dealing with them in

such a way, that it enables the organization to implement its market strategy best in the way it has decided to serve the end customers‟ needs (Kamann 2007). The purchasing function thus fulfils among

(20)

12 | P a g e

Furthermore, there should be processes that aim to warrant a proper ICT infrastructure. According to Piccoli (2007) information systems are formal, socio-technical, organisational systems designed to

collect, process, store, and distribute information. Policies regarding these information systems are

dependent on the interplay of technology with other aspects, such as structure, people and processes. These policies, influenced by the quality programme, should express what procurements should be made. The purchasing process performs the actual purchase.

Production planning and control (PPC) exists of two sub definitions. Production control, including material management, primarily focuses on the control of production quantities along time (Bertrand, Wortmann, and Wijngaard, 1998). This production control and material management are part of logistical control which is defined as “the supply and discharge of material and capacity sources for the benefit of the transformation process (Bertrand et al, 1998)”. Planning refers to the allocation and

coordination of different capacity sources among production departments. Obviously, production planning and control is significant in the creation of workload balance and the release of work in progress. Furthermore, by stipulating the work procedures the quality programme is important too. Illogical work procedures will, after all, harm the balance of workload due to „starving‟ and „blocking‟. Customer Relation Management (CRM) is a term for the methodologies, technologies, and

e-commerce capabilities used by firms to manage customer relationships (Jobber, D., 2004 ). CRM is

more than simply the technology. It also concerns the processes and deliverance of customer propositions. Therefore, it is partially influenced by the quality programme and the available ICT- facilities. Also PPC systems are crucial tools for meeting increasingly high customer demand (Stevenson, Hendry, Kingsman, 2005) and therefore influence the CRM. Typical functions of PPC are for instance the determination of level of flexibility or throughput time control.

(21)

13 | P a g e

Output Black box

Human Resource Policy

Input

Average throughputtime Recruitment & Training proces

Quality Programme Purchase Policy

Information Technology Policy Production Planning and Control Policy Sales Policy

Customer Relation Management (system)

Figure 3: Simplified conceptual model of preliminary cause-and-effect diagram

2.2 Research assumption

According to among others Walliman (2001) eminent academics disagree on the fundamental aspect of science. One movement argues that science should be seen as a social construct. In contrast, others take a more traditional point of stance and claim that to reveal the truth that is buried in the nature‟s complexity, one should use a scientific approach to stimulate theoretical development refined and inspired by experimentation (Walliman, 2001). Since these kind of assumptions are fundamental in the development of proper research, clarity is given.

First, the assumption about the very essence of phenomena under investigation, called ontological nature. The author of this report perceives the world as external to the individual cognition, made up of hard, tangible and relatively immutable structures, according to the realism (Burell and Morgan ,1979) .

The second assumption is about the epistemological debate, which tells something about the grounds of knowledge. According to the author the nature of knowledge can be seen as hard, real and capable of being transmitters in tangible form. Consequently, the stance of the „positivist‟ (Burell and Morgan, 1979) is assumed.

Another vital debate concerns the methodology with the two mainstreams, ideographic and nomothetic. According to Burell and Morgan (1979), the ideographic theory explains methodology as getting close to the subject under investigation with a detailed exploration of background and history. “The

idiographic approach emphasises the analysis of the subjective accounts which one generates by „getting inside‟ situations and involving oneself in everyday flow of life (Burell and Morgan, 1979)” On

(22)

14 | P a g e

In sum, taken into account the above assumption the objectivist notion (Burell and Morgan, 1979) is occupied. Linking this to the work of Walliman means that the author does not believe in science as a social construct. Reversely, the scientific method is preferred to develop and apply theoretical meanings.

