The board of a high volume manufacturing organization has hired a quality expert to improve their performance and market position. A report has been prepared recommending tools and approaches to be applied to the manufacturing process. The report includes significant impacts that could generate business benefits, the type of expected business impacts, mechanism by which the business impacts would be generated, and the supporting factors necessary to maximise the results.
Historically design and development of products have been focused on customer specifications. However, due to the complex nature of the process of designing and manufacturing products, customer requirements have not been completely satisfied. Customer requirements have been muddled over precedent, capability of production, and technological capability. Today it has become all the more imperative for organisations to be customer focussed, as customers are likely to exercise the best choices available to them. A simplistic view of the complex nature of customer satisfaction has been illustrated in the Kano quality model (see appendix a, figure 1). Spoken performance involves satisfying stated customer requirements. Basic quality involves customer assumption that requirements that have not been explicitly stated will be satisfied. In this type of performance, there exists a risk of unsatisfied customers. Excitement quality refers to satisfying requirements beyond customer expectations. In a competitive environment, it is desirable to meet at least spoken performance levels of customer satisfaction (Quality Techniques, 2006).
Quality Function Deployment
Kobe shipyard of Mitsubishi was the first to define and apply Quality Function Deployment (QFD). A matrix was prepared where customer demands were put on a vertical axis and methods to accomplish them on a horizontal axis. The system has evolved to encompass a wide range of activities within most manufacturing organisations. The technique has been defined as: “A system for translating customer requirements into appropriate company requirements at every stage, from research through product design and development, to manufacture, distribution, installation and marketing, sales and service.” In QFD general requirements are translated into specific primary, secondary and tertiary requirements. Primary requirements are top level definitions that are customer requirements, whereas secondary requirements are functional requirements and tertiary requirements are the smallest levels of detail to achieve those requirements. As a manufacturing organisation, the focus is on tertiary requirements to satisfy all the customers involved. Commonly deployed data gathering techniques include questionnaires and discussion techniques. The objective of the data gathering process is to understand the needs of all customers and accurately translate them into functional requirements. A QFD matrix has been illustrated in figure 2 (see appendix a). Customer requirements are located on the left side of the matrix as ‘whats’. Design requirements are located on the top of the matrix as ‘hows’. The relationship matrix is the combination of ‘whats’ and ‘hows’. Symbols are used to denote relationship strengths, such as strong, medium or weak. ‘What’ importance ratings are located on the left of the matrix alongside the ‘whats’. ‘How’ importance ratings are located at the bottom of the ‘hows’. These are generated from ‘what’ importance ratings and relationship strengths, and summarised for each ‘how.’ Relationships could be either synergistic or trade-offs. This allows for early selection of trade offs or synergies in favour of the customer. Customer rating and competitive assessment serve as a measure of present competitiveness and the proposed design. Also, a determination could be made whether current capabilities are leading or lagging and gaps are present. The checks within the QFD process ensure that the quality deployment process does not go off track. The QFD process could be extended to other components of the product development life cycle (see appendix a, figure 3). The product development life cycle includes product planning, part deployment, process planning, and production planning. Product planning includes translating customer requirements into functional requirements. Part deployment includes identifying component characteristics that satisfy functional requirements. Process planning involved identifying processes for production of the components. Production planning involves transition of the product plan into manufacturing. Thus the QFD approach is multidisciplinary involving representatives from marketing, design, production, engineering and control, and sales. Though preparation of the first matrix is often time consuming, industry experts have reaffirmed that the time spent in up-front effort is repaid many times in downstream activities. QFD results in benefits including lesser development time, fewer changes that are undertaken at an early stage, lesser problems at start-up and thus lower costs, lesser field problems, customer satisfaction and improved company knowledge base (“Quality Function Deployment,” 2006).
Statistical Process Control
Recent developments in manufacturing technology have shifted the emphasis for achieving quality to preventive techniques from traditional inspection techniques. This is achieved by the use of Statistical Process Control (SPC). SPC involves a statistical method of data collection and analysis that allows monitoring of the operation and controlling the process. SPC is based on the work of Walter Shewart, who introduced the use of control charts to detect process variation in 1927. W. Edwards Deming was influential in the development of SPC as a technique. SPC has been popularised by the success of Japanese companies, and has been widely adopted by manufacturing companies worldwide. The basic principle in SPC involves understanding variation in a process and managing it. Over the long term, SPC aims to minimise variation in processes and meet customer requirements more closely than ever before. This is achieved by the evaluation of capability, provision of control systems, and provision of guidance for continuous improvement. Central tendency and spread are the two basic elements in a process. Variations that can affect a process include common causes, special causes, and cyclical. Common causes are inherent in the process and can be changed by fundamental action. Special causes are caused by transient causes outside the process that could be tracked to the cause and action taken to prevent them from recurring. Cyclical variation is caused by non-random repeating pattern. In addition to these elements, an understanding of the shape of the distribution is necessary. Most industrial processes could be classified as normally distributed (see appendix a, figure 4). The proportion of population within limits as multiples of standard deviation from the average has been illustrated. For example, 99.73 percent of the population lie within three standard deviations on either side of the average. Using these factors process capability could be assessed and controlled. Process capability involves the determination whether a process is stable and able to comply with the specifications. An instable process will be constantly subject to transient and special causes. Knowing the process enables us to adjust the process within the allowable tolerance limits. A process control chart has been illustrated in figure 5 (see appendix a). The spread is the distance between the upper and lower control limits. It represents the common cause variation. Thus if no special causes of variation exist, we can predict that the output will lie between the two limits represented by three standard deviations on either side of the average. The use of SPC enables us to predict the presence of assignable causes and take appropriate action, by distinguishing between the effects of assignable causes because of normal circumstances, and unassignable causes because of special circumstances. Since the nature of process control has evolved to being predictive rather than reactive, a system of continuous sampling is desirable. It must be noted that it is not desirable to run SPC on processes that are not capable, since they will result in production of out of tolerance products and require complete inspection of components. A system of ongoing sampling is based on the Central Limit Theorem. According to the Central Limit Theorem, distribution of sample averages are normally distributed even when the parent distribution is not normal. The control limits are three standard deviations on either side of the sample average distribution. Setting the control limits to three standard deviations enables us to overcome alpha and beta risks associated with using two standard deviations or lower. The probability of taking action when an assignable cause is not present and the probability of failing to take action when a special cause is present are overcome by setting the standard deviations. A lower standard deviation causes an increased risk of tampering, which is undesirable. Several types of control charts are available for variable control. These are based on the principle of controlling two elements; sample average or sample median and sample range or standard deviation of the distribution. Commonly used charts include sample average and range. Issues with the use of process control charts include decisions involving statistical abilities and practical considerations. Hence, SPC must be used as a tool to enhance engineering or management decisions. Guidelines on the use of SPC include processes where output could be measured, concentrating on areas of immediate benefit. This could include areas such as customer complaints, high costs of quality, etc. Characteristics that are important to the customer or those facing problems are characteristics to be controlled. A subgroup size of three to five is desirable depending on time/cost considerations. The frequency of samples is dependent on the process. It should be balanced between the rate of process change and effort required for taking samples. Another consideration is the number or value of items produced between samples, which are a measure of the quantity at risk. An important consideration is the selection of subgroups randomly from the population and samples within the subgroups must to be consecutive. Conventionally control limits are calculated after twenty subgroups have been generated. Control limits should be recalculated on significant changes to the processes or changