You are on page 1of 32

Accepted Manuscript

Six Sigma Performance for Non-Normal Processes
Tariq Aldowaisan , Mustapha Nourelfath , Jawad Hassan
PII:
DOI:
Reference:

S0377-2217(15)00550-0
10.1016/j.ejor.2015.06.036
EOR 13043

To appear in:

European Journal of Operational Research

Received date:
Revised date:
Accepted date:

8 February 2015
2 May 2015
11 June 2015

Please cite this article as: Tariq Aldowaisan , Mustapha Nourelfath , Jawad Hassan , Six Sigma
Performance for Non-Normal Processes, European Journal of Operational Research (2015), doi:
10.1016/j.ejor.2015.06.036

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Accepted Paper for Publication in European Journal of Operational Research, June 2015

Highlights
We analyze Six Sigma performance for non-normal processes.
We examine changes in failure rates for exponential, Gamma and Weibull processes.
Higher quality improvement effort may be required for non-normal processes.
Reporting the sigma level as an indication of the quality can be misleading.
Wrong Six Sigma projects can be selected when systematically assuming normality.

AC

CE

PT

ED

M

AN
US

CR
IP
T





1

ACCEPTED MANUSCRIPT
Accepted Paper for Publication in European Journal of Operational Research, June 2015

Six Sigma Performance for Non-Normal Processes
Tariq Aldowaisana, Mustapha Nourelfathb, Jawad Hassana
a

Interuniversity Research Centre on Enterprise Networks, Logistics and Transportation
Department of Mechanical Engineering, Laval University
Quebec (QC.) G1V 0A6, Canada

CR
IP
T

b

Department of Industrial and Management Systems Engineering
College of Engineering and Petroleum, Kuwait University
P.O. Box 5969, Safat 13060, Kuwait

AN
US

E-mail addresses: tariq.aldowaisan@ku.edu.kw (T. Aldowaisan); mustapha.nourelfath@cirrelt.ca (M. Nourelfath);
j.hassan@ku.edu.kw (J. Hassan).

Abstract

Six Sigma is a widely used method to improve processes from various industry sectors. The
target failure rate for Six Sigma projects is 3.4 parts per million or 2 parts per billion. In this

M

paper, we show that when a process is exponential, attaining such performances may require a
larger reduction in variation (i.e., greater quality-improvement effort). Additionally, identifying

ED

whether the process data are of non-normal distribution is important to more accurately estimate
the effort required to improve the process. A key finding of this study is that, for a low k level,

PT

the amount of variation reduction required to improve an exponentially distributed process is less
than that of a normally distributed process. On the other hand, for a higher k level, the reverse

CE

scenario is the case. This study is also analyses processes following Gamma and Weibull
distributions, and the results further support our concern that simply reporting the sigma level as

AC

an indication of the quality of a product or process can be misleading. Two optimization models
are developed to illustrate the effect of underestimating the quality-improvement effort on the
optimal solution to minimize cost. In conclusion, the classical and widely used assumption of a
normally distributed process may lead to implementation of quality-improvement strategies or
the selection of Six Sigma projects that are based on erroneous solutions.
Keywords: Quality management, Six Sigma, non-normal distributions, optimization, project
selection.
2

English and Taylor (1993) provide a good study on the robustness of the normality assumption when the distribution is not normal. For example.4 parts per million (ppm) or 2 parts per billion (ppb).. fill rates and the probability to meet specified customer service levels are not governed by normal distributions (Nourelfath. A major drawback of this trial-and-error approach is the lack of physical meaning in the resulting 3 . e. in particular. tools. in project management where a project will more PT likely finish late. for service or transaction systems.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. Montgomery and Runger (2011) studied the error associated with non-normality in the estimation of defects in parts per million. AlliedSignal had savings of $1. and a 10 to 30% capital reduction” for each sigma improvement (Harry and Schroeder. a 20% margin CR IP T improvement. this assumption remains widely used and recognized in the field of Quality. the normality assumption does not hold. the authors recommended applying transformational functions to the underlying non-normal variable until a normal transformed variable is found. Identified in a survey of the relevant literature. robots).g. Additionally. Another example is the procurement process of oil and gas companies. Although not all manufacturing processes follow a normal distribution. Invariably.. where it was observed that cycle time data more closely resemble a Gamma distribution rather than a normal AC distribution. the normality assumption ED comes into question. June 2015 1 Introduction Since its launch in the 1980s. even in manufacturing ones. which have become increasingly predominant in various applications. 2011). Important assumptions in Six Sigma AN US are that the process is normal and its specifications are two-sided. GE reported 11% revenue and 13% earnings increases from 1996 to 1998. depending on certain assumptions. However. machines. Six Sigma has generated great expectations and hype in the business world. this is the case when the number of activities is too small to assume a normal distribution based on the central limit CE theorem. and Harry and Schroeder claimed “10% net income improvement. in supply chain management and goods production systems. These business gains were based on management and technical methods for improving processes with a theoretical aim of a failure rate of 3.5 Billion in 1998.g. 1999). The assumption of normal distribution is generally appropriate for many technological manufacturing systems (e. the normality test should be a first step in any approach using the normality M assumption. driven largely by reports of significant business gains.

and Section 5 provides our concluding remarks. The remainder of this paper is organized as follows: Section 2 presents the relevant literature. we first analyse the amount of variation reduction required to improve the process. The main objective of this paper is to evaluate non-normally distributed data in Six Sigma CR IP T performance methodology. as often is the case in service applications. Then. This ED could lead to erroneous estimation of the efforts and costs to achieve that goal when the process is not normal. 1989. The contribution of this research is that it provides guidelines to help managers to more accurately estimate the efforts required to PT achieve the performance goal of Six Sigma. and the second model addresses the optimal selection of processimprovement strategies. 1996. These optimization models analyse the impact of a poor estimation of quality-improvement effort on optimal solutions to minimize costs. June 2015 transformed variable. Section 4 presents two optimization models to illustrate the effect of inaccurate estimations of probability distributions. Gamma and Weibull distribution processes. 1992). Rodriguez. The efforts required to achieve the performance goal of Six Sigma are then compared between the normal and exponentially distributed cases. Clements. Other researchers suggested fitting the appropriate distribution to the data and then calculating the parameters that would yield the desired ppm (Somerville and Montgomery. AC Section 3 analyses the Six Sigma performance methodology in exponential. Using the exponential distribution case. 1996. two optimization models are developed: the goal of the first model is to AN US find the optimal trade-off between quality-improvement program costs and costs incurred as a result of poor quality. 4 . Finally. M It is often the case in the professional and academic communities of quality to set 3.4 ppm as the goal of Six Sigma without due consideration for the underlying process distribution. Farnum. we generalize our study to processes using Gamma and Weibull distributions.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. rendering it unappealing to non-academic practitioners. and analyse the robustness of proposed quality CE optimization models under inaccurate probability distributions.