Furthermore, the instrumental perspective of organisations is assumed, which sees management as purposeful control of processes, with the organisation as a mean (de Leeuw, 2002). Analogous to the objectivist assumption, this research uses a hard system approach corresponding to the first order cybernetics that presumes univocal observation independent of the observer (de Leeuw P. 89, 2002). The problem situation is regarded as predictable and independent of the researcher. However, it is acknowledged that the history of the researcher or observer is influential. Pluriform investigation will improve the objectivity as more respondents with different backgrounds is more reliable than an equal quantity of respondents with similar backgrounds. Therefore, pluriformity is a central condition to this work. Consequently, the list of interviewee‟s consists of people with different functional responsibilities and backgrounds. On the work floor almost all engineers, project managers and certification managers from the business cluster appliances and tools are interviewed. Furthermore representatives of the departments quality and legal; support office; archive; inventory management; sales and account management, business development and marketing; finance; facilities; information technology and human resource will be interviewed. Semi-structured questionnaires for al different departments and functionalities will be used, in view of different subjects of expertise.

2.3 Methodology

2.3.1 Methodological View

Arbnor and Bjerke (1997) distinguish four methodological views, the analytical approach, the system approach, the actors approach and the relative view. The creation of knowledge and their methodological roots is central point of the book. A clear foundation of knowledge creation and about doing research efficiently and effectively is provided. Due to the preferred hard system approach, the actor approach is irrelevant. Both the system approach as the analytical approach perceives the reality as consisting of fact-filled system structures in the objective reality (Arbnor and Bjerke, 1997). The difference is however, that the system approach acknowledges subjective opinions as well and treats them as facts. Given that the author should make an appeal on the knowledge and opinions of the shop floor workers, the system approach appears the most suitable methodological view. The reality is not summative in the system approach and scientific ideal is to find these wholes and patterns as

objective structures and try to make every new system picture better than the last, partly as most valid system structures, partly as pragmatically more favourable concepts (Bjorne and Bjerke, 1997). Since

the methodological view encompasses a rather broad perspective „the general system approach‟ suites best in this matter. A striking definition stated by Arbnor and Bjerke (1997) runs as follows: “There exist models, principles, and laws that apply to generalized systems or their subclasses,

(23)

15 | P a g e

“forces” between them. It seems legitimate to ask for a theory, not of systems of a more or less special kind, but of universal principles applying to systems in general (von Bertalanffy, 1955, P.77)”. This

rather broad perspective encompasses several system movements which will be briefly discussed during the reflection.

The system approach generally requires extensive interviews and questionnaire activities (Arbnor and Bjerke, 1997). Furthermore, the creator of knowledge should look for secondary data and the systems history.

2.3.2 Procedural rationality

De Leeuw, (2002) defines procedural rationality as the way of how to make a decision, methodologically. The kind of business problem selected form the problem mess is of „design focused‟ nature. It is expected this research comes up with an intervention or redesign of certain processes in order to decrease the average throughput time. Two typical design focus methodologies are the Design Diagnose Change (DDC) model of De Leeuw (2002) and the Design-Focused and Theory-Based Business Problem Solving methodology (DFTB BPS) of van Aken et. Al (2007). The latter methodology follows the logic of a regulative cycle consisting of 1) the problem mess, 2) the problem definition, 3) analysis and diagnosis, 4) formulation of an action plan, 5) intervention, 6) and the evaluation that enters the problem mess again. Due to limitations in time, step 5 and 6 are beyond the scope of this research. The remaining four steps are analogous to the DDC model. Therefore to characterize the procedural rationality in the research context the DDC (diagnose, design, and change) model is used. De Leeuw (2002) and van Aken et. Al (2007) emphasizes the employment of pluriform diagnosis and the essence of a proper problem diagnosis, corresponding to the research assumption.

