Supply Chain Quality Management: A Simulation Study

Information and Management Sciences Volume 19, Number 1, pp.
131-151, 2008
Supply Chain Quality Management: A Simulation Study

Chu-hua Kuei Pace University U. S. A. Janice K. Winch Pace University U. S. A. Christian N. Madu Pace University U. S. A.
Abstract This paper demonstrates the application of a sequential experimentation and simulation metamodeling to supply chain quality modeling. The supply chain simulation model examines the eectiveness of supply chain operations, demand uncertainty, supply chain speed, and quality and distribution issues. The eects of critical supply chain factors on quality and speed of a supply chain network are investigated. Our results show that demand and supply uncertainties do not determine the viability of the supply chain networks. Rather, the viability of supply chain networks is primarily determined by supply chain quality (SCQ). The approach presented in this paper can be extended to all areas of supply chain quality management and development. This study is based on the use of statistical experimental design methods and simulation metamodeling.
Keywords: Simulation, Supply Chain Quality Management, Taguchi Design. 1. Introduction Enterprises depend on the eectiveness of supply chain networks to oer better and cheaper products, shorter response times, and higher service levels. The term supply chain networks is broadly dened as a set of networked organizations that work together to respond rapidly and correctly to customer needs (see Figure 1). Supply Chain Management (SCM), Just-in-time (JIT), Total Quality Management (TQM), Six Sigma Quality Management, Enterprise Resource Planning (ERP), and Supply Chain Quality Management (SCQM), as a result have been recognized as major initiatives that can help
Received May 2006; Revised October 2006; Accepted February 2007. Supported by ours.
Information and Management Sciences, Vol. 19, No. 1, March, 2008
Figure 1. Supply Networks.
rms to enable conditions and nurture capacities for supply chain eectiveness (Chow et al. 2006, Robinson and Malhotra 2005, Kannan and Tan 2005, Takahashi and Hirotani 2005, Kelle and Akbulut 2005, Madu and Kuei 2004, Tari and Vicente 2004, Sheu et al. 2004, Kuei and Madu 2003, Kuei et al. 2002, Kuei and Madu 2001). Supply chain management is a holistic and strategic approach to demand, operations, procurement, and logistics process management (Chow et al. 2006, Madu and Kuei 2004, Kuei et al. 2002). Just-in-time is a sell-one-make-one and a supplier-enterprise-co-production system (Carreira 2005, Madu and Kuei 2004, Kannan and Tan 2005, Takahashi and Hirotani 2005). Total Quality Management calls for developing a system emphasizing continuous improvement, employee focus, data driven decision making, and the voice of the customers (Kannan and Tan 2005, Tari and Vicente 2004, Kuei and Madu 2003). The focus of Six Sigma Quality Management is on developing and implementing highly capable business and operations processes to meet the needs of stakeholders (El-Haik and Al-Aomar 2006, Kuei and Madu 2003). Conversely, ERP is a system whereby all the functional units of an enterprise are integrated onto a single computer system that will serve dierent needs of the end users (Kelle and Akbulut 2005, Madu and Kuei 2004, Sheu et al. 2004). Robinson and Malhotra (2005) take a process centric view and dene SCQM as the formal coordination and integration of business processes involving all partner organizations in the supply chain channel. This mega process involves measuring, analyzing, and continually improving products, services, and process. The aim is to create value and achieve satisfaction of intermediate and nal customers in the market place (Robinson and Malhotra 2005). Kuei and Madu (2001) dene Supply Chain Quality Management (SCQM) with three simple equations where each equation represents the letters that make up SCQM. The denition is as follows: SC = a production-distribution network,
133
Q = meeting market demands correctly, and achieving customer satisfaction rapidly, and protably, and M = enabling conditions and enhancing trust for supply chain quality. To gain competitive advantage, enterprises need to make the most out of the above initiatives to master the challenges of speed and quality in the 21th century (Madu and Kuei 2004). While supply chain quality management appears to be one of the key determinants of supply chain eectiveness, there is still less documented evidence of its critical success factors and their impact on key supply chain performance indicators. Consequently, of particular interest in this study is the inuence of competitive factors in supply chain quality management. The interest in identifying the inuential competitive factors in supply chain quality management has been growing over the past few years (Chow et al. 2006, Lin et al. 2005, Kuei et al. 2005). The objective of this study is to assess the eects of critical supply chain factors on quality and speed of a supply chain network. A supply chain simulation model is established to look into the eectiveness of supply chain operations, demand uncertainty, supply chain speed, and quality/distribution issues.