Service organizations in the USA and Europe are front-runners in implementation of Six Sigma. They include a hospital. Exploratory empirical evidence was provided through many case studies of organisations. Additionally. (1995) presented a case study to illustrate how the concept of zero defects. The major findings of the authors include an understanding about the suitability of Six Sigma implementation in service organisations. Their findings also showed that organizations from the healthcare and banking sectors have largely implemented the Six Sigma methodology. represent the life-blood of the modern economy (Aboelmaged. 2012). The second concerns approaches to the evaluation of Six Sigma performance without assuming normal probability distributions. a consultancy service and a PT hotel. arguably. we discuss each of these CR IP T two research areas. can be applied to the measurement of customer satisfaction and to examine the impact of customer expectations on the company’s strategies for improving satisfaction. are gradually incorporating Six 5 . Behara et al. as confirmed by many literature reviews. with Asian countries such as India and Singapore following behind. a public service organisation. such as information technology. Chakraborty and Chuan (2013) CE highlighted that “top management commitment and involvement” are indeed important and critical factors for success. nevertheless. as measured by Six Sigma.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. 2010. AC The authors investigated the implementation of Six Sigma in service organizations through questionnaire surveys. and other services. telecommunication services and transport services. Next. relatively few studies have reported the successful implementation of Six Sigma in AN US service systems that. Chakraborty and Chuan (2012) suggested that the implementation and impact of Six Sigma can be affected by contextual factors such as service ED types. June 2015 2 Literature Review There are two main bodies of literature that are related to our research. They found that Six Sigma implementation in service organizations is limited across the globe. and the continuity of Six Sigma very much depends on these factors. The information presented is based on actual studies conducted for a M high‐tech manufacturing company. The first is the literature of Six Sigma implementation in service systems that are more often characterized by non-normal distributions. Several studies have been conducted in regard to implementation of the Six Sigma methodology in large manufacturing companies. Stone. The authors argued that management support and team member support are primary factors for success.

June 2015 Sigma. linking Six Sigma to business strategy. Talib et al. common myths. Setijono (2010) presented a method of estimating left- CE side and right-side Sigma levels. (2013) empirically investigated the relationship between total quality management practices and quality performance. Pan et al. the primary data were “replicated” by first generating random numbers that followed a normal standard distribution and then re-calculating these random numbers with the mean. Simulation technique was then applied to generate a larger amount of secondary data as the basis for estimating left-side and right-side Sigma levels.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. In the context of M Indian service companies. project management skills. They developed a model based on service quality principles to explain the outcomes of BPI adoption at the operational level. Because most service quality data follow a non-normal distribution. (2007) presented some of the most common challenges. ED Our analysis of the literature regarding Six Sigma performance without the normality assumption reveals that this assumption is usually “taken for granted” without a proper inquiry PT into whether this assumption has been fulfilled or not. difficulties. customer focus. The calculated Sigma levels suggested that the developer’s performance was still far below Six Sigma level of performance. development and deployment of Six Sigma. By analysing case studies of financial institutions in Thailand. and implementation issues in the application of Six Sigma in service industry settings. Antony et al. They showed that it is important to link Six Sigma to the strategic as well as the operational level. The results of their study showed that the majority of service organizations in the UK have been engaged in a Six Sigma initiative for just over three CR IP T years. Given that critical attributes may contain data sets that are non-normally distributed. organizational infrastructure. This method was applied in a Swedish house-building construction project. Based on the non-normal process 6 . standard AC deviation. Management commitment and involvement. (2010) developed a new key performance index (KPI) and its interval estimation for measuring the service quality from customers’ perceptions. Buavaraporn and Tannock (2013) formulated relationships between specific business process-improvement (BPI) initiatives and customer satisfaction. Chakraborty and Leyer (2013) developed a framework that defined organizational conditions by which to implement Six Sigma in financial AN US service companies. and the skewness of the primary data. to fulfil the assumption of normality. They also discussed the benefits of Six Sigma in service organizations and presented the results of a Six Sigma pilot survey in the UK. In this method. and understanding of the Six Sigma methodology were identified as the most critical factors for the successful introduction.

(2008). The performance of the proposed method was validated by a simulation study for different non-normal distributions. Chou et al. (1998) PT and Amiri et al. The results showed that the new KPI allowed practicing managers to evaluate the actual service quality level delivered and to prioritize the possible improvement projects from customers’ perspectives. engineers were able to determine the actual process capability more accurately. Other existing approaches consist of transforming data such that they satisfy normality conditions. CE For example. Using the adjusted process capability formula in Hsu et al. a new KPI suitable for measuring service quality was developed.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. Hsu et al. and the BOX-COX method to estimate performance indices of non-normal processes. Furthermore. A real-world semi-conductor AN US production plant was investigated and presented to illustrate the applicability of the proposed approach. compared with the traditional CR IP T method of sample size determination. For example. Our literature review also shows that the 7 . The process capability index also can be evaluated using approximate heuristic approaches. unilateral specification with target value. The above-mentioned papers confirm that Six Sigma theory and implementation have not been sufficiently studied for service processes. ED (2008) compared these methods and concluded that estimations by Burr’s method yielded better results that are closer to the real performance factors. June 2015 capability indices used in manufacturing industries. Burr’s method. Only a few papers have dealt with this important issue. and unilateral specification without target value. Ahmad et al. There are also some approaches using Clements’s method. In addition. in Abbasi (2009). an artificial neural network is proposed to estimate the process capability index (PCI) for right skewed distributions without appeal to the probability distribution AC function of the process. kurtosis and upper specification limit as input variables. (2008) considered the problem of determining the adjustments for process capability with mean shift when data follow a Gamma distribution. The proposed neural network estimated PCI using skewness. (2012) considered data transformations to convert non-normal data into normal data. A challenge to applying Six Sigma methodology in service processes is the fact that. the confidence interval of the proposed KPI was established using the bootstrapping method. the underlying processes are non-normal. it was shown that a substantial amount of cost savings could be expected using the suggested sample sizes. Pan and Wu (1997) developed new performance indicators for non-normal data by breaking quality characteristics into bilateral specifications. in most cases. They proposed formulas for performance measurement and put forward a flowchart for calculating performance indices of normal and non- M normal data.

CE Step 4 – Analysis of the effect of inaccurately estimating the quality effort: For this. we evaluate the Six Sigma performance using exponential. Step 1 – Illustration of the Six Sigma failure rate goal of 3. two optimization models are developed. this is the first time that the Six Sigma performance methodology has been evaluated to analyse the robustness of quality optimization models under inaccurate probability distributions. They are proposed to analyse the robustness of quality improvement strategies. Guidelines are provided to help managers to more accurately estimate the efforts required to achieve the performance goal of CR IP T Six Sigma. Next. M Step 2 . Our work rather develops an approach for Six Sigma performance evaluation without assuming normal probability distributions. June 2015 majority of the existing studies are based on the normality assumption. Gamma and Weibull distributions.4 ppm or 2 ppb: This is based on the use of a simple example from the manufacturing industry. this analysis is based on an analytical evaluation and a ED comparison of the values obtained for exponential and normal distributions. First.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. Step 3 – Generalization of the analysis: PT The results of Step 2 are extended by considering Gamma and Weibull processes. Our first model is a step-loss quality model. 8 . and AC the second model addresses a problem that arises with Six Sigma project selection. The next Sections detail each of the four steps above.Analysis of the Six Sigma performance for exponential processes: Using exponentially distributed data. AN US The proposed methodology can be summarized by the following steps. two optimization models are developed to analyse the effect of inaccurately estimating the quality effort. To our knowledge. The contributions of this study are twofold. These optimization models are presented to support our analysis of Six Sigma performance.

Any shaft produced outside the 1.06 lower and upper AN US specification limits (LSL and USL) is denoted as defective (Figure 1).1 Six Sigma Performance Goals: an illustrative example The 3. in the parabolic view of quality.94 and 2.00±0.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. The following example from the manufacturing industry sheds some light on these performance goals.06 CE PT ED 1. AC Defective LSL Defective TV USL Figure 1: The parabolic view of quality.00 2. June 2015 3 Six Sigma performance analysis 3. However. In this example. These goals are founded on two key assumptions: the process under analysis is normally distributed.06 cm diameter specifications.94 M (nominal) value (TV) (and away from the upper and lower specification limits).4 ppm and 2 ppb process failure rates are major performance goals of Six Sigma. CR IP T and the process specifications are two-sided. a manufacturer needs to produce cylindrical shafts with 2. 9 . the process performance increases as the diameter of the manufactured shaft is closer to the target 2. a shaft produced with a diameter that falls anywhere within the specification limits is considered non-defective. In the traditional goal-post or binary view of quality.

94  6 0. This results in a 6 process with a failure rate of 2 AN US ppb.e. the process is said to be a 3 process (i. For statistical references.00 cm (equal to the TV) and a standard deviation () of 0.06  2.. The assumption of a normal distribution is generally appropriate for machine manufacturing.00  1. for example. thereby reducing the process variation  to 0. we calculate . however. amount of variation on either side of TV) with a calculated failure rate of 2. 6 process AC CE PT 3 process Figure 2: Normal 3 and 6 processes. as illustrated in Figure 2. Because many professionals do not confirm normality. as follows: USL  TV   TV  LSL    2.02 CR IP T  Therefore.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. then  = 12. the difference between LSL or USL and TV in terms of the process variation . for service or transaction systems. the normality assumption comes into question. or calculate the difference between the 10 .00   2.02 cm. see. June 2015 Assuming that the cylindrical shaft manufacturing process follows a normal distribution with a mean of 2. Montgomery and ED M Runger (2011) or Chandra (2001). If the engineer can improve the quality of the cylindrical shaft-making process.700 ppm using the normal distribution function.01 cm.

5 process shift to the right.5 shift to the right.810 ppm. Figure 3 shows the effect of a 1. consequently. their specification limits are not necessarily two-sided. 11 . different failure rates. However.2 Non-Normal Six Sigma As previously mentioned. 6 process ED M AN US Shifted 6 process PT Figure 3: A normal 6 process with a 1. This would generally yield a different approach and.5 allowance. many transactional and service processes do not follow a normal AC distribution. we show that for certain applications.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. in this section. Moreover.5 on either side of the target value (Bothe. Such non-normality is commonly accounted for in the Six Sigma projects using the ±1.5 process with a failure rate of 3. This leads to  = 9. a 4.4 ppm. a common practice is to account for normality deviation by allowing for a shift of ±1. Note that such a shift has a greater consequence in the case of 3 processes. CE 3.5 is not always adequate. 2002). a seemingly ample allowance of 1. June 2015 process mean and the target value. resulting in CR IP T a defect rate of 66.

