Improving on-time performance of the composite parts production for ProductX

(1)

All rights reserved. Reproduction or disclosure to third parties of this document or any part thereof, or the use of any information contained therein for purpose other than provided for by this document, is not permitted without prior and express written permission of CX

Improving on-time performance of

the composite parts production for

ProductX

Master thesis Technology Management

By Peter Krook

S1831208

p.j.krook@student.rug.nl

December 2010; Public version

University of Groningen

Supervisor: Prof.dr.ir. J. Slomp Co-assessor: Dr.ir. D.J. van der Zee

(2)

Public Master thesis - Peter Krook II

Preface

In front of you is lying my final thesis written in occasion of the regular diagnosis research project for the Technology Management masters degree at the University of Groningen. I conducted this research at CompanyX (CX) in CityX. CX is a company specialized in designing, developing and producing complex light-weight aerostructures for aerospace and defence industry. During an internship of eight months, I carried out an investigation into the area of on-time performance of composite parts intended for ProgramX ProductX (PX). Without the help of many people this research project would not have had the current quality. Therefore I would like to take this opportunity to show my appreciation to those who helped me to finish this research successfully.

First, I would like to thank all the people from CX who helped with this research. Special thanks to X for his guidance and the opportunity he gave me to perform this research. Moreover, I would like to thank X for his guidance and the possibility to participate in ‘his’ MES PoC team (appendix 1). I would like to thank X for his inspiration and his contribution by sparring with me.

Second, at the university I would like to show my appreciation to my supervisor prof.dr.ir. Jannes Slomp for his guidance, motivation and constructive feedback. Subsequently, I would like to show similar appreciation to my co-assessor dr.ir. Durk Jouke van der Zee for helping to bring this research to a successful ending.

Finally, I am graceful to Jan-Willem Krook MSc and drs.ing. Frank Beverdam for contribution to the quality of this report by their critical view and valuable feedback.

CityX, december 2010

(3)

Public Master thesis - Peter Krook III

Management summary

This research is done for the ProgramX program-management of CompanyX (CX). The research is performed within CX’s composite parts production facility which is characterized by low volume and high variety of customized parts. These parts have different routings with multiple directions; however some kind of dominant order flow is signalled. Therefore this facility can be characterised as a general job shop. The facility supplies to assembly departments of different programs, which deliver aerostructures for specific customer end products based on long-term make-to-order contracts.

The motive for this research has been the X% on-time performance of the PROGRAMX ProductX (PX) assembly department. This department gets PX composite parts supplied from the above described composite parts production facility. The X% is not matching with CX’s operational excellence strategy. This on-time performance will even decrease when the expected increase in production rates and product variations become reality. Due to urgency of the problem, several initiatives have been initiated to increase the on-time performance of X%. One of these initiatives has been this research, which investigated the on-time performance of the PX composite parts production. Their on-time performance was calculated on 53,7%. Improving the on-time performance of the PX composite parts production will contribute to an increasing on-time performance of the PX assembly. The reason for this is the ability to start on-time at the PX assembly department. A quantitative problem analysis leaded towards the understanding that the research had to focus at cycle time variability of the non-destructive-inspection- (NDI), milling- and measure department. Cycle time variability concerns the deviation between the time parts spend as work-in-progress and the time allotted for production of a part on that routing. Therefore decreasing cycle time variability at the introduced departments should improve the on-time performance of PX composite parts by increasing lead time reliability.

(4)

Public Master thesis - Peter Krook IV

3) conflicting performance evaluation. These categories are causing cycle time variability in two ways. First, inappropriate control measures are used for scheduling capacities and dispatching priorities of especially the bottleneck. Second, an insufficient information base for decision making is set by current insight in capacities, planned start dates and performance.

The designed solutions for eliminating the above discussed causes will be briefly introduced. First, the inappropriate control measures should be solved by controlling workload input and -output and using the first-in-first-out (FIFO) priority dispatching rule. Workload control should match capacities for the entire chain of workstations at the moment of order release. As a result, the milling department will be treated like a bottleneck. Consistent use of the FIFO rule at the debag-, NDI-, milling- and measure department should eliminate the switching order sequences. Concluding, workload control and FIFO should provide the control measures to realize an increasing on-time performance.

Second, the insufficient information base should be solved by supporting the planning and control system with a shop floor control (SFC) system and by setting the right key performance indicators (KPI’s). The SFC system should provide the required information about orders and system status for using workload control and FIFO. The KPI’s should systematically trigger management’s perspective for focus of action towards improving on-time performance of the entire chain. Therefore all levels of the PX composite parts production and elements per level should be evaluated consistently, equally and at least at lateness and workload related KPI’s. Concluding, a SFC system and the right set of KPI’s should provide the information base to realize an increasing on-time performance.

The implementation of the designed solutions could improve the on-time performance of the PX composite parts production by 27,3%. Moreover, the capital used for work-in-progress can decrease with 1.250.000 euro. Next to this, a cost advantage of 1 full-time-equivalent for planning is expected. The cycle time can decrease with about 8 days. These results can only be achieved when the following requirements will be met:

- Sufficient commitment is present. - Workload norms are determined

appropriate.

- Quality and reliability are improved. - Pre-calculated hours are adjusted and

(5)

Public Master thesis - Peter Krook V

1. Introduction

This report elaborates on a research performed for the ProgramX program-management of CompanyX (CX). The motive for this research is the urgent delivery performance problem of the PROGRAMX ProductX (PX). The PX assembly department delivers X% of the PX-orders on-time. This X% is measured by the management information system (Hermes) within the period of August 2009 till June 2010. The PX delivery performance is considered to be a problem due to the target of 98% on-time performed orders. The customer is complaining about this performance. Besides, CX got a penalty clause of X% price reduction for not satisfying all PX key customer values of delivery performance, quality and affordability. Moreover, the low delivery performance is not matching CX’s operational excellence strategy, which strives for delivering products at competitive prices and with minimal inconvenience. Concluding, this research is performed due to a mismatch between the X% delivery performance and CX’s strategy for satisfying customers.