2. Literature Review 2.1. Supply Chain Simulation model Simulation models are increasingly being used to analyze complex supply chain problems (Reiner 2005, Holweg et al. 2005). To satisfy customer demands for shorter lead times, increased delivery frequency, and improved product and process quality, a simulation model was adopted by Jansen et al. (2001) to study a catering supply chain in the Netherlands. In their study, various distribution scenarios were tested for logistical and nancial performances. Jain et al. (1999) used simulation to understand the behavior of supply chain networks and were able to identify the major issues surrounding such networks. They identied logistics and business processes as the two major issues: Logistics: The transportation of materials between facilities should be included. Business Processes: Basic operations processed such as forecasting, production, inventory control, and customer orders should be considered in the simulation model to meet the objective of the study.
134
Towill et al. (1992) compared dierent quick-response-to-orders strategies based on a supply chain simulation model. They concluded that collaboration between all players within the supply chain could result in great benets. Jain et al. (2001) also proposed a framework for supply chain analyses. Critical elements in the framework included product characteristics, demand volumes and proles, distribution center lead times, supplier lead times, cost data, strategic options, supply chain structure scenarios, and performance indicators (e.g., service levels, lead times, and inventory turnover). Reiner (2005) also establishes a discrete-event simulation model to evaluate supply chain process improvements. Using a simulation of a multi-tier vehicle supply system, Holweg at al (2005) investigate potential improvement opportunities in the area of scheduling. The Taguchi method is employed to determine the number of simulation trials. 2.2. Supply Chain Quality management Does SCQM really pay? Lin et al. (2005) note that, by using a structural equation model, quality management practices are signicantly correlated with the supplier participation strategy. This association may inuence tangible business results and customer satisfaction levels. Kuei et al. (2005, 2001) also tested the association between organizational performances and SCQM practices such as manufacturers knowing their suppliers quality records, considering a suppliers capability when designing products, and establishing long-term relationships with individual suppliers. Their results show that perceived improvements in organizational performance are associated with improvements in supply chain quality management practices. A structured quality approach thus needs to be adopted to (1) design quality into a supply network, (2) optimize materials/products/services systems, (3) stabilize the SC-wise quality system, and (4) maximize ERP systems. Anthony et al. (2005) also note that in todays global market, large enterprises are heavily dependent on small and medium sized rms to provide high quality products at low costs. Through the empirical studies on supply groups, they learned that quality practices do benet small and medium sized rms within the supply chain in UK. Key benets to business include reduction in process variability, increased protability, and reduction of operational costs. While there are empirical studies to verify the association between supply chain quality management practices and organizational performance, to the best of our knowledge, there are no simulation studies to date to associate SCQM variables and supply chain
135
eectiveness. Furthermore, empirical results generated from questionnaire surveys frequently used in these studies are often limited because they measure the perceptions of the respondents rather than actual behavior. Simulation on the other hand mimics the real life behavior of the system and may give some insight into supply chain networks. This paper, therefore, uses simulation to develop associations between supply chain quality management practices and organizational performance. A simulation metamodel is developed to facilitate the use of the model and the specic relationships that are observed. 2.3. Taguchi Designs and Metamodel Madu and Kuei (1992, 1993), and Lin et al. (1994) were perhaps among the rst groups that elaborated the use of Taguchi designs and regression (or metamodel) models. Friedman and Pressman (1988) dened metamodel as any auxiliary model which is used to aid in the interpretation of a more detailed model. Initially, Taguchi experimental designs were used primarily in the area of quality assurance for screening out signicant and robust experimental factors. A unique aspect of the Taguchi designs is their use of linear graphs to generate fractional factorial design plans. When dealing with a large number of experimental factors in a complex operations situation, the construction of fractional factorial designs can result in the reduction of experimental trials. Holweg et al. (2005) also adopt Taguchi methods when conducting their supply chain simulation studies. The benets of constructing a metamodel in a simulation study include model simplication, enhanced exploration and interpretation of the model generation to other models of the same type, sensitivity analyses, optimization, answering inverse questions, and providing the researcher with a better understanding of the behavior of the system under study (Friedman and Pressman 1988). Further, as noted by Madu and Kuei (1992), in solving complex problems, simulation normally becomes attractive. However, there are problems with the use of simulation, especially when researchers need to consider numerous experimental factors. In this case, even with the help of fractional factorial designs, a prohibitive number of simulation runs is still required. Madu and Kuei (1992) thus adopted the multiple-stage experimentation approach. At the rst stage of the simulation study, independent factors are grouped based on their common characteristics. If the direction of the factor is unclear, it is put into one group by itself. The Analysis of Variance (ANOVA) test is used to determine
136
the signicance of these group factors. At the conclusion of the rst stage of the experiment, signicant group factors will be disaggregated into single factors in the follow-up experiment. The newly found single factors will be further tested to determine their possible eects on the performance measure of interest. A metamodel might be constructed at this stage to meet the research interests. This procedure is followed in the present simulation study.