The left side of figure 4 depicts the different stages of the DDC model. The upper left square explicates three different analyses. According to van Aken et al. (2007) these three approaches should be combined to produce a comprehensive diagnosis. Important note is the presence of the preliminary cause-and-effect diagram, which serves as input for the three diagnoses. The naming and framing of this diagram aids at selecting the right causes for long throughput time, which will be subject to interventions. This approach is different from theory forming strategies such as the grounded theory approach (Strauss and Corbin, 1998), which aims to build theory from scrap out of raw qualitative data, without explicit reference models. In contrast, we are only interested in the formation of a local theory including relevant problems, causes and possible consequences derived from the preliminary cause-and-effect diagram.

The use of multiple sources of data collection enhances reliability and validation (van Aken et al, 2007). “The least reliable option is to exclusively build upon the opinions of organization members as gathered

(24)

16 | P a g e

frequently rely on such interviews. Hence, we argue that the exclusive utilization of the grounded theory approach is not sufficient in this research. Nevertheless, the three diagnosing analyses contain several similarities with the coding strategies of Grounded Theory. According to Straus and Corbin (1998), the notion of conditions, actions/interactions, and consequences is essential in (local) theory building. This corresponds to our preliminary cause-and-effect diagram, which is divided into causes, problem variables and consequences for throughput time. The process-oriented analysis investigates the conditional factors (causes) for the problem variables. It thereby assures the business focus by using the principles of Value Stream Mapping (VSM), represented by the clouds in column 2. Both QMS as the current value stream are sources to identify what is going on.

Throughout the empirical analysis possible causes of long throughput time, experienced by the shop floor workers, should be further explored and validated. This corresponds to axial coding which refines the interplay of conditions, actions/interactions and consequences. However, prior to the validation, one should identify which factors of the preliminary cause-and-effect diagram are experienced as problematic by the shop floor workers. In order to fulfil this analytically, the process of open coding (also part of grounded theory approach) and Analytic Hierarchy Process is used. The theoretical analysis will reinforce these validation asserts by refining the local theory. A mingle focus on both the cause-and-effect diagram and the actual business process will contribute to exploring and validating the causes. D ia g n o s is D e s ig n C h a n g e Problem situation Identified problem(s) Solution Improved situation QMS value stream Current value stream Gaps analysis

Qualitative data gathering: questionnaires (semi-standard) Observation

Open Coding (Grounded Theory Approach) Analytic Hierarchy Process Secondary data gathering Value Stream Mapping Flow charts

Statistical anaylsis: F-test, T-test

Queuing theory Production Planning and Control

Methods/Tools/Literature Reticulation (level 1)

Literature research: Production Planning and Control

Cybernetics Workload control Organisational design Preliminary cause-and-effect diagram

Express and validate business problem

Typfication of diagnosis

Empirical analysis

Theoritical analysis Process oriented analysis

Reticulation (level 2)

Secondary data gathering. Qualitative data gathering: questionnaires

Design of improved system model

Beyond the research scope Engineers

PM and SO Predictability

(25)

17 | P a g e

The process-oriented analysis starts with a description of the practical way of operating, followed by the elaboration of the quality management system. Semi-structured questionnaires will be used so that the explorative nature of the problem diagnosis phase is preserved, while the researcher dependency is limited (van Aken et al., 2007). So, a list of specific questions is designed, leaving sufficient room for additional information. Subjective interpretations are treated as objective facts within the system view (Arbnor and Bjerke, 1997) and hence during this research. Concerning the questionnaires, a useful

method of getting information from a sample of the population that you think knows most about a subject is theoretical sampling (Walliman, 2001). Moreover, reflecting upon the assumption of

pluriformity, interviewee‟s from different functional areas should be selected. Of course, the functions situated on the work floor have substantial knowledge of the operational processes. Justifying this importance, a sample size of nearly 100% is taken among the shop floor workers. Furthermore, all back office departments will be covered by at least one respondent. The high variety of respondents (Operations, Quality and Legal, Sales, Finance, ICT, Business Development, Archiving, Sample management, and Facilities) and relatively large sample size of work floor population attempts to deal with „respondent unreliability” (van Aken et al. 2007). The questionnaires will be taken on different moments in time, concerning reliability matters dependent on particular circumstances. The questionnaires are elaborated in a genuinely anonymous way by coding all names of the respondents.