The urgency of the delivery performance problem can be explained by considering the program master schedule till 20XX. This document shows that the amount of PX’s is going to increase approximately X times. Moreover, the total amount of PX types will grow with X%. These higher production rates and product variations will cause escalation of the problem, if the production system which has to deal with them is going to be designed like the current system. Due to the urgency of the problem several initiatives are ongoing for increasing the PX delivery performance. One of these initiatives is this research which investigates the delivery performance of the PX composite parts. The reason for this is that late delivered PX composite parts cause the inability to start on-time at the PX assembly department which processes the composite parts into PX’s. Hence, this research elaborates on delivery performance of the PX composite parts production in order to improve the ability to start on-time at the PX assembly department.

(8)

2. Problem context

This chapter elaborates on the PX composite part production to understand the system which has the delivery performance problem. The first paragraph discusses the products produced within the system. Then the composite part production process is discussed in order to understand the way the products are manufactured. Finally, the strategy and organization of the composite part production will be discussed for creating general understanding about the organizational objective and structural design. These topics form the required knowledge basis for understanding the research environment.

2.1 Products

In order to understand the output of the composite parts production this paragraph elaborates on products which this system manufactures. The composite parts production process produces composite. This material is built out of different component layers. Usually the term composite refers to fibre reinforced synthetic material. The composite products at CX are built of laminated prepreg material. Prepreg material is a fibrous mat of thermoset synthetic material impregnated with resin. Prepreg material characteristics are changing negatively by room temperature. This explains the importance of controlling the time which rolls and plies spend within room temperature, called the ‘open time’. Sometimes honeycomb and syncore foam are parts used to improve stiffness of the composite. However, composite parts are mainly built of prepreg material.

Understanding the output means also understanding the use of output. The composite parts are used by assembly departments of different programs, which process them in complex light-weight aerostructures for aerospace and defence industry. Programs are projects to deliver parts for a specific customer end product with an average duration of 18 years. At the moment the company has contracts for X programs in five business lines on a make-to-order basis. Stevenson et al. (2005) defines different types of make-to-order sectors. CX can be assigned to the Repeat Business Customizer (RBC) sector which provides customized products on a continuous basis over the length of a contract. Goods are customized, however made more than once permitting a small degree of predictability. Concluding, CX can be defined as RBC of aerostructures.

(9)

composite parts production has to deliver even a wider variety of composite parts. In total CX has X active composite part article numbers for all programs. Currently the PX assembly department demands already X composite part types. Next year, when another PROGRAMX type will be produced, the PX assembly department demands X part types more. Then the system needs to produce already X composite part types with varying characteristics and configurations for the PX’s. Therefore, the composite parts production is characterised by low volumes and high variety. Concluding, the system is repetitively producing low volumes of highly varying customized composite parts for aerostructures based on long-term make-to-order contracts.

2.2 Process

This paragraph creates an understanding of the several steps needed to manufacture composite parts. Every individual part has its own production routing, because every part requires its own specified compilation of sequential production steps for realization. All X routing variants are analyzed and shown in the ‘from-to-diagram’ of figure 1. This diagram visualizes the frequency of movement from one operation to another by numbers and the thickness of the arrows between operations. Besides, every operation is numbered between brackets. Concluding, figure 1 visualizes the production process for PX composite parts.

(10)

Figure 1 can be explained in the following way. First, the supplied prepreg material is pre-cut (1) into plies. These plies are going to the lay-up department (2) where they will be placed on the mould. When the mould is bagged and the vacuum is installed, the mould is ready to run an optional charge in the oven (5) or autoclave (3) for hot-debulk or bonding. After the optional charge the product is going back towards the lay-up department (2) for placing the next layers. Then the product is running an autoclave charge (3) for polymerization of the resin in order to connect the thermoset synthetic fibres. After the charge the product goes to the debag department (4). The bag will be removed from the mould and the product will be unloaded from the mould. After debagging the frequent optional steps are the oven (5) and measure (10) operations. The last standard step is that the products go towards the non-destructive inspection (NDI). NDI (6) inspects the inner of laminated products for porosity and inclusion. Concluding, the standard production steps are: pre-cut, lay-up, autoclave, debag, oven and NDI.

After the NDI (6) step the product has two options at the assembly department (7): - Packing the product to transport it to CityY

- Assembling the product on the milling fixture

Within CityY the product has the optional step of X-ray (8) in order to guarantee the inner quality of the product. At the milling department (9) the contour of the product is modified for realizing a net composite part instead of gross composite part at on of the following machines: Fooke, Belotti and Jobs. After the milling operations the product is going always towards a 3D measure machine (10) which will measure geometrical characteristics. Finally, after the milling operation the NDI department (6) will check the product for delamintation, which can be defined as composite layers which loose. The planning and control elements of figure 1 will be deepened out within paragraph 4.1. A more detailed description per operation can be found within appendix 2. In order to get more insight in the process using composite parts, appendix 5 elaborates on the assembly process. To conclude, after NDI the optional step of milling is followed by a measure and another NDI operation.

(11)

When analyzing figure 1 in combination with the physical process flow (figure 2) the following conclusion can be drawn up. First, the composite parts production has a kind of functional lay-out design. Second, the composite parts production process has high routing variability with some kind of dominant flow. General job shops are characterised by multi-directional routings with a dominant flow direction, unlike the general flow shop which consists of work travelling one direction (Stevenson et al., 2005). Therefore, the conclusion can be drawn that the composite parts production process is a general job shop.

2.3 Strategy & organization

The organizational strategy and structure of the organization gives insight in what the organisation strives for and how it is designed for fulfilling the strategic objective. First, based on the CX Annual report (CX, 2009) and the production strategy (CX DEVISION, 2010) can be concluded that CX strives for the technology leadership- and the operational excellence strategy (Treacy and Wiersema, 1993). More specific, the PROGRAMX production strategy focuses at quality, affordability and delivery performance, which are the key value and performance drivers of the customers (Y et al., 2006). Thus, the organization and especially the ProgramX have delivery performance as important objective.

(12)

Second, the organizational structure is discussed. When considering the organizational configurations of Mintzberg (1991), CX can be defined as machine organisation. Supporting- and technical staff departments are important within machine organisations, because these formalize behaviour by standardisation of rules, procedures and function descriptions. The organization is structured as a matrix-staff-organisation (Jones, 2007) with the functional and program dimensions like figure 3.

The operations and supply chain management (SCM) organisation is elaborated in more detail by figure 4, because this research is done within the operations & SCM organisation. The vice president (VP) operations and SCM reports to the President of CX. The departments under the VP are divided as follows:

- Operations of the location CityY - Operations of the location CItyX

- Overall Purchasing - Overall Logistics

Figure 4: Operations & SCM organization structure

The research is performed within the production department of the location CityX and overall logistics department. Therefore these departments are deepened out. First, the location CityX has a production director who is responsible for the entire production facility.

(13)

This production director is supported by the staff production engineering department. The department is responsible for delivering employees who have knowledge of production preparation and -processes. The production facility of CityX is split up into four production responsibility areas:

- Composite parts - Machining parts

- Assembly of the defence business line products - Assembly of the civil business line products

Under these operations managers all team leaders per department are positioned. Most of the operations discussed in paragraph 2.2 are controlled by the composite parts operations manager. The measure operation is covered by the assembly defence operations manager. Concluding, research is done mainly within the composite parts responsibility area. Second, the director logistics is responsible for the planning and material handling of the CX organisation. Both responsibility areas are organised per location. The entire logistic organisation is supported by the staff logistics, which has for example the responsibility to do the functional management of the ERP system. To conclude, the research is done within CX’ matrix-staff-organization which strives for technology leadership and operational excellence. More specific, the research will focus particularly on the composite parts- and planning CityX responsibility area of the operations & SCM organization.

2.4 Conclusion

The background concludes that the system is repetitively producing low volumes of highly varying customized parts for aerostructures based on long-term make-to-order contracts. These composite parts are manufactured from prepreg material within a general job shop which has a kind of functional lay-out design. On a high level the production steps for manufacturing composite parts are:

- Pre-cut - Lay-up - Autoclave - Debag - Oven - NDI - Milling - Measure

(14)

3. Research design

This chapter elaborates in more detail on the delivery performance problem of the PX composite parts production. This creates the ability to determine the methodology, objective and questions to answer. Therefore this chapter deals with three topics. First, paragraph 3.1 analyses the delivery performance in detail. Second, the methodology for dealing with the problem is determined. Finally, paragraph 3.3 elaborates on the problem statement and sub-questions of the investigation.

3.1 Problem analysis

This paragraph elaborates on the PX composite parts delivery performance problem in more detail for understanding what needs to be investigated. Before going into detail about CX’s situation, the importance of delivery performance needs to be defined. An area of concern for many manufacturing companies is the ability to deliver their products on time (Sridharan & Xiaoming, 2008) and more specific, meet quoted dates and quantities (Sarmiento et al., 2007). Due to intense competition, appeal for more reliable delivery times has reached practically all industries (Kumar & Sharman, 1992) and is considered as a source of potential competitive advantage (Skinner, 1969). This competitive advantage is in particular of interest for make-to-order (MTO) companies, because high delivery reliability is one of the order winning performance criteria for MTO companies (Soepenberg et al., 2008). Concluding, higher delivery performance will give CX a competitive advantage.

Now the delivery performance problem will be discussed in more detail. The measure of delivery performance is lateness, which can be calculated by subtracting the planned delivery date from the actual delivery date (Soepenberg et al., 2008). Delivery performance of a delivered order can be characterized as an on-time performed order or a late delivered order. Therefore a population or sample of orders can be characterized by the performance measure of on-time performance and late delivered orders which are shown in figure 5.

Figure 5: Formula dashboard of delivery performance

(15)

set includes reschedulings, which could have influenced the results. The result of the analysis is shown within the probability plot of PX composite parts lateness in figure 6. Out of figure 6 can be concluded that the PX composite parts have an OTP of 53,7%. The remaining 46,3% late delivered orders were also analyzed separately. This led towards an average lateness of 23 days and standard deviation of 18 days. This lateness needs to be placed in perspective by comparing it with the average cycle time of 31,67 days. The average standard deviation (the average of all individual article cycle time standard deviations) belonging to this average cycle time is 11,31 days. By comparing the lateness of 23 days and the average cycle time of 31,67 days can be concluded that the 46,3% of late orders have high impact on cycle time.

The delivery performance problem could be assigned towards a specific article type or group of article types due to for example quality problems. Therefore the OTP per PX composite article type is analyzed. Appendix 3 contains the analysis of the OTP per PX composite article type. 25 of the 40 article types have an OTP lower than 75%. Therefore, the research can not focus on a specific PX composite part type in order to improve the OTP, because the delivery performance problems apply to 62,5% of all types. Concluding, the research will elaborate on the delivery performance of all PX composite part types.

To conclude, 53,7 % on-time performance is ascertained to be a problem for the competitive advantage of the PX composite parts production. In order to create deeper understanding of this problem the upcoming paragraphs will unravel the delivery performance problem.

3.1.1 Starting moment & lead time realization

The delivery performance is dependable on two variables: starting moment and lead time realization. The reason for this is that the two possible causes of late orders are late

(16)

starts or longer cycle times than lead times. This paragraph discusses both variables in detail in order to determine the variable which has most negative impact on delivery performance.

The first variable which influences delivery performance, as discussed previously, is the starting moment of the order. The starting moment is the moment an order is released onto the shop floor. The measure of starting moment is start lateness, which can be calculated by subtracting the planned start date from the actual start date. The performance measure of the starting moment is the percentage of on-time starts, which is calculated with the formula shown in figure 7.

Figure 7: Formula dashboard of starting moment

The data set of figure 6 is statistically analyzed for creating insight into the measures of figure 7 for the PX composite parts production. The result of the analysis is shown within figure 8, which shows the probability plot of start lateness. Figure 8 shows that 61% of the orders started on-time. The remaining 39% of late starting orders were analyzed separately and had an average start lateness of 19 days with a standard deviation of 23 days.

The second variable which influences delivery performance is the lead time realization. Lead time realization depends on the appropriateness of the quoted lead times and the way the lead times are realized by the cycle time. The reason for this is the management constant of lead time which highly influences the time phasing function of MRP. In contrast, cycle times are generally random. The measure of lead time realization is cycle time variability, which can be calculated by subtracting the lead time from the cycle time. The

(17)

Cycle time variability:

The quality of nonuniformity of the time parts spend as WIP compared to the time allotted for production of a part on that routing or line.

Law of lead time:

The manufacturing lead time for a routing that yields a given service level is an increasing function of both mean and standard deviation.

definition of cycle time variability can be realized by combining Hopp and Spearman’s (2008) definitions of cycle time, lead time and variability shown in table 1. Combining these definitions leads towards the following definition of cycle time variability:

Concept Definition

Lead time The lead time of a given routing or line is the time allotted for production of a part on that routing or line.

Cycle time The cycle time of a given routing is the average time from release of a job at the beginning of the routing until it reaches an inventory point at the end of the routing. Or the time which parts spend as WIP.

Work-in-progress (WIP) The inventory between the start and end points of a product routing.

Variability Variability is formally defined as the quality of nonuniformity of a class of entities. Table 1: Definitions of lead time realization related concepts (Hopp and Spearman, 2008)

The performance measure of the lead time realization is the service level percentage, which is calculated with the formula shown in figure 9. The understanding of the relationship between cycle time variability, lead time and service level can become more clear by considering the law of lead time. This law explains the fundamental principle relating cycle time variability to required lead times (Hopp and Spearman, 2008):

Figure 9: Formula dashboard of lead time realization

(18)

When the measures of start lateness and cycle time variability are compared, then can be concluded that the service level performance is worse than the on-time start performance. As a result, lead time realization has the most impact on CX’s actual delivery performance of 53,7%. This conclusion can be strengthened by analyzing the 46,3% of late orders separately. In other words, the demarcation towards cycle time variability becomes more clear when unravelling the late orders separately. Figure 11 visualizes the partition of the late orders into the following causes:

- 25,66% had a late start.

- 38,16% had a longer cycle time than lead time.

- 36,X% had a late start and longer cycle time than lead time.

Figure 11 shows that the amount of late delivered orders which have problems with cycle time variability is higher than the amount of orders which have problems with start lateness. Therefore the research needs to focus at cycle time variability. Within the diagnosis of chapter 4 will be investigated if the late started orders influence the cycle time variability by rush orders. Hopp and Spearman (2008) quote the following about the importance of reducing cycle time variability when delivery performance needs improvement:

Figure 10: Probability plot of the lead time realization (days)

(19)

The key to keeping customer service high is a predictable flow through the line. In particular, we need low cycle time variability. If cycle time variability is low, then we know with a high degree of precision how long it will take a job to get through the plant. This allows us to quote accurate due dates to customers, and meet them. Low cycle time variability also helps us quote shorter lead times to customers (Hopp and Spearman, 2008, page 360).

From this quote can be concluded that a stable cycle time will improve predictability of the production system. This increases reliability of the planned lead times and thereby the delivery performance of the production system. Concluding, this research is going to focus on lead time realization for reducing cycle time variability of PX composite parts.

3.1.2 Lead time realization per department

The lead time realization, measured by service level, includes the biggest share towards late delivered orders in comparison with the starting moment. The service level of the PX composite parts production depends on the lead time realization of all individual departments in the routing chain. Therefore this paragraph discusses per department the cycle time variability measure and service level performance measure shown in figure 9.

The analysis of the cycle time variability and service level is performed on data of all processed orders per department. This data contains the registration of lead time and cycle time per order per department. The times within this data are based on actual start- and finish date by bookings of finished operations on productive cost centres. Within this data the milling department is split up as follows: Fooke, Belotti, Jobs and Hand/coat. The statistical analysis is performed for the period between 29 March 2010 and 13 June 2010. Concluding, this paragraph analyses the cycle time variability and service level per department.

First, a statistical analysis is performed on the cycle time variability per department. The results are shown in table 2. Table 2 shows that the departments with the highest average and standard deviation of cycle time variability are the Fooke, Measure and Jobs.

CT variability Pre-cut

Lay-up

PROGRAMX Oven Autoclave Debag

PROGRAMX NDI Fooke Belotti Jobs Hand/coat Measure

-1,86 -1,52 -0,04 -0,92 0,57 -0,64 3,46 1,00 1,12 1,05 3,01

median -2,00 -1,83 -1,00 -1,00 -0,13 -2,00 1,92 -0,04 -1,02 -0,06 1,96

sigma 1,48 4,05 2,06 1,96 2,81 3,57 7,66 4,40 6,79 3,59 6,23

min -5,00 -6,88 -3,00 -7,00 -2,00 -7,00 -5,00 -5,00 -5,00 -0,38 -50,00

(20)

Second, the service level is determined per department for the same data set. Besides, the impact of the service level is determined by the total amount of orders having a shorter, equal or longer cycle time than the lead time. Figure 12 shows the results.