3. Problem Description A typical supply chain problem involves a manufacturer who sources items or component parts from dierent vendors. These items are sent to a distribution center where the nal assembly is done and then shipped to fulllment centers that are closer to the products demand or the marketplace. To illustrate and develop the simulation metamodel, a hypothetical supply chain problem is proposed. Consider a single supply chain network that operates with one distribution center located in Taipei, Taiwan. Currently, the distribution center purchases monitors, keyboards, and power supply modules from suppliers in China or Korea. The nal assembly is done in the distribution center. Customer orders from the North American market are consolidated in the order fulllment oce located in Edison, New Jersey, USA. They are fullled subsequently by the distribution center in Taiwan. Data have been gathered to analyze the supply chain network. The data indicate that the time between consolidated customer orders is approximated by a normal distribution with a mean of 48 hours and a standard deviation of 10 hours. The processing time through each supply chain partner and the associated quality records and repair costs are presented in Table 1. Table 1. Original Data.
Distribution Center Power Supply Monitor Keyboard Lead Time N(168, 36) N(72, 12) N(48, 10) N(48, 10) Defect rate 5% 5% 5% 5% MTTR EXP(72) EXP(48) EXP(48) EXP(48) Repair Cost $100/hr $40/hr $80/hr $40/hr
N(mean, standard deviation): Normal Distribution; EXP(MTTR): Exponential Dist (Mean Time to Repair).
This paper develops the supply chain quality model to address the following question: What eects do the critical success factors for supply chain network have on the perfor-
137
mance measure of supply chains? The measures of performance considered here include response time and the cost of non-conformance in the distribution center. Some of the primary assumptions of the supply chain quality model discussed in this paper are summarized as follows: Customer orders are consolidated and follow the normal distribution. The lead time between the distribution center and the order fulllment center follows the normal distribution. The lead time between the distribution center and suppliers follows the normal distribution. The defective rate at the distribution center is a constant. The defective rates at the supply groups are identical. The defective items are repairable and are assumed to be completely rejuvenated after each repair. The repair time follows the exponential distribution. The repair cost is a constant.
4. Simulation Model GPSS/H is used to code the supply chain quality simulation model. There are three major segments and one module in the simulation model. The simulation model adopted for this study is expressed as a block diagram (see Figure 2). Each gure or block, with its own distinctive shape, plays a particular role in our Supply Chain Simulation model. The rst segment consists of market demands and customer orders. The second segment represents the operation of the distribution center. The third segment depicts the operations of three supply groups and the assembly process in the distribution center. In the last two segments of the model, quality modules are introduced to simulate the failure and repair processes (see Figure 3). From the problem description, ve input (or independent group) factors aect this supply chain system. They are presented below (their assumed levels are enclosed in parentheses): Demand uncertainty (Normal distribution with mean= 48 and standard deviation= 10; Normal distribution with mean=24 and standard deviation= 2) Distribution lead time (Normal distribution with mean= 168 and standard deviation = 36; Normal distribution with mean= 48 and standard deviation=6)
138
Quality level - Distribution Center and Supply Group (3 Sigma: Yield= 99.73%; 4 Sigma: Yield= 99.9937%) Mean Time to Repair - Distribution Center and Supply Group (Exponential distribution with mean 72 or 48; Exponential distribution with mean 18 or 8) Supply lead time - Power Supply Module and Monitor/Keyboard Module(Normal distribution with mean= 72 or 48 and standard deviation= 12 or 10; Normal distribution with mean= 72 or 48 and standard deviation= 3 or 2) Input (or independent group) factors are also shown in Table 2. Further, two response variables considered in this study are: Ltime = response lead time CON = cost of nonconformance, i.e., the repair cost per hour of operation at the distribution center.
Figure 2. SCQM Model.
139
Figure 3. Quality Module. Table 2. Screening Experimentation - Group Factors and Levels.
Group Number 1 2 3 4 5 Content Demand Uncertainty Distribution Lead Time Quality Level Mean Time to Repair (MTTR) Supply Lead Time Level 1 N(48, 10) N(168, 36) 3 Sigma* 3 Sigma** EXP (72)* EXP (48)** N(72, 12)# N(48, 10)## **: Supply Group #: Power Supply Module ##: Monitor and Keyboard N( ): Normal Distribution EXP ( ): Exponential Distribution Level 2 N(24, 2) N(48, 6) 4 Sigma* 4 Sigma** EXP (18)* EXP (8)** N(72, 3)# N(48, 2)##
*: Distribution Center (3 Sigma: Yield= 99.73%; 4 Sigma: Yield= 99.9937%)
The discrete event simulation was conducted for 2400 hours. A total of thirty replications were carried out. The response lead time (Ltime) and cost of non-conformance (CON) are, thus, the average of the thirty replications. El-Haik and Al-Aomar (2006), Carson (2002), and Madu and Kuei (1993) discussed procedures for validating models of this form. This procedure consists of the following two steps: (1) validation, and (2) verication. The former calls for reevaluating the simulation program for errors especially
140
with the logic, while the latter is concerned with the internal consistency of the simulation model (e.g. the output measures may be observed to see if they are reasonable). Those procedures were also followed in this paper to make sure every phase of model construction is validated. 5. Experimentation When large numbers of variables are involved, it is unwise to investigate all the input factors in an experiment simultaneously. Madu and Kuei (1993) suggested the use of a sequential approach as an investigation strategy. This four-step procedure is summarized as follows: Screening Experimentation and Group Factors Follow-up Experimentation and Individual Experimental Factors Developing Regression Metamodels Validation Tests In the initial stage of experiment, group-factors are evaluated as if they were single factors. In the follow-up experiment, only the individual factors within the signicant groups are tested. Regression metamodels are developed subsequently to better understand the supply chain system and the interaction among the systems variables. This procedure is concluded with metamodel validation tests. The essence of this experimentation is to be able to make an informed decision. 5.1. Screening experimentation and group factors From the problem description, ve group factors are identied that may aect supply chain networks operation. The denition of the group factors and the levels used are presented in Table 2. Notice that each factor is examined at two levels. Thus, only the linear eects of each group factor on the dependent variables such as response lead times (Ltime) and cost of non-conformance (CON) can be assessed. This group screening technique is also adopted and discussed in Madu and Kuei (1992). The corresponding mathematical model is shown below:
K K K
Y = 0 +
i=1
i xi +
ij j=2
ij xi xj + e.
(1)
141
Where Y is the value of the supply chain measure of performance such as response lead times (Ltime) or cost of non-conformance (CON); xi is the value of i-th group factor; 0 is the grand mean; i is the coecient of group factor i; ij denotes the coecient of the interaction between group factors i and j; and e is the experimental error. Based on the proposed mathematical model, the L16 factional factorial design (Taguchi and Wu 1980, Peace 1993, Madu and Kuei 1993) is adopted (see Table 3). As a result, the main eects of the independent group factors and their two-factor interaction eects can be estimated with only sixteen simulation runs (Madu and Kuei 1993). Table 3 shows the column assignment of each group factor on the L16 orthogonal array (Taguchi and Wu 1980, Peace 1993, Madu and Kuei 1993). This experimental design plan is prepared based on Taguchis linear graph (Taguchi and Wu 1980, Peace 1993, Madu and Kuei 1993). The results obtained from the sixteen runs are also given in Table 3. Madu and Kuei (1993) discussed the analysis of variance (ANOVA) procedure for testing the significance of both the main and two-factor interaction eects. The procedure was followed here to form the ANOVA tables for both response lead times (see Table 4 (a)) and cost of non-conformance (see Table 4 (b)). Notice that the pooling technique is employed to estimate the experimental error. For example, for the case of Ltime, since the Sum of Squares (SS) values of group 1 factor, group 5 factor, and all the two-factor interactions (except for group 3*group 4) are very small, they are aggregated and used to estimate the experimental error. Table 4 (a) shows that group 2 factor, group 3 factor, group 4 factor, and group 3*group 4 are all signicant (at a level of signicance = 0.01) in estimating response lead times (Ltime). Further, Table 4 (b) shows that group 2 factor, group 3 factor, group 4 factor, group 2* group 4, and group 3*group 4 are also all signicant in estimating the cost of non-conformance (CON). It is also observed from Figure 4 and Table 4 (b), demand uncertainties (group 1) and supply uncertainties (group 5) do not matter much in the supply chain networks CON in this study. Similar conclusions are drawn for the overall response time. Thus, for supply chain performance indicators such as Ltime and CON, three group factors that really matter in our study are: Distribution Centers Lead Time (group 2), Quality Level at both the distribution center and the supply sites (group 3), and the Mean Time to Repair at both the distribution center and the supply sites (group 4). These ndings are also shown in Figure 4. Further investigation of these three critical group factors is required.
142
Table 3. Screening Experimentation - Systematic Assignment to Five Group Factors and Multivariate Simulation Results.
Group1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Group 2 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Group3 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Group4 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Group5 Level 1 Level 2 Level 2 Level 1 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 1 Level 2 Level 2 Level 1 Avg. Ltime (hrs) 237 230 228 228 128 120 118 117 238 230 229 228 129 119 118 117 Avg. Cost $5447 $1826 $3559 $472 $5800 $1919 $1457 $652 $7804 $1757 $3800 $757 $6430 $1707 $1636 $465
Table 4(a). ANOVA - Group Factors and Ltime.

Source Group 1 Group 2 Group 3 Group 4 Group 5 Group 1 * Group 2 Group 1 * Group 3 Group 2 * Group 3 Group 4 * Group 5 Group 1 * Group 4 Group 2 * Group 4 Group 3 * Group 5 Group 3 * Group 4 Group 2 * Group 5 Group 1 * Group 5 Pool Error: #: At least 99% condence SS 0.25 48620.25 144.00 81.00 0.25 0.25 0.00 1.00 0.00 1.00 1.00 0.00 56.25 0.25 0.25 0.386364 56.25 145.59# 48620.25 144.00 81.00 125840.60# 372.71# 209.65# MS F
143
Table 4(b). ANOVA - Group Factors and CON.

Source Group 1 Group 2 Group 3 Group 4 Group 5 Group 1 * Group Group 1 * Group Group 2 * Group Group 4 * Group Group 1 * Group Group 2 * Group Group 3 * Group Group 3 * Group Group 2 * Group Group 1 * Group Pool Error: SS 649636.00 1792921.00 24730729.00 43487430.00 248502.30 361201.00 299209.00 722500.00 111556.00 805506.30 1701720.00 86142.25 6459222.00 542432.30 596756.30 MS 1792921.00 24730729.00 43487430.00 F 4.05 55.91# 98.31#
2 3 3 5 4 4 5 4 5 5
1701720.00 6459222.00
3.85 14.6#
442344.10
#: At least 99% condence
Figure 4. Response Plots - Main Eects.
5.2. Follow-up experimentation and individual experimental factors For the purpose of this study, only the main eects of single factors are considered
144
at this stage. The mathematical form used in this follow-up experimentation is shown below: Y = 0 +
i=1 K
i xi + e.
(2)
Where Y is still the value of the supply chain performance such as response lead times (Ltime) or cost of non-conformance (CON) and e is the experimental error; xi is now the value of i-th individual experimental factor; and i is the coecient of single factor i. The denition of those individual experimental factors and the levels used are presented in Table 5. The L16 factional factorial design is again adopted here (see Table 6). The systematic assignment of the levels to the ve single factors are also shown in Table 6, with the corresponding values of Ltime and CON, after sixteen simulation runs with thirty replications. Table 5. Follow-up Experimentation - Individual Factors and Levels.
Factor A B C D E Name Distribution Lead Time Distribution Centers (DCTR) Defective Rate Distribution Centers (DCTR) MTTR Supply Groups Defective Rate Supply Groups MTTR Level 1 N(168, 36) 3 Sigma EXP (72) 3 Sigma EXP (48) Level 2 N(48, 6) 4 Sigma EXP (18) 4 Sigma EXP (8)
N( ): Normal Distribution EXP ( ): Exponential Distribution
Table 6. Table 6 Follow-up Experimentation - Systematic Assignment to Five Factors and Multivariate Simulation Results
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Factor A Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Level 2 Factor B Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Level 1 Level 1 Level 1 Level 1 Level 2 Level 2 Level 2 Level 2 Factor C Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 2 Factor D Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Level 1 Level 2 Factor E Level 1 Level 2 Level 2 Level 1 Level 2 Level 1 Level 1 Level 2 Level 2 Level 1 Level 1 Level 2 Level 1 Level 2 Level 2 Level 1 Avg. Ltime (hrs) 237 235 229 232 229 229 235 228 123 123 124 119 124 118 119 118 Avg. Cost $5445 $7628 $1520 $2060 $1007 $2516 $332 $472 $7385 $6589 $2064 $1744 $453 $1475 $325 $364
145
5.3. Developing regression metamodels In this study, regression analysis is employed to test the newly proposed mathematical models. The regression metamodel derived for CON can be stated as: Cost of non-conformance (CON) = 53693(DCTR Defective Rate)*** + 54.67(DCTR MTTR)*** (*** : P < 0.01) (3)
It is observed that CON is signicantly associated to the following two single factors: Distribution Centers (DCDR) Defective Rate, and Distribution Centers (DCTR) Mean Time to Repair (R2 = 0.810959, P = 0.002). To demonstrate the utility of this model, consider the following example: if the DCTR Defective Rate = 5% and DCTR MTTR = 50, then CON is $5418.15/hr (53693(0.05) + 54.67(50)). Putting this into perspective, one could explore options on how to optimize the supply chain. Based on the CON metamodel, it appears that controlling the defective rate or improving quality, as well as reducing the mean time to repair, would signicantly reduce the cost of non-conformance. There are also signicant associations between Ltime and four main factors as shown in the model below. The regression form is as follows (R2 = 0.999096, P = 0.001): Ltime = 70.35*** + 0.92(DCTR Ltime)*** + 42.97(DCTR Defective Rate)** +35.16(Supply Group Defective Rate)* + 0.07(Supply Group MTTR)** (* : P < 0.10, ** : P < 0.05, *** : P < 0.01) (4)
For example, if our DCTR Ltime = 100, DCTR Defective Rate = 5%, Supply Group Defective Rate = 5%, and Supply Group MTTR = 20, then Ltime is 167.67 hours (70.35 + 0.92(100) + 42.97(0.05) + 35.16(0.05) + 0.07(20)). Again, quality is prominent in reducing the lead time. Therefore, to develop an ecient supply chain network would, in part, require developing strategies to reduce the cost of non-conformance and lead time. These could be achieved by improving quality and reducing the mean time to repair. It is also intuitive that an increase in quality would perhaps help to reduce the mean time to repair. 5.4. Validation tests Intuitively, as input factors values (e.g. DCTR Defective Rate) increase, our performance values (i.e. Ltime and CON) should increase accordingly. It is observed from
146
equations (3) and (4), that the functional relationships derived from regression analyses conform to our intuition. To further validate our metamodels, we randomly generated ten design points that were not considered in the follow-up experimentation. Each design point is implemented in both simulation and metamodels. The results for the Ltime are shown in Table 7. The average percentage error is estimated as:
10 i=1
Simulation(i) M eta(i) Simulation(i)
10.
(5)
As the result in Table 7 shows, the average percentage error for estimating Ltime is 1.3%. The metamodel presented as equation (4) is therefore valid in estimating the response lead times (Ltime). Table 7. Validation Test for Ltime.
A* 1 2 3 4 5 6 7 8 9 10 N(160, 30) N(58, 8) N(70, 14) N(150, 20) N(90, 20) N(140, 28) N(60, 12) N(120, 24) N(80, 15) N(110, 25) B 0.06 0.009 0.04 0.03 0.01 0.008 0.05 0.04 0.06 0.03 C 60 50 20 30 45 55 25 48 52 69 D 0.009 0.06 0.03 0.04 0.01 0.008 0.06 0.03 0.05 0.04 E 28 12 45 36 18 10 42 32 20 16 Simulation 223.81 128.81 141.48 214.53 158.60 203.75 136.16 187.43 151.84 177.93 Metamodel 222.40 127.05 140.67 213.57 155.19 200.48 132.75 185.76 154.29 175.37 Average: *: Use mean only for the metamodel % Error 0.006 0.014 0.006 0.005 0.022 0.016 0.025 0.009 0.016 0.014 0.013
As for the cost of non-conformance (CON), with the same ten random design points, the average dierence between our simulation and metamodels is $1,697. Since our CON is the product of repair cost ($100) and average repair time in the distribution center, it appears that our metamodel can not reect the change in CON with a small change in the distribution centers defective rates and MTTRs. Since CON is very sensitive to small changes in defective rates and MTTR, it would probably be better to use simulation models rather than metamodels for estimating CON.
147
6. Implications and Discussions for Supply Chain Quality Management From the preceding analyses of the metamodel of supply chain networks, several observations are summarized as follows: First, the primary aim of supply chain networks is to supply customers with quality products and services. Knowledge of discriminating factors can help enterprises of all sizes along the supply chain to have a strategic focus in achieving their system-wide goals. The simulation results and metamodels presented in this study provide guidelines on critical success factors to realize this goal of supply chain quality management. The critical issues in the realm of our simulation study are: defective rate reduction, problem solving capability, supply chain speed, and supplier involvement. This nding supports an earlier work by Madu and Kuei (2004) and Kuei and Madu (2001). As noted by Madu and Kuei (2004), to achieve supply chain quality, initiatives such as continuous improvement or Six Sigma Quality Management may be adopted. Leadership support for such initiatives must be present. Further, a systemic view of supply chain quality and performance metrics must be established and endorsed by the entire supply chain (see Figure 5).
Figure 5. SCQM. Second, in pursuit of supply chain quality and operational excellence, our metamodels show that supply chain competencies do play a signicant role in the supply chain setting. As noted, for example, Mean Time to Repair (MTTR) is essential with respect to response lead times (Ltime) and cost of non-conformance (CON). It appears that enterprises need to develop competence to respond rapidly to customer needs in order
148
to improve overall supply chain performance. This conclusion is consistent with that of Wisner et al. (2005), Reiner (2005), and Madu and Kuei (2004). Third, as noted by Juran (1992), todays managers must be bilingual (they need to present the case in terms of costs and things). Metamodel development is essential for these two fronts. With our CON model, for example, managers would get a good sense of prot or loss when launching supply chain quality management initiatives. With our Ltime model, project team members know what it will take to accomplish the mission which is linked to customer requirements and supply chain strategy. Fourth, the use of group screening, Taguchi designs, and metamodels in solving complex supply chain quality problems can be extended to all areas of supply chain management and development. Since a supply chain simulation experiment involves a large number of variables, the use of group screening is especially suitable in reducing the number of simulation runs. The Taguchi approach also helps to ensure that appropriate levels of the experimental factors are studied. Metamodels can help decision makers to address sensitivity analysis issues. Finally, our approach presents a pragmatic and eective way to explore the eectiveness of supply chain operations, demand uncertainty, supply chain speed, quality issues, and distribution issues without dwelling on complex mathematical models or empirical studies. Through the use of simulation, the system of interest is adequately analyzed without unrealistic assumptions. The nal metamodels are easily derived and are easy to comprehend and to use.