To ascertain that all processes are captured, the questionnaires are drawn up analogous to the QMS. Each step in the QMS is evaluated along practical experiences of the employees. Apart from that, the questions have been discussed with the operational manager to assure completeness. The questionnaires should give an indication of possible problem areas which cause long and uncertain throughput times. If more than 10 respondents reveal similar problem variables (based on open coding), further analyses have to validate the problems and relations with throughput sub times. To ascertain that the correct problem variables for further analysis are selected, Analytic Hierarchy Process (AHP) is applied. The operational manager fills in a two questionnaires making pair-wise comparisons. The first one determines to what extent the sub-throughput times leverage the total throughput time. Second questionnaire identifies the influence of the problem variables to the different sub-throughput times. By combining these two pair-wise comparisons one is able to select the problem variables which are most significant in impeding the total throughput time.

(26)

18 | P a g e

process. On the other hand, this flow chart is used to trigger suggestions of the shop floor workers by hanging the chart on the wall where remarks can be written down. (Participative) Observation is employed too since, to some extent I became a member of the organisation (van Aken et al., 2007). Consequently I walked frequently through the test lab in order to note conspicuities, and attended workshops and cluster meetings.

Based on the questionnaires and observation, the current value stream will be designed at which the principles of Value Stream Mapping (VSM) are particularly helpful. VSM “allows a company to

document, measure, and analyze a complex set of relationships as well as plot a course to create an improved operating strategy and organizational design” (Keyte and Locher 2004, p. 2). In addition

Lasa, de Castro, and Laburu (2009) claim that VSM is a practical tool in productive environments on academic and theoretical level. VSM is also suitable, taking into account the methodological view as the idea is to gradually move towards an improved future value stream. Besides, it focuses on the whole value stream in stead of individual work stations and aims to reduce wasteful activities. The output of the diagnosis phase will consist of a clear functional problem divided into instrumental problems or causes.

(27)

19 | P a g e

3. PROBLEM DIAGNOSIS

Starting point of a proper diagnosis is a validity check of the selected business problem and a fully fledged, integral explanation of the problem (van Aken et al, 2007). Once the validity of the problem has been recognized, the kind of diagnosis can be presented. The definite diagnosis starts with a process oriented analysis where the researcher digs into the actual processes and the „normative‟ Quality Management System using the first two steps of Value Stream mapping. Aim of the process oriented analysis is to get acquainted with the different processes and to generate points of special interest. The problem variables from the preliminary cause-and-effect diagram which are experienced by more than at least 10 respondents and which are indicated as significant by AHP, are subject to further analysis assuring validity. Even though the different analyses are described sequentially, the actual project is performed iteratively.

3.1 Validation of the business problem

Throughput time is defined as the average time between order confirmation and the shipment of the final test report. Concerning the throughput (sub) times no data is stored, nor are acceptable norms formulated. The available operational data are obviously relatively limited. However, there are activity lists of the project manager obtainable, which record all activities, including start date and agreed delivery date. By exploring the time a job number is withdrawn from the list, one can identify the throughput time of this project. 30 projects were examined accordingly, as presented in appendix 7. The 13 projects still running are up to 143 weeks in progress with an average of 53 weeks. The remaining 17 projects encompass an average throughput time of 30 weeks. Only one single project was delivered before the agreed delivery date.

WOWOS is an information system that records, among others, all hours devoted to a project. Each week an employee has to account for the hours spent. Since the employees are judged on the allocation of these hours, reliability is strictly monitored. By means of WOWOS it was feasible to compare the actual hours spent on a project with the throughput time. It appeared that among the finished projects, on average only 11% of throughput time were considered to be effective hours, while this was even lower (9,7%) when ongoing projects were included. The remaining time consisted of relatively wasteful activities as waiting and transportation time.