0 500 1000 1500 2000 2500 3000 3500 Pr e-cut Lay-u p JSF Oven Autoc lave Deba g JSF ND I Fook e Belot ti Jobs Hand /coat Meas ure O rd er s (# ) 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% 80,0% 90,0% 100,0% S er vi ce le ve l ( % ) # CT>LT # CT LT % Service level Figure 12: Overall analysis of the service level per department

Figure 12 shows that the worst performing departments regarding service level percentage are the Fooke, measure and debag PROGRAMX. However, figure 12 shows that the Fooke, NDI and Belotti have the highest amount of orders with a longer cycle time than the lead time. Therefore, it can be concluded that the NDI department is still having a big impact on the overall order flow due to the high amount of orders with a longer cycle time than the lead time. Moreover, the debag PROGRAMX has low impact on the overall order flow due to the low amount of orders with a longer cycle time than the lead time. Concluding, the NDI, Fooke, Belotti and measure department perform worse regarding service level and amount of orders with a longer cycle time than the lead time.

To conclude, the detailed analysis of the lead time realization per department showed that the NDI, milling and measure department perform worst with reference to the other departments. The reason for this is that these departments have the highest cycle time variability, lowest service level and highest amount of orders with a longer cycle time than the lead time. Thus, this research has to focus at improving the cycle time variability of the NDI, milling and measure department.

3.1.3 Conclusion

(21)

moment and lead time realization of a production order. The PX composite parts production had 61% on-time start and 55,8% service level. A separate analysis of the 46,3% late delivered orders led towards the following conclusion: the amount of late delivered orders which have problems with cycle time variability is higher than the amount of orders which have problems with start lateness. Therefore this research is going to focus on lead time realization for reducing cycle time variability of PX composite parts. The influence of start lateness on cycle time variability will be investigated within the diagnosis.

In addition, the cycle time variability and service level of the NDI, milling and measure department is ascertained worst with reference to the other departments. Therefore the research needs to focus at cycle time variability of these departments. A more stable cycle time will improve the predictability of the production system. This increases the reliability of the planned lead times and thereby the delivery performance of the production system.

3.2 Methodology

In order to structure the research a methodology needs to be chosen for application, which will be discussed within this paragraph. The methodology of a research needs to fit the problem type of which the methodology is intended for. Jackson (2003) emphasizes this by describing the importance of fit between methodology and stakeholder objectives towards the system of analysis. These objectives can be unitary, diverging or coercive (Jackson, 2003). 26 employees of CX legitimized the delivery performance problem of the PX composite parts production system. These employees included the following functions:

- Program & operations managers - Team leaders

- Production & manufacturing engineers - Planning & logistic employees

- Change office employees (black-belts)

Due to consensus of these stakeholders the delivery performance problem can be defined as one of the unitary objectives concerning the PX composite parts production system.

(22)

DOV-Public Master thesis - Peter Krook 16 Research objective:

Provide CX with recommendations for a design, which in case of implementation will decrease cycle time variability of the NDI, milling and measure operations in order to increase on-time performance of the PX composite parts production system.

Research question:

Which system characteristics need to be changed in which way in order to decrease cycle time variability of the NDI, milling and measure operations for increasing on-time performance of the PX composite parts production system?

methodology considers the subjective perspective of the problem possessor pluralistic and objective (Prins, 2008). This methodology aims for explanation of the undesirable system output and subsequently designing and changing the system in such a way that desired system output arises (Prins, 2008). Therefore the DOV-methodology includes the following phases: diagnosis, design and change. This report elaborates on the diagnosis and design of the problem by answering questions. The methodology for answering these questions is discussed within the introduction of every paragraph which deals with a question. The change phase can be fulfilled by implementation of the designed solution by CX’s management. Concluding, the research will come up with a design for the legitimized delivery performance problem of the PX composite parts production system by applying the DOV-methodology.

3.3 Problem statement

This paragraph will form the problem statement of this research. The problem statement includes the research objective and the research question (de Leeuw, 2002). The intended end result of the research is a design, which in case of implementation will improve or solve the urgent delivery performance problem of the PX composite parts production system. Besides the design needs to be capable of dealing with demand dynamics. Based on the detailed analysis of paragraph 3.1 the research is going to elaborate on the following research objective:

(23)

Paragraph 3.2 determined the DOV-methodology which is going to be used for realizing the design for the problem. In order to structure the process of answering the research question, the research determined sub-questions per phase of the DOV-methodology. These sub-questions will be discussed per phase now.

Diagnosis sub-questions:

First, ensuring that products are made in the required quantities at the right time is a production planning and control issue (Nicholas, 1998). Therefore is the first sub-question for the diagnosis of the delivery performance problem formulated in the following way:

1. How is the PX composite parts production planned and controlled?

Second, all possible causes of the cycle time variability need to be investigated for being able to consider the subjective perspective of the problem possessor pluralistic and objective. Therefore is the second sub-question formulated in the following way:

2. Which causes are possibly influencing cycle time variability?

The diagnosis is finalized when is investigated which root causes explain that the cycle time variability occurs in the PX composite parts production. Therefore the third question is formulated in the following way:

3. Which causes are the factual explanations for the cycle time variability?

Sub-question 2 and 3 structure the causes by fishbone diagrams (Ishikawa, 1985). The reason for using this method is that it is appropriate for identifying multiple potential factors causing an overall effect. The overall effect in this case is cycle time variability. Concluding, the diagnosis’ end result is the factual explanation of the cycle time variability.

Design sub-questions:

When the origin of the cycle time variability became clear out of the diagnosis, the design will come up with a solution to eliminate the root causes. Therefore the fourth sub-question is formulated in the following way:

4. How can the causes which explain the cycle time variability be eliminated?

(24)

variations. Moreover, the capability of dealing with capacity ramp-ups is required either. Therefore the fifth sub-question is formulated in the following way:

5. Is the design robust for changes in demand and production environment?

Finally, paragraph 3.2 stated that the change phase can be fulfilled by implementation of the designed solution by CX’s management. Before implementation the management needs to decide whether the designed solution will be implemented. Therefore CX should have insight in the requirements and expected results for being able to take a deliberate decision. Besides, these insights are necessary for being able to make an implementation plan. Therefore, the sixth sub-question is formulated in the following way:

6. What are the requirements and expected results from implementation of the design?

Concluding, the design’s end result is the robust designed solution for elimination of the cycle time variability root causes. Thus, the research will investigate how to decrease cycle time variability of the NDI, milling and measure operations in order to increase the on-time performance of the PX composite parts production system.

3.4 Conclusion

This chapter analyzed the delivery performance problem of the composite parts production in more detail for determining the methodology, objective and questions to answer. The conclusions are discussed now. First, the problem analysis leaded towards the conclusion that this research needs to focus at cycle time variability of the NDI, milling and measure department. Decreasing cycle time variability at these departments will improve the competitive advantage of on-time performance by increasing lead time reliability.

(25)

4. Diagnosis

Paragraph 3.3 described the end-result of the diagnosis as follows: the factual explanation of the cycle time variability. In order to realize this factual explanation this chapter is structured according to the sub-questions defined in paragraph 3.3 of the research design. First, the way the planning and control of the PX composite parts production functions is discussed. Second, the possible causes of cycle time variability are going to be determined. Finally, the third paragraph determines relevant causes which are the factual explanation of the problem.

4.1 How is the PX composite parts production planned and

controlled?

This paragraph explains how the PX composite parts production is planned and controlled in order to understand how CX ensures current delivery performance. Empirical research of Lane and Szwejczewski (2000) investigated that a responsive planning and control system is the most important facilitator of good delivery performance on products made-to-order. The research of Wang and Lin (2009) is also strengthening that in today’s manufacturing enterprise, performance of customer service level is highly dependent on the effectiveness of its manufacturing planning and control system. Therefore planning and control is investigated.

The research conducted in-depth interviews with employees from operations, logistics and finance in order to realize the investigation. Besides, the researcher worked along with the different planners and observed the PX composite parts production. The insights gained from the gathered information is discussed by dealing with the following topics. First, the structure of the planning and control is discussed. Then is discussed how planners work within three responsibility areas. The second last paragraph discusses the key performance indicators (KPI’s) which are controlled per organizational responsibility level of the PX composite parts production. Finally, the conclusion answers the first diagnosis sub-question.

4.1.1 Planning and control decision making levels

(26)

Figure 13: Planning and control elements per level

Figure 13 is further explained now. The starting point of the current control mechanism is the demand on a make-to-order basis. The information about this demand is delivered in the format of a sales- and operations plan. The program support officers deliver this plan to the logistic staff department. The sales plan consists of the amount of ordered ship sets split up into the amount of customer specific end-item number configurations which need to be produced at a certain date. A ship set is a complete set of the end-items which CX supplies in order to assemble one end-item for the customer. Hereby the operations & SCM organization expect programs to decide how to spread demand over time into takts by using the levelling strategy. After the sales plan is filled, the operations plan translates this information into the required amount of capacity per department over time. Thus, the sales and operations plan is the highest information level about material and capacity requirements per program.

(27)

Due to the fact that demand is dynamic, the sales- and operations plans change too. Every time the sales- and operations plan changes, these changed documents have to go through the demand management cycle (DM). The first step of the cycle is that the program has to do a demand management proposal to the logistic staff. The logistic staff is evaluating the feasibility for the material- and capacity requirements by the impact-analysis twice per month. This analysis shows the impact on the required capacity and materials per department with a time scope of half a year. In other words, the document shows the deltas for the MPS and RCCP. Hereby the logistic staff is sending demand management proposals to the operations and purchasing management per facility. Thereby logistics guarantees realistic required materials and capacities. These requests need to be accepted by the operations and purchasing departments in order to update the demand on sales lines level. If the demand management proposals are not feasible, the operation’s capacities and purchasing’s suppliers need to be pressured or the demand needs to be pushed rearwards. Therefore, the impact analysis is the basis for the operations management to make decisions about required- and available capacity when demand characteristics change.

On the second lowest level the sales line end-items are placed in time for having even better insight in the material requirements over time. This is realized by calculating the planned start dates backwards daily by subtracting the lead time from the planned delivery date. These subtractions are done by the following tools: Project Requirements Planning (PRP) and Material Requirements Planning I (MRP). The lead time of a given routing or line is the time allotted for production of a part on that routing or line (Hopp and Spearman, 2008). This lead time consists of the standard waiting time per department plus the article task times per routing operation. For the ProgramX the task time includes the setup time. As a result the setup time is not planned separately. The standard waiting times per department which are added to the article task times are shown in table 3.

Department Wait time (days)

Pre-cut 3 Lay-up 5 Autoclave 2 Debag 1 Oven 1 Fooke 3 Jobs 3 Belotti 2 Hand/ coat 0 NDI scan 2 Assembly 1

(28)

Moreover, the PRP/MRP calculates with infinite capacity and uses during calculations article safety time and not article safety stock. However, the PX composite part articles have no safety time. The calculations are done for all anonymous sub assemblies and anonymous parts for realizing a customer specific end-item. Parts and subs become fixed to a specific configuration when these are used within an end-item. If it happens during or after parts and subs are produced, the produced product configuration will be adapted when necessary. Concluding, PRP/MRP gives CX insight in material requirements over time.

On the second lowest level the PRP/MRP do not provide CX with information about capacity. Decisions about required- and available capacity are realized by the Capacity Requirements Planning (CRP) which is connected with the PRP/MRP system. The specific tool used for these decisions is the capacity status graph which shows the amount of required capacity per program per week. By analyzing this graph the management can determine the required amount of shifts and full time equivalents (FTE’s). Thus, CRP provides CX with information about required capacity over time.

On the lowest level Purchasing orders (PUR) and Shop Floor Control orders (SFC) (also called Production Activity Control) are controlled by the PRP/MRP level. On this low level three planners (FTE) release and priority dispatch PX end-items, sub assemblies and parts. If released orders arrive at the department the team leaders control required- and available capacity by deciding when to overlabour. The PX planning distinguishes three responsibility areas: composite manufacturing, composite milling and PX assembly.

Overall can be concluded that the MPS and RCCP give the integrated view on material and capacity requirements of the sales and operations plans per program. PRP/MRP places the material requirements in time by calculating backward. The connected CRP system shows capacity requirements. Finally planners release and priority dispatch SFC- and PUR orders. The upcoming paragraphs describe the way of working for releasing and priority dispatching the SFC orders by the planners of the three responsibility areas.

4.1.2 Composite manufacturing planning

(29)

been drawn according to the Business Process Modelling Notation (BPMN) (White and Miers, 2008).

Figure 14: Composite manufacturing planning information scheme

The planning steps which are shown in figure 14 will be discussed next. The first step of the planning process is the generation of a list with orders which need to be planned. On Monday the planner starts with collecting an order list per lay-up department out of the ERP system for a time scope of 3 weeks. The order list contains the following information:

- Planner code - Production order - Project

- Article number - Description

- MRP calculated start date - Delivery date

- Order quantity - Stock shortages - Production time

(30)

When this order list is ready per lay-up department, the second step of the planning process is to fill the week loading per lay-up department. The composite parts production has more lay-up departments which are organized per program. The week loading shows an overview of all the articles per lay-up department and a schedule of the week. This overview is filled with the amount of articles per day in order to get insight in the required lay-up and debag hours. Priority dispatching is done by looking to the planned start dates of every order per individual mould type within the order list. During the filling process the planner is taking into account the material shortages, because orders without material will not be planned due to open time. Material availability is considered by checking the following information:

- The mark for stock shortages on the order list

- The ‘material shortages list’ with all material shortages per order - The amount of stock and reserved stock within the ERP system

During the filling process the planner considers also the amount of required capacity. The planner uses a levelling strategy during filling, because a guideline is used of a 12 week average required amount of hours which need to be processed per lay-up department. The planner chooses the specific days for filling an order based on personal experience with the standard available men capacity per day. This is compared with the filled amount of hours for laying-up and debagging the articles. Every Monday the available men capacity per lay-up department is reported to the planner by the team leaders. Then the planner is able to check extreme differences between the amount of required- and available men hours. To conclude, the planner tries to match required- and available men capacities within the week loading.

Another factor considered during filling of the week loading is the idle orders. Idle orders their cycle time exceed the lead time within a specific department and do not move within the queue. Considering idle orders will guarantee that the right order priorities are used, because idle orders usually need to be processed as soon as possible. The last issue considered during filling is the mould availability. When more than one article needs to be planned within the same week the lead times of lay-up, autoclave and debag are important to consider. The reason for this is that these lead times will determine the mould availability for the second product in the week. Concluding, the filled week loadings per program are realized by the consideration of material availability, men capacity, idle orders and mould availability.

(31)

needs to be filled with the necessary amount of moulds per type and then calculates if it is possible to load the autoclave with these moulds based on mould volume. In order to form the autoclave charges per day, all curve numbers are outlined within the week loading per lay-up department. This step consists of adjusting the articles per day within the week loading in order to fit the charge and the other way around. Therefore, priority dispatching of lay-up is influenced by the planned charge moments. Concluding, autoclave charges are planned in the week loading documents and thereby release dates are determined.

When a charge is planned, the charge is also placed within the integrated autoclave planning which contains the helicopter overview of the scheduled curves per autoclave per day. There are some lay-up departments which share curves. Therefore all lay-up week loading documents are picked together for planning the autoclave charges. Next to the lay-up week loading documents, the ProgramXYZ request for charges. These charges need to be planned within the integrated autoclave planning too. Another extra dimension of the autoclave planning is that some curve charges are only allowed to run within a certain time slot due to their high energy consumption. In short, the autoclave planning plans the charge capacity, which constrains the amount of composite parts that can be made.

In the fourth step of the planning’s process the planner makes an overall capacity overview of all the lay-up departments. This document integrates the required- and available hours out of the week loading documents per lay-up department per day. The capacity overview is necessary for fitting available capacity of men hours with the required amount of hours. After making this capacity overview, it is discussed on the capacity meeting with all team leaders.

(32)

After the capacity meeting the orders are released daily. All planning documents need to be finished on Wednesday, because then the first orders are released, despite that the capacity will be discussed on Thursday. Orders are released daily three days before the planned start date of the lay-up for printing the order, getting materials out of the freezer and performing the cut operations. On the same day as the SFC document goes to the pre-cut department, the planner also releases the material request for the warehouse. Concluding, orders are released daily three days before the planned start date of lay-up.

According to Wielen et al. (1995) the consideration of operational characteristics regarding the production units is important during diagnosis or design of the production planning and control. Operational characteristics are the characteristics of the production units which are important to consider during central controlled order release. The four operational characteristics defined by Wielen et al. (1995) are:

- Series size constraint - Sequence constraint

- Workload constraint - Capacity constraint

Due to the importance of considering operational characteristics, this paragraph will discuss them also from the composite parts production planning and control. Out of figure 14 can be concluded that during order release of composite parts production the planner is also considering operational characteristics. Table 4 shows which operational characteristics are currently considered per department during order release on the composite parts production.