7. Conclusions In this simulation study, we have demonstrated the links between critical supply chain quality factors and their inuences on two performance indicators, namely, response lead times (Ltime) and cost of non-conformance (CON). By building and examining simulation models, we have a better understanding of the supply chain process. Following the sequential experimentation and two-level Taguchi designs, we identify signicant factors with a minimal number of simulation runs. Emphasis can therefore be placed on the critical factors that add value to the supply chain networks. In the global economy, it is paramount to devote a supply chains limited resources to creating reliable processes and to improving quality and operational eciency. Our results show that the most impor-
149
tant factors in improving the eectiveness of supply chains are quality and operational eciency. Acknowledgement The authors gratefully acknowledge the helpful comments of James Preskar of the Lubin School of Business in the development of this manuscript. References
[1] Antony, J., Kuman, M. and Madu, C. N. Six Sigma in small and medium sized UK manufacturing enterprises: some empirical observations, International Journal of Quality & Reliability Management, Vol. 22, No. 8, pp.860-874, 2005. [2] Carreira, B., Lean Manufacturing that Works, AMACOM: NY, NY, 2005. [3] Carson, J. S., Model verication and validation, Proceedings of the 2002 Winter Simulation Conference, pp.52-58, 2002. [4] Chow, W. S., Madu, C. N., Kuei, C., Lu, M. H., Lin, C. and Tseng, H., Supply Chain Management in the U.S. and Taiwan: An Empirical Study, Omega - International Journal of Management Science, 2006, (forthcoming) [5] El-Haik, B. and Al-Aomar, R., Simulation -based Lean Six-Sigma and Design for Six Sigma, Wiley, Hoboken, NJ, 2006. [6] Friedman, L. W. and Pressman, I. The metamodel in simulation analysis: can it be trusted? Journal of Operations Research Society, Vol. 39, No.10, pp.939-948, 1988. [7] Holweg, M., Disney S. M., Hines P. and Naim. M. M., Towards responsive vehicle supply: a simulation-based investigation into automotive scheduling systems, Journal of Operations Management, Vol. 23, pp.507-530, 2005. [8] Jain, S., Lim, C., Gan, B. and Low, Y., Criticality of detailed modeling in semiconductor supply chain simulation, Proceedings of the 1999 Winter Simulation Conference, pp.888-896, 1999. [9] Jain, S, Collins, L.M, Workman, R.W, Ervin, E.C. Development of a high-level supply chain simulation model. Proceedings of the 2001 Winter Simulation Conference, pp.1129-1137, 2001. [10] Jansen, D. R., Weert, A., Beulens, A. J. M. and Huirne, R. B. M., Simulation model of multicompartment distribution in the catering supply chain, European Journal of Operational Research, Vol. 133, pp.210-224, 2001. [11] Juran, J. M., Juran on Quality by Design. The Free Press: NY, NY; 1992. [12] Kannan, V. R. and Tan, K. C., Just in time, total quality management, and supply chain management: understanding their linkages and impact on business performance, Omega, Vol. 33, No. 2, pp.153-162, 2005. [13] Kelle, P. and Akbulut, A., The role of ERP tools in supply chain information sharing, cooperation, and cost optimization, International Journal of Production Economics, pp.93-94, pp.41-52, 2005. [14] Kuei, C., Madu, C. N., Chow, W. S. and Lu, M. H., Supply chain quality and excellence in the new economy: an empirical study of Hong Kong based rms, Multinational Business Review, Vol. 13, No. 1, pp.33-53, 2005. [15] Kuei, C, and Madu, C.N. Customer-centric six sigma quality and reliability management. International Journal of Quality & Reliability Management, Vol. 20, No. 8, 954-964, 2003. [16] Kuei, C., Madu, C. N., Lin, C. and Chow, W. S., Developing supply chain strategies based on the survey of supply chain quality and technology management, International Journal of Quality & Reliability Management, Vol. 19, No.7, pp.889-901, 2002.
150
[17] Kuei, C., Madu, C. N. and Lin, C., The relationship between supply chain quality management practices and organizational performance, International Journal of Quality & Reliability Management, Vol. 18, No. 8, pp.864-872, 2001. [18] Kuei, C. and Madu, C. N., Identifying Critical Success Factors for Supply Chain Quality Management, Asia Pacic Management Review, Vol. 6, No. 4, pp.409-423, 2001. [19] Lin, C., Chow, W. S., Madu, C. N., Kuei, C. and Yu, P., A Structural Equation Model of Supply Chain Quality Management and Organizational Performance, International Journal of Production Economics, Vol. 96, pp.355-365, 2005. [20] Lin, C., Madu, C. N. and Kuei, C., Experimental Design and regression analysis in simulation: an automated owline case study, Microelectron Reliability, Vol. 34, No. 5, pp.845-861, 1994. [21] Madu, C. N. and Kuei, C., Group screening and Taguchi design in the optimization of multi-echelon maintenance oat simulation metamodels, International Journal of Computers and Operations Research, Vol. 19, No. 2, pp.95-105, 1992. [22] Madu, C. N. and Kuei, C., ERP and Supply Chain Management. Chi Publishers: Faireld, CT; 2004. [23] Madu, C. N. and Kuei, C., Experimental Statistical Designs and Analysis in Simulation Modeling. Quorum: Westport, CT; 1993. [24] Peace, G. S., Taguchi Methods. Addison-Wesley: NY, NY; 1993. [25] Reiner, G., Customer-oriented improvement and evaluation of supply chain processes supported by simulation models, International Journal of Production Economics, Vol. 96, pp.381-395, 2005. [26] Robinson, C. J. and Malhotra, M. K., Dening the concept of supply chain quality management and its relevance to academic and industrial practice, International Journal of Production Economics, Vol. 96, pp.315-337, 2005. [27] Sheu, C., Chae, B. and Yang, C. C., National dierences and ERP implementation: issues and challenges, Omega, Vol. 32, No. 5, pp.361-372, 2004. [28] Taguchi, C. and Wu, Y., Introduction to O-line Quality Control. Central Japan Quality Control Association: Nogoya, Japan; 1980. [29] Takahashi, K. and Hirotani, M. D., Comparing CONWIP, synchronized CONWIP, and Kanban in complex supply chains, International Journal of Production Economics, pp.93-94, pp.25-40, 2005. [30] Tari, J. J. and Sabater, V., Quality tools and techniques: Are they necessary for quality management, International Journal of Production Economics, Vol. 92, pp.267-280, 2004. [31] Towill, D. R., Naim, M. M. and Wikner, J., Industrial Dynamics Simulation Models in the Design of Supply Chains, International Journal of Physical Distribution and Logistics Management, Vol. 22, No. 5, pp.3-13, 1992. [32] Wisner, J. D., Leong, G. K. and Tan, K., Principles of Supply Chain Management. South-Western: Mason, Ohio; 2005.
Authors Information
Dr. Chu-hua Kuei is Professor of Management and Management Science, Lubin School of Business, Pace University, New York. He has over forty refereed publications. He is the co-author of Strategic Total Quality Management (1995) and ERP and Supply Chain Management (2004). His research interests are computer simulation, management science, supply chain management, and quality management. Department of Management and Management Science, Lubin School of Business, Pace University, 1 Pace Plaza, New York, N.Y. 10038, U. S. A. E-mail: ckuei@pace.edu Dr. Christian N. Madu is Research Professor and Management Science Director at the Lubin School of Business, Pace University, New York. He also serves as co-editor of the International Journal of
151
Quality & Reliability Management. He is the author of the following books on quality management and technology management: Quality Function Deployment (2006), Competing on Quality and Environment (2004), and Managing Green Technologies for Global Competitiveness (1996). Dr. Madu is also the coauthor of Experimental Statistical Designs and Analysis in Simulation Modeling (1993), Strategic Total Quality Management (1995), and ERP and Supply Chain Management (2004). He has over one hundred refereed publications. Department of Management and Management Science, Lubin School of Business, Pace University, 1 Pace Plaza, New York, N.Y. 10038, U. S. A. E-mail: cmadu@pace.edu Janice Winch is an associate professor of management science in the Lubin School of Business at Pace University, New York. She has a PhD in Operations Research from Rutgers University. Her publications include articles in Manufacturing and Service Operations Management, International Journal of Production Research, and Probability in Engineering and Informational Sciences. She is a member of Institute for Operations Research and the Management Sciences (INFORMS) and Decision Sciences Institute (DSI). Her research interests include production scheduling, simulation, supply chain management, and applied queueing theory. Department of Management and Management Science, Lubin School of Business, Pace University, 1 Pace Plaza, New York, N.Y. 10038, U. S. A. E-mail: jwinch@pace.edu

Supply Chain Quality Management: A Simulation Study

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Supply Chain Quality Management: A Simulation Study

Uploaded by

Copyright:

Available Formats

Information and Management Sciences Volume 19, Number 1, pp.