(28)

20 | P a g e

the questionnaires) , the product is not conform the standards and hence is disapproved. Then, the customer often makes the necessary modification and additional tests are needed. This extends the throughput time and consequently the initial due date is exceeded, without adjustments having been made. This high dependency of the customer will receive further attention during the remaining report.

Albeit throughput time norms are not available, based on the above numbers it is concluded that long throughput time is a real problem. Real problems are defined as problems where one should find the solutions in changing the reality (de Leeuw, 2002).

3.2 Kind of diagnosis

De Leeuw (2002) addresses four important attributes to characterise a certain diagnose. Since a business problem is already selected from the problem mess this diagnosis encompasses a predefined problem indication.

Aim is to identify which causes or instrumental problems influence the throughput time and to what extent. Therefore a broad diagnosis concept is used. Furthermore the preliminary cause-and-effect diagram serves as an explicit reference model, since it is used to track relevant causes. The diagram presented prior to the empirical analysis is discussed with management, to assure that the important causal relations are further analysed. Furthermore, the QMS would also be a helpful reference model, since it explains how to execute the activities in order to accomplish the company goals. Therefore the QMS will be compared with the practical way of testing.

Due to the assumption of pluriformity an integral diagnosis suites best.

3.3 Process oriented analysis

3.3.1 Current Value Stream

(29)

21 | P a g e

Lasa et al. (2009) list several main arguments for the potential of VSM, which are acknowledged by the author of this report too:

The use of graphical interface simplifies the evaluation of material and information flows. The systemic vision reflects manufacturing system inefficiencies.

It serves as a common language.

The possibility of being the starting point of a strategic improvement plan.

Although VSM is frequently used within the concept of Lean implementation, the author argues, with regard to the above reason, that it should not go hand in hand with lean implementation per se. Aim in this context is to analyse the practical way of testing in order to compare it with the normative way. Subsequently possible bottlenecks can be discovered.

In contrast to pure production companies where capacity is determined by workstation or production equipment, KEMA‟s capacity depends on human actors. Planning is based on human capacity and work flows from one function or actor to another. Consequently, the different functions take a central position in the representation of the current values stream. During the design of the diagram a number of areas for improvements were noticed, signified by small red and orange stars. The value stream may give the impression that only the engineer executes the tests. In practice, however, a project manager performs tests as well.

(30)

22 | P a g e

After the order confirmation, a project manager must fill in a job preparation form (JPF) containing the customer information, the kind of services and tests, sample information, price, how many hours are needed to perform the test, amount of hours needed for project management, which engineer could do the test, and what is the starting date and agreed deadline of the jobs. Based on a JPF the support office books an order in Axiant, which is the customer relation system. Due to its connection with WOWOS (workorder information system) the information implemented in Axiant is automatically copied to WOWOS. In WOWOS one is able to find all information concerning a particular order. The support office makes a copy of the JPF and hands this over to the planner. The original JPF is placed in the project file together with all documents related to correspondence and is passed trough to the project manager. Furthermore, a work-in-progress map is created on the J directory in the right client map. During the test process an engineer should use this map to save all the documents and test results.

One person is responsible for the planning of the business cluster. The planner verifies whether the proposed engineer by the project manager (indicated on the JPF) has enough capacity to perform the tests and possess the necessary qualifications. In practice only on occasion the planner deviates from this advice after consideration with the project manager. All projects are planned in the plan-tool on the basis of the JPF. The amount of calculated hours is allocated among the project manager and an appropriate engineer. A project manager thus declares his own hours for a certain job. This evidently increases the option of estimating superfluous hours for the project manager. Subsequently, the hours are adopted and planned into Planboard. Since the hours are not specified up to activity level, Planboard only represents rather general descriptions of tasks together with the start/end- date and quantity of hours needed. Hours are attributed to employees in weeks by dividing total hours needed by available weeks. For example, a project requiring 20 hours of an engineer finishing in 4 weeks, attributes 5 extra hours a week for this particular engineer. Still, an engineer might fulfil the 20 hours job in the first week. Apparently, the planning is not prescribing what to do at what time. Only a general representation of expected workload among engineers and project manager is given. Hence we should speak of workload allocation instead off planning. The project manager remains responsible for reaching the deadline and fulfilling the calculated hours