Series size Sequence Workload Capacity

Pre-cut Fixed for open time weekly

Lay-up

MRP planned start

Fixed for open time weekly Matching men hours # FTE's, # LPS's

Autoclave Max batch volume Fixed for open time weekly Max # charge hours 3 autoclaves, 3 shifts

Debag see lay-up see lay-up

NDI

Milling

Measure

Table 4: Considered operational characteristics during current parts release

(33)

4.1.3 Composite milling planning

The second responsibility area which will be discussed is the composite milling planning. The way of working of this planner is discussed in detail in order to understand the current way of planning and control. The composite milling planner plans the milling department with the operations of deburring, coating and the three milling machines. The planner plans these departments with a frequency of 3 to 4 days with two days fixed. The fixation of those two days is necessary for procuring and delivering tools by the delivery of the tool preparation department. All the orders which are planned by the milling planner are already released by the composite manufacturing planner(paragraph 4.1.2) or PX assembly planner (paragraph 4.1.4). Therefore, the planner is only dispatching priorities. The steps for realizing the composite milling planning are shown in the information scheme of figure 15 which is drawn according to the BPMN (White and Miers, 2008).

Figure 15 is discussed in more detail now. The orders which require milling operations become known by the planner through three different ‘channels’. These incoming channels which the planner of the milling department needs to control are the priority list, the PROGRAMY lay-up department and the PROGRAMX assembly department. The first channel is the priority list. This list contains all orders within the queue of the milling department. Orders emerge at this list when orders are booked finished at the operation on the routing before the milling operation. This list per milling sub department contains all the

(34)

WIP orders in order of prioritization and planned start date. The list contains also orders with the following characteristics which disturb the planner’s overview:

- Orders which finished milling and not finished deburring or coating - Orders which are stagnated

- Orders which have the non-conformance status

- Order without the product available at the milling department

The second channel is the mail of the PROGRAMY lay-up department which informs the planner about the articles which are expected to arrive at the milling department. The motivation for planning this department apart from the priority list is the long article task times. The third channel is the overview of PROGRAMX assembly. This overview informs the planner about available orders which are allowed to be planned for the milling operation per day. The motivation for considering this department apart from the priority list is the dependence of the PROGRAMX assembly on the fixture for milling. The reason for this is that these fixtures are used for assembly operations and milling operations. Thus, the three demanding information channels are: the priority list, the PROGRAMY mail and the PROGRAMX assembly overview.

When all the information is available the planner starts with an empty ‘checklist workable work’ per milling machine in order to plan the upcoming 3 to 4 days. Orders can be planned within this list by putting the following information in the document:

- Day / time / shift - Program

- Article

- Machine task time

- Order number - TPO/FAI mark - Operation type

(35)

and drill-foam articles are planned for the weekend, because these articles are standard milling operations. These articles are planned out of the priority list. Finally the planner is planning program specific time slots for milling operations based on the sequence of the priority list. To conclude, the sequence of checklist filling is:

1. TPO orders

2. PX assembly orders 3. High urgency orders

4. Co-bond/drill foam 5. Priority list

When the checklist is filled, the document is reviewed by the milling team leader in order to check the feasibility regarding the availability of moulds, tools and men capacity. Then the planning is adjusted and provided to the milling department. Concluding, the ‘checklist workable work’ is the tool used for dispatching priorities by the milling planner.

Overall can be concluded that the composite milling planning controls the following incoming flows: composite manufacturing, the PROGRAMY lay-up and the PX assembly. The milling planner dispatches priorities of already released orders by filling the ‘checklist workable work’ per milling machine with a fixed sequence of orders or time slots.

4.1.4 PX assembly planning

The third responsibility area which is discussed is the PX assembly planning. The way of working of this planner is discussed in detail in order to understand the current way of planning and control. The PX assembly planning is relevant for the composite parts production due to the fact that the milling and measure department perform operations for this planner. Figure 16 shows an information scheme of the PX assembly planning steps which is drawn according BPMN (White and Miers, 2008).

(36)

The steps shown in figure 16 are discussed now. First, the planner gets the overview of the orders which need to be released by looking into his mailbox. These orders get into the mailbox 7 days before the planned PRP/MRP release date. The mailbox contains the following information per order:

- Planned release date - Article number - Order number

- Order quantity - Start date

- Material shortage or not

Priority dispatching and release are based on the planned release dates. Thus, demanded subs and assemblies are released based on PRP/MRP calculated planned release dates.

Before orders are released the material availability is considered by the PX assembly planner. Therefore, the mailbox contains also an information field with the sign if material is short, however the scope of the mailbox is only 7 days before order release. If the planner wants to have insight into the material availability with a longer time scope, the VETA list with a time scope of 8 weeks can be used. This VETA list shows all end-items which (expect to) have material shortages on the planned start date of the item operation. For the end-items on the list is shown which parts are becoming short at the planned start date and who needs to deliver these parts. This person, another planner or purchaser, will be contacted if material availability is becoming a problem for releasing the order. To conclude, the material availability is considered and controlled before release by the VETA list.

If all the materials are available, the planner can release the order. During the release of order into the PX assembly department the planner considers the amount of required capacity. The required amount of capacity is considered by releasing with the guideline of a 12 week average required amount of hours which need to be processed within the PX assembly. Available capacity of the assembly department and mould availability are not taken into consideration. Paragraph 4.1.3 already explained the importance of considering operational characteristics during order release. The only operational characteristic considered by the PX assembly planner is the sequence constraint of MRP planned starts. Concluding, PX sub and assembly release is guided by a 12 week average required amount of hours and MRP planned starts.

Improving on-time performance of the composite parts production for ProductX