This concept also requires that the one-sided normal be redefined so that TV is not equal to the mean of 15 days. the standard deviation equals the mean.5 0.5 45.5 5 3 0.002 2 AC CE 3 PT k 12 .335 7.500 18.311 135. compares much less favourably against the 6 normal process claim of 0.700 ppm in the normally distributed case.56E-6 0.143 2. which industry practice requires be paid within 30 days upon receipt of an accurate bill. a k process should be redefined such that the entire CR IP T deviation.34 335.1 Exponential processes As an example.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. which in this scenario. Because the specification is onesided with a target value (TV) of zero.002 ppm M (Table 1). This performance compares favourably against the 3 process of 2. For the exponential distribution. and therefore the total failure rate is also on one side of the TV. and the resulting 3 AN US process has a failure rate of 2. is one-sided. This process could be modelled by an exponential distribution with a mean of 5 days. take the payment by oil companies of their suppliers’ bills.479 ppm.33E-9 0.832 8 1.5 0 0.700 2. Furthermore.  (days) Normal (ppm) Exponential (ppm) 1 15 317.75 63.316 5 2.015 10 1.144 7 2.875 1.573 45.4 6 2. June 2015 3.002 6. ED Table 1: Normal and Exponential k processes. we obtain a 6 process with a failure rate of 6. 2k.5 days.479 4 3.667 0 0.113 9 1.2.144 ppm. to improve the process average to 2.

5 days as shown in AC Figure (5). However. then.5 shift in the process mean.5 shift in the process mean to the left is inconsequential because the mean would be below the TV of zero. Figure 4 shows the behaviour of k processes for the two distribution types on a logarithmic CR IP T scale.5 to the right from its current value of 2. any shift to CE the left side should not be considered because it is in the direction of improvement. a 10 process needs to be targeted with mean and standard deviation of approximately 1. a 1. if we assume a 1. this would produce a failure rate of 8.5 days. and in all cases.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. if we were to allow the mean to shift 1.230 ppm (Table 2). By applying a similar analysis to the 3 process. the calculated failure rate is 90. the situation reverses with an increasingly greater difference in failure rates. It is interesting to note that the exponential distribution has a lower failure rate than the normal distribution up to approximately 3.0001 0. 13 . normal 1000000 exponential Fallout Rate 100 1 0.01 0 1 2 0. then the direction of shift becomes an PT important consideration. For the 6 exponential case. Now.000001 1E-10 3 4 5 6 7 8 9 10 k M 1E-08 AN US 10000 ED Figure 4: Failure rate behaviour of normal and exponential distributions. June 2015 Table (1) also demonstrates that for the normal distribution 6 goal of 2 ppb to be valid for an exponential distribution.720 ppm for the shifted mean exponential.

Table 2: Normal and Exponential k processes for the case of a 1.770 201. .5 3.  Shifted Normal Shifted Exponential (days) (ppm) (ppm) 1 15 697.75 6.698 CE 7 PT 6 ED M k 1.5 .810 90.5 308.019 3.210 40. . .5 shifted mean.75 0. 15.19E-8 747 10 1.897 3 5 66. .718 4 3.667 3. .4 ppm to be valid for an exponential distribution.662 9 1.5 0 335.762 5 3 232.316 2. . .230 2.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research.143 0.672 449. a 15.4 8.6 18.875 4. for the shifted 6 normally distributed goal of 3.9523 0 3.4 AC 8 From Table (2).75 process must be targeted with a mean and standard deviation 14 .329 2 7. June 2015 CR IP T 6 process Shifted 6 process AN US Figure 5: Exponential 6 and shifted 6 processes.02E-5 1.

and then the situation reverses with an increasingly greater difference in failure rates. in the un-shifted case. Figure (6) shows that the exponential distribution has a lower failure rate than the normal distribution up to somewhere between 2 and 3.000001 1E-08 M Figure 6. For example. The variation reduction required is even larger for the shifted case (more than four times). However.0001 AN US 0. 15 .9523 days. as defined by the amount of reduction in .01 0 1 2 3 4 5 6 7 CR IP T 1000000 8 9 10 k 0. normal shift exponential shift 10000 Fallout Rate 100 1 0. exponentially distributed data from a process performing at less than 3 would achieve PT improvement faster than normally distributed data. June 2015 of approximately 0. From the analysis of the exponential distribution example.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. required to improve performance. it is clear that one must first ED determine the distribution of the underlying data in order to estimate the efforts. It is AC important to emphasize that higher sigma processes for a given distribution are generally more difficult to improve than lower ones. Similar to Figure (4). if the process is performing at higher than 3 and the target is to achieve the goal 6 failure rate of 2ppb. then the exponential CE process would require more than twice the variation reduction as that of a normal distribution. Failure rate behaviour of normal and exponential distributions for the shifted case. which would further add to the total effort required to achieve Six Sigma performance level goals.

the failure rate is greater than 25% for a 6 Sigma level of performance. which consists of minimizing the total cost.2 Gamma and Weibull processes The tables given in the Appendix show the failure rates for Gamma and Weibull distributions with respect to the sigma level. For example. two optimization models are developed to analyse the effect of inaccurate estimations of probability distributions on the optimal solutions. As shown in the Appendix. We need to select the most appropriate alternative. For example.1 Quality cost model ED quality model. The first model is a step-loss PT 4.1 Mathematical model This optimization model is voluntarily simplistic. June 2015 3. and the second addresses a problem in Six Sigma project selection.2. On the one hand. a 6 sigma quality-improvement strategy is more expensive than that for 5 sigma. for a quality characteristic that follows the Weibull distribution with shape parameter 3. These results support our concern that simply reporting the sigma level as an indication of the quality of a product or process can be misleading. Consider that many quality-improvement strategies are available. selecting the best quality-improvement strategy is 16 . It is developed for a preliminary illustration AC of the effect of underestimating the effort required to improve quality on the total cost. The objective function. This is Weibull distribution performance that is worse than the 2 ppb failure rate for the normal AN US distribution. CE 4. the failure rate with respect to the sigma level for both distributions is affected only by the shape parameter  and the scale parameter  has no effect. Each strategy is characterized by cost and a potential quality level. Of note. is the sum of (1) the cost incurred in deploying the selected quality-improvement strategy and (2) the cost incurred as a result of defective items or services.1.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. 4 Optimization Models M In this section. the failure rate for the Weibull distribution is much more sensitive to CR IP T the value  than the failure rate for the Gamma distribution.

and r as the rejection costs. The goal of the optimization model is to determine the most favourable trade-off between qualityimprovement costs and poor-quality costs. 1 . Constraint (4) specifies that the decision variable is binary. On the other hand. June 2015 more expensive but leads to fewer defective items or services. …. The objective function (2) is 17 .ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. i = 1. Loss function (1) is consistent with the idea that all values within the specification range are equally desirable (binary view of quality). 2. We use a step-loss function based on a conformance quality definition of a quality characteristic 0 L Y    r CR IP T being acceptable if it falls inside predetermined specification limits: if LSL  Y  USL. otherwise (1) with L(Y) as the loss associated with quality characteristic Y. n (4) i 1 Constraint (3) specifies that only one quality-improvement strategy can be selected. The following notations are used: number of possible quality-improvement strategies ci: yearly cost of quality-improvement strategy i FRi: failure rate under quality-improvement strategy i Q: number of services or products per year r: unitary rejection cost PT ED M n: Xi: decision variable. equal to 1 if quality-improvement strategy i is implemented (0 otherwise) CE The mathematical model is: AC Minimize n  X i  ci  r FRi Q  (2) i 1 n Subject to  Xi  1 (3) X i 0. LSL and USL as the lower and AN US upper specification limits. less expensive improvement strategies lead to higher costs as a result of poor-quality items or services.

the optimal solution corresponds to a 9 Sigma AC level with a total cost of $2. this is equal to 3. For each i 1 fraction of a sigma improvement. and this corresponds n to a cost.000 for an exponential process (instead of $403. represents an estimation of the annual cost incurred CR IP T as a result of poor quality.1. We assume that the unitary rejection cost is r = $10.2 Numerical example Consider that 10 quality-improvement strategies (n = 10) are possible.230 ppm for a shifted exponentially distributed one AN US (see Table 2). On the other hand. 4. for a 6 sigma quality level. in ED most cases. The first one. The failure rate FRi depends on the selected quality level. a quality effort needs to be implemented. However. if the probability distribution is instead exponential. June 2015 n composed of two terms. 10). repair or scrapping.000. rework.  X i  FRi i 1 r Q  . Table 3 presents the costs of each k process with corresponding failure rates. the actual total cost if we consider a 6 Sigma level is $8. this may correspond to the costs of annual production.400 calculated under the normal assumption). In service systems. more effort will be required to improve a process from a 5 sigma to 6 sigma level compared to improving a process from a 3 sigma to 4 sigma level. etc. is the annual cost spent on quality. For example. this may correspond to any cost incurred because of failure to fulfil the service on time. The costs in Table 3 correspond with the fact that.  X i ci .000.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research.630.400. The second term. For tangible products. The results are presented in Table 4 and show that when we consider a normal distribution. ….4 ppm for a shifted normally distributed process and to 8. 18 . The average number of PT services per year is Q = 100. CE the optimal solution is found for a 6 Sigma level with a total cost of $403. Each strategy i M corresponds to a k process in Table 2 (k = i = 1.847.000.

000 90.100.335.5 400.000 8 1.000.000 232.922.762 18.962.000 6.000 4 6.5 4.795.000 9 2. Shifted Normal Shifted Exponential i Total cost ($) Total cost ($) 697.329 7.000 201.897 3 5 50.862.4 7 2.000 3.682.5 25.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research.662 M i Table 4: Total costs and optimal solutions.300.000 4.398.847.000 5 432.875 1.5 shifted mean).000 7 700.000 66.100.770 201.667 2.339.310.000 3.000 3 66.718 4 3.210 5 3 200.000 2.500 AC CE PT 1 ED Quality strategy 19 .400 8.000 4.000 697.860.698 AN US  (days) 1.143 700.316 8. June 2015 Shifted Normal Shifted Exponential FRi (ppm) FRi (ppm) 10.019 8 1.19E-8 747 10 1.516.000 6 403.000 449.6 6 2.000.000 2.230 3.300.600 18.5 ci ($) 1 15 2 40.768.000 0.672 449.630.75 100.000 4.000 40.810 90.000 2 308.000 10 4.000 0 335.02E-5 CR IP T Table 3: Costs and failure rates for Normal and Exponential k processes (1. 9 1.000 308.

it is not unusual to select a strategy with higher failure cost and lower prevention cost if it minimizes the Total Cost (i. 1. reputation.2 Six Sigma project selection model ED 4. improvement projects). many organizations managed to estimate the various components of CR IP T quality costs. image. airplanes). n j is characterized by a cost and a quality level. We consider that the objective consists of maximizing the total profit. and Appraisal and Prevention costs (e. June 2015 The Total Quality Cost of the objective function in Model (2)-(4) consists of: Failure/Rejection costs (e. .g.g. We need to select the most suitable set of Six Sigma projects and the corresponding quality levels. This remark was explicitly pointed out for example by Juran and Gryna (1993) and Kubiak and Benbow (2009). scrap. In this case. the lowest failure rate is preferred as long as the available quality budget allows this. reviews.e. it is a goal that is not possible to achieve. M 4.g. audit.g. selecting many processes and the best quality-improvement projects is more expensive but leads to fewer defective items or 20 . investigation. correction. as defined by the selling price minus the total cost. the Failure/Rejection. even in life threatening operations (e.2.1 Mathematical model This optimization model is developed to highlight the effect of assuming a normal PT distribution on the selection process of Six Sigma projects. Note that AC project 0 corresponds to the case where no quality improvement is implemented (status quo). Therefore. It remains that for companies that are very AN US sensitive to service level. the failure cost will be relatively high (because the estimated costs related to loss of customers. loss of customers). It is true that not all costs are easy to quantify (e. The latter is the sum of (1) the cost incurred in implementing the selected quality projects for the selected processes and (2) the cost incurred because of defective items or services in all selected projects. let us consider a set of processes with nj quality-improvement candidate projects for each process. reputation. warranty. 2. loss of customers). rework. Because many of the most successful manufacturing and oil and gas companies implement thousands of Six Sigma projects every year. As for zero defects. inspection. image and reputation are very high). plus Prevention and Appraisal Costs). However.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. On the one hand. CE   Each project i i  0.

1..ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. while taking into account the trade-off between quality-improvement and poor-quality costs. 1 .. 2. .. i  0. as the objective is to maximize the total profit. m (8) j 1 i 0 21 . j  1.. the selling price of the conforming items is also taken into account. j 1 i 0 j  1.. equal to 1 if project i is implemented for a selected process j (0 otherwise) ED M AN US m: PT The mathematical model is: m nj Profit = CE Maximize n AC Subject to   m nj   X ij u j Q j 1  FRij   X ij cij  rj FRijQ j j 1 i 0  X ij  1. fewer selected processes and less expensive projects lead to higher costs incurred as a result of poor quality. The goal of the optimization model is to select the appropriate set of processes/projects and quality levels. On the other hand. m  (5) (6) i 0 m nj   X ij cij  B (7) X ij 0. . CR IP T The following notations are used: number of processes as candidates for Six Sigma improvement nj: number of possible quality-improvement projects for a process j cij: yearly cost of quality-improvement project i for process j FRij: failure rate under quality-improvement project i for process j Qj: number of services or products per year for process j rj : unitary rejection cost for process j uj: unit selling price of the final product or service for process j B: budget available for Six Sigma projects Xij: decision variable. Indeed. . n j . June 2015 services..

718 3. the first project corresponds to a 3 Sigma quality level.2. The failure rates of normal and exponential distributions are grouped in Table 5. Table 5: Failure rates used in Example 4. is the annual cost of quality for the selected projects. Therefore. Constraint (6) specifies that for each process. m nj   X ij cij .672 449. j 1 i 1 is the total revenue generated when units are sold or services are fulfilled. All other data are in Table 6.329 1 (Three Sigma) 66. The last term.230 Quality project i 2 (Six Sigma) 22 . For both processes.2. only one qualityimprovement project can be selected. M 4. CR IP T m nj represents an estimation of the annual cost incurred as a result of poor quality j 1 i 1 for all selected projects. The AC CE available budget is B = $200.810 90.2. constraint (7) guarantees that the budget limit is not exceeded.000. The second one. and the second is a 6 Sigma PT quality level. Thus. 2) and that 2 potential quality-improvement projects are possible (i = 1.4 8. The last constraint specifies that the decision variables are binary. June 2015 m nj   The objective function (5) is composed of three terms.  X ij u j Q j 1  FRij . The first term. Shifted Normal Shifted Exponential Failure rate in ppm Failure rate in ppm 0 (One Sigma) 697. Constraint (7) specifies the budget constraint. 2). most companies do not have unlimited funds to implement process-improvement projects. In fact.2 Numerical example Consider that we must choose between 2 candidate processes or operations for improvement ED (j = 1.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. We assume that the current situation (i = 0) corresponds to a One Sigma quality level. j 1 i 1   X ij rj FRij Q j . AN US there is generally a limited budget allocated to all Six Sigma projects.

000 120. and Process 2 with Project of Quality level 1 Infeasible Process 1 with Project of Quality level 2.030 Only Process 2 with Project of Quality level 1 1.706.592. respectively.592.000 2 200. of items or Unit. Profit (in $) Status quo -416.000 0 0 120. rejection Unit selling (i) in $ services (Qj) cost (rj) in $ price (uj) in $ 0 0 100.030 Only Process 1 with Project of Quality level 2 1.000 12 20 1 10.000 1 5. and Process 2 with Project of Quality level 2 Infeasible 23 .197 Process 1 with Project of Quality level 2. and Process 2 with Project of Quality level 2 3. ED Table 7: Results when normal distribution processes are assumed.2.786.000 2 150.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research.777.728 PT Selected processes and projects 1.000 CR IP T 1 Quality project Cost (cij) 12 20 12 20 10 18 10 18 10 18 AN US Process (j) M The results are given in Tables 7 and 8 for normal and exponential distributions.726 AC CE Only Process 1 with Project of Quality level 1 Process 1 with Project of Quality level 1.000 100. and Process 2 with Project of Quality level 1 3.000 120.968 Only Process 2 with Project of Quality level 2 1.438 Process 1 with Project of Quality level 1.2. 2 No.000 100.697. June 2015 Table 6: Data for Example 4.

402 is realized for the exponential distribution case.728). and Process 2 with Project of Quality level 1 Infeasible Process 1 with Project of Quality level 2. the model instead selects Process 1 with AC - Quality level 1 and Process 2 with Quality level 1. ED M - This is because the failure rate of an exponentially distributed process is less than that of a normal process for less than 3 Sigma quality levels. On the other hand. - A profit of $1.197. 24 . this same solution is infeasible for exponential processes under the fixed budget constraint. and Process 2 with Project of Quality level 1 3. project) selection for Process 2.e. Profit (in $) Status quo 1.402 Only Process 1 with Project of Quality level 1 2. PT the situation reverses.890 AN US Process 1 with Project of Quality level 1. June 2015 Table 8: Results considering exponential distribution processes. for 3 and 6 Sigma levels. When distributions are considered exponential. the optimization model selects Process 1 with Quality level 1 CE - and Process 2 with Quality level 2.534.957 Only Process 1 with Project of Quality level 2 Infeasible CR IP T Selected processes and projects Only Process 2 with Project of Quality level 1 2.335 Only Process 2 with Project of Quality level 2 Infeasible Process 1 with Project of Quality level 1.786.212.$416.339. The aforementioned difference is the quality level (i. We conclude the following from this example: Under the normality assumption.. The optimal profit realized under the normal distribution is 3.212. and Process 2 with Project of Quality level 2 Infeasible We note the following for the status quo situation (One Sigma level): A loss is observed under the normal distribution assumption (.407.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. and Process 2 with Project of Quality level 2 Infeasible Process 1 with Project of Quality level 2.

However. once data are collected. Weibull and Gamma 25 . CR IP T 5 Conclusions In the existing literature. Six Sigma theory and implementation have not been sufficiently studied for service processes. Managerial insights were provided through the paper using many illustrative numerical examples. by using the exponential distribution CE rather than the normal distribution. the majority of the existing studies are based on the normality assumption. That is. in most cases. this paper has developed an approach for Six Sigma performance evaluation without assuming normal probability distributions. two optimization models were developed to analyse the effect of inaccurately estimating the quality effort. the contributions of this study are twofold. the underlying processes are non-normal. we have evaluated the Six Sigma performance using exponential. Unlike the prior literature.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. the variation reduction required to improve a process at a lower than given k level is less than that when the process is beyond that level. We have shown that achieving the performance goal of Six Sigma in the exponential. The consequences of non-normality on the failure rate were analysed for Gamma and Weibull distributions. we AC have shown that achieving the performance goal of Six Sigma when the process is exponential is more demanding than that for the normally distributed case. this is the first time that ED the Six Sigma performance methodology has been evaluated to analyse the robustness of quality optimization models under inaccurate probability distributions. a special emphasis must be put in identifying the right probability distribution. A challenge to applying Six Sigma methodology in service processes is the fact AN US that. Only a relatively small number of papers have dealt with this important issue. it reasonable that the distribution type is accounted for and specifically calculated when an organization plans for its next Six Sigma initiative. In comparison to the existing studies. instead of merely assuming a normal distribution as it is often the case in the current practice. To the best of our knowledge. Moreover. Gamma and M Weibull distributions. First. Next. These analyses demonstrate that attaining the Six Sigma performance goals require varying different effort levels based on the distribution type. Guidelines were then PT provided to help managers to more accurately estimate the efforts required to achieve the performance goal of Six Sigma. Therefore. We demonstrated that. June 2015 The results above confirm that it is mandatory to accurately estimate the efforts required to achieve the performance goal of Six Sigma.

36. (2008). This research is supported by Kuwait National Petroleum Corporation (KNPC) project CE number XP 01/14. Finally. 26 . Instead. A neural network applied to estimate process capability of non-normal processes. M. 3093–3100.4 ppm or 2 ppb target performance goals made by Six Sigma practitioners. S. & Zeefungsecul.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. That is. (2010). The results indicated that an incorrect estimation of the probability distribution... Aboelmaged. In addition. June 2015 cases requires a greater variation reduction than that for the normally distributed case. 27 (3). P. 642-665. we have analysed the robustness of the optimal solution to minimize cost when an exponential process is assumed to be normal. The purpose of this article is not to advocate for changing the term Six Sigma or cast doubt on its business value. Future work will aim to apply the results of this study to the Commercial Department of a Gas PT Acknowledgement ED and Oil company in Kuwait. (2009).G. using two optimization models. it is possible that less variation reduction is required. B. and thus the quality effort. 49. Process capability calculations for nonnormal quality characteristics a comparison of Clements. This paper does not identify which distribution is better for service systems. Abdollahian. Six Sigma quality: a structured review and implications for future research. Ahmad. ANZIAM Journal. AC References Abbasi. We are currently developing a theoretical model to explain why the normality assumption is highly questionable M for cycle times in service processes. we hope to highlight the effect of distribution types in an effort to promote professionalism and accuracy in regard to the 3. International Journal of Quality & Reliability Management. Expert Systems with Applications. although for less than the Six Sigma level. Burr. no field study is provided to support our claims. and Box-Cox methods. managers need to make significant efforts and a sense of awareness to identify the right probability distribution for each process. M. may lead to CR IP T erroneous solutions when selecting quality programs and Six Sigma projects. 268-317. instead of systematically assuming normal distributions AN US in Six Sigma projects.

A. & Gresham. Case study analysis of Six Sigma implementation in service organisations. M. (1995). (2013)... G. International Journal of Quality & Reliability Management. M. International Journal of Quality & Reliability Management. International Journal of Production Research. International Journal of Lean Six Sigma. Juran.. Chakraborty. 30 (2). (2008). Quality Progress.J. PT Chou Y. Six Sigma: The breakthrough management strategy revolutionizing the world’s top corporations. P.. (2007). 19 (6). McGraw-Hill. (1993). Clements. N. 517–529. Using Johnson curves to describe non-normal process data. Capability adjustment for gamma processes with mean shift consideration in implementing Six Sigma program. B. 27 . (1993). Process capability calculations for non-normal distributions. 31. 18 (6). W.M. Transforming Non-Normal Data to Normality in Statistical Process Control. AC Farnum.H. (3rd ed. Developing a Six Sigma framework: perspectives from financial service companies. Fontenot. & Wu. T. Pearn. 141-170. Scientia Iranica. Quality Engineering. 12 (3). 1894–1905... An empirical analysis on Six Sigma implementation in service organisations. Business process improvement in services: case studies of financial institutions in Thailand.J. 479–487. J. Business Process Management Journal.. Chakraborty. Doubleday. J. CRC Press. CR IP T Behara. 918.M.. Antony. Quality Planning & Analysis. & Tannock. 4 (2).ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. 14 (3).. A. R..L. & Leyer.L. R. Kumar.M. (2013). 24 (3). & Gryna.R. & Mason R. & Schroeder. 30 (3).F. & Chuan.K. June 2015 Amiri. F. AN US Buavaraporn. CE English. International Journal of Quality & Reliability Management. M. & Cho. 294-311. Journal of Quality Technology. (2001). Harry. Bashiri. & Doroudyan. International Journal of Quality & Reliability Management. 95-100. 22. 992-1019. 30 (3).. M Chakraborty. challenges and difficulties. 319-340. J. (2002). (1989). Y. T. (1996). Non-normal multi-response optimization by multivariate process capability index. G. 335-339. Quality Engineering. Inc. & Taylor. M.). H. A. (2013). 9. common myths.. D. Process capability analysis – a robustness study.K. 191.C. Polansky A. & Chuan. A. 256-279. (2012).. M. (2012). ED Chandra.C.. J.. (1999).. Random House. F. 1621-1635. Customer satisfaction measurement and analysis using Six Sigma. Mogouie.. (1998). Hsu...S. J. European Journal of Operational Research.R. Bothe. Statistical Quality Control. empirical observations and success factors. Statistical reason for the 1. M. A.5r shift. Six sigma in service organizations: benefits. 133-141. N.M. Antony.

). 81-88. & Benbow. 112-132. Montgomery. 27 (3). & Runger. A.. Pan. Microelectronics Reliability. D. M.. Quality Engineering.. 318-332. 110 (6). T. The Certified Six Sigma Black Belt.). (2nd ed. 3 (2). International Journal of Lean Six Sigma. An empirical investigation of relationship between total quality management practices and quality performance in Indian service companies. T.N. 421–428.. 30 (3). 28 . (1992).ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. D. Z.C. DePMO.L. Somerville. F. S.N. Pan. (1996). 305-316. (2012). Journal of Quality Technology. 823-840. 176-186. ED Stone. G. 24.W. & Bretholt. International Journal of Quality & Reliability Management. Kuo. CR IP T Nourelfath. John Wiley & Sons. European Journal of Operational Research. (2009). AN US Rodriguez. M. (2013). (2010). & Wu. R. K.. (1997). 280-318. International Journal of Quality & Reliability Management. Process capability indices and non-normal distributions.. Normal approximation through data replication when estimating DisPMO. Process capability analysis for non-normal relay test data. June 2015 Kubiak. (2010).. left-side and right-side Sigma levels from non-normal data. (2011). D.B. & Montgomery. D. & Qureshi. Applied statistics and probability for engineers. 212. (2011). J. Service level robustness in stochastic production planning under random machine breakdowns. S. Industrial Management & Data Systems.L. 9. 37 (3). Recent developments in process capability analysis. ASQ Quality Press. Developing a new key performance index for measuring service quality.. J.M. AC CE PT Talib. Four decades of lean: a systematic literature review. M Setijono. (5th ed. Rahman.

respectively:  x M    f  x. June 2015 APPENDIX In this section.    x  1 x 1e       AN US    CR IP T calculations below.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. and variance for the Gamma  2   2 whereas. the failure rates for the Gamma and Weibull distributions are presented for different sigma levels.     x 1e       ED     1  1   2 AC CE PT   2    1  2      1     1               2 29 . mean. . the scale parameter  is fixed at an arbitrary value of 2 in all the distribution are given by. mean. respectively: f  x. and variance for the Weibull distribution are given by. . The probability density function. Because the failure rates for both distributions are only affected by the value of their shape parameter . the probability density function.

312011 2478.09E-05 7.56922 Failure 3.97367 20.54E-05 1.001944 0.571092 1943.828427 3.000170 4.002005 AN US  USL 8.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research. June 2015 1.828427 3.464102 4 sigma level              Mean 1 2 3 4 5 6 Variance 2 4 6 8 10 12 Std.68936532 10.000150 0.28427 31.464102       6 sigma level              Mean 1 2 3 4 5 6 AC CE  Variance 2 4 6 8 10 12 ED       Mean 1 2 3 4 5 6 PT  Variance 2 4 6 8 10 12 Failure (ppm) 3580.8E-05 6.000769 0.386628432 0.92284241 7.8235242 150.44949 2.752177 2094.414214 2 2.97056 24 29.966576 2005.000335 0. dev.828427 3.834923 1958.003580 0.5041988 45.414214 2 2.057500709 Std.828427 3.162278 3.29822 27.4626279 205.282253 122.464102 USL 14.162278 3. dev.87E-07 2.414214 2 2.78461 CR IP T  30 .485281 12 14.06E-06 5.44949 2.055632298 5. Failure rates for Gamma distribution 3 sigma level              Mean 1 2 3 4 5 6 Variance 2 4 6 8 10 12 Std.24E-07 Failure (ppm) 37.62278 34.000206 0.96389309 6.44949 2.414214 2 2. 1.69694 16.002479 0.766196071 0.06E-06 Failure (ppm) 169.39992976 19.44949 2.0212981 106.94113 37.39388 33.66E-07 3. dev.000106 Failure (ppm) 769.62742 25. 1.423022 USL 11.223636375 M 5 sigma level  Failure 0.000122 0. 1.64102 Failure 0.464102 USL 16.162278 3.4949 28.59592 22.97056 18.001959 0.3695568 335.162278 3.850780182 0.85E-06 7. dev.14E-06 1.94733 41.002095 0.31371 16 19.414285 Std.14214 20 24.144212353 1.97E-05 1. 1.71281 Failure 0.

944272 2 1.39992976 6547.234590382 0.37699E-13 0.576587 0.370488383 Failure (ppm) 0 45.759333 0.074665 5.55418 16 9.805491 1.926503 0.555999 3.53999E-05 0.576587 0. 8.785959 Variance 80 4 1.111997 7.649101 USL 53.1588 690936.225872 0.9542 Mean 4 2 1.774528 1.774528 1.14421E-06 0.759333 0.180118031 0. Failure rates for Weibull distribution 3 sigma level              Mean 4 2 1.225872 0.774528 1.9102 299021. dev.061397259 0.104679491 0.772454 1.502761 0.299021372 0.772454 1.000335463 0.87313 180118.806973 7.25947 234590. 8.649101 USL 71.225872 0.35523 5.468671 54805. 8.649101 USL 89.4911 253956.025269028 0.253956951 Failure (ppm) 0 6.805491 1.593331 6. 8.ACCEPTED MANUSCRIPT Accepted Paper for Publication in European Journal of Operational Research.894603 Failure 2.020280255 0.25452 115651.772454 1.421332 Std.0309 370488.412022 6.805491 1.576587 0.002081569 0.192804 Failure 0 0.421332 Std.805491 1.66563 12 7.9507                                   AC       CE 6 sigma level ED  PT  M 5 sigma level Mean 4 2 1.502761 0.772454 1.147 Mean 4 2 1.926503 0.3716 519333.115651910 0.785959 Variance 80 4 1.3313 24 14.3817 472426.421332 Std.568597 25269.926503 0.926503 0.858407 0.774528 1.502761 0.559017 4.71046 11.472426159 0. dev.054805873 0.144212353 2081.789207 Failure 0 6.491006 Failure 0 4. June 2015 2.858407 0.002478752 0. dev.690936954 Failure (ppm) 2.649101 USL 107.785959 AN US 4 sigma level CR IP T  31 .11803 9.006547469 0.576587 0.858407 0. dev.944272 2 1.3827 Variance 80 4 1.785959 Variance 80 4 1.02772 104679.37699E-07 2478.858407 0.25872 9.265028 7.421332 Std.759333 0.519333147 Failure (ppm) 0 335.752177 61397.502761 0.225872 0.944272 2 1.4626279 20280.759333 0.944272 2 1.44272 20 12.