Plan board: hours / delivery time JPF: Estimated hour allocation among Eng. and PM Delivery time Required qualifications Degree of capacity utilization per Eng./PM Activity list per Eng./PM

Figure 5: schematic representation of ‘planning’.

(31)

23 | P a g e

takes more time in comparison with the Planboard. For one thing this has to do with irregular update meetings with the planner, on average once every two weeks. For another, shop floor workers admitted that they not fully inform the planner, due to a lack of faith in the system and the conviction that planning is only a labour intensive activity without any advantages. Complaints were made about the overview and understandability of Planboard. One week many hours are negative, while the next week suddenly everything is solved.

Planboard is also used for the planning of scarce test equipment and the record of worked hours by the shop floor workers. At the end of each week an employee has to justify his worked hours through a week report.

Once the project is planned, the engineer receives the project file including all relevant documents. Because quite a few engineers or technicians are under instruction, various project managers make a detailed test plan. However, there is no standard in constructing test plans, causing different procedures among different project managers. This is obviously not beneficial for the standardisation and clarity of the test processes. Nevertheless, the engineer discusses with the project manager how and which tests should be executed. Subsequently, the engineer collects the samples by means of a BOB form. In the central warehouse (entitled as BOB, Beproevings Objecten Beheer) all test samples are collected, unwrapped and booked in a system called BOB+. A BOB form is printed and distributed to the support office which puts it either in the project file or gives it directly to the engineer. The test sample, including a manual and product specification documents is analysed and an initial check on completeness is performed. According to the respondents in at least 15% of the orders, the engineer or technician discovers incompleteness in a later stadium. Thereupon, the customer is contacted by the engineer or project manager to demand for the absent documents or products. It seemed that due to extensive time pressures this initial sample check is quickly performed without going into detail. Furthermore, the customer is often too late in delivering the samples, although no formal delivery dates are agreed. It frequently occurs that a project can not start at the planned date, because the sample is not yet present.

(32)

24 | P a g e

be executed in accordance with the project manager. Common activities in the physical test procedure are:

The evaluation of the manual.

Pictures are made of the products and the different parts. The different parts are named.

Thermocouples are made. The wiring circuit had to be made.

The physical test execution which generates the measures dependent of the product and kind of test.

Certain engineers are already reviewed by their project manager directly after they executed the tests. Again this approach of interim reviews differs between the project managers. When tests are completed, the test report is sent (both manual as electronically) to the support office, where the lay out is modified, the completeness of the project file is guarded by means of a checklist and the documents and certificates are printed. Each certificate is registered and produced by a different information system. The use of different certification registration systems, sometimes for the same clients, all generating a different lay-out is clearly rather inefficient. Besides, the systems are reasonably obsolete according to the respondents.

The printed documents together with the certificates are returned to the engineer, who makes sure that everything is complete. Subsequently, the documents are reviewed by a qualified person, selected by the engineer. A reviewer evaluates the validity and reliability of the conducted test process. As there is no formal planning tool, the review of the project file frequently has to wait until the project manager makes time. As soon as the engineer assimilated all the remarks the report goes back to the support office which again prints the changed pages and returns it to the engineer so he can sign the report. At this moment the certification manager will be involved. An engineer is able to select his own certification manager. Numerous respondents notified that there is difference in quality among the certification managers. Because an engineer wants his work to be approved, the less strict certification managers are selected.

Throughput time reduction of the Test and Certification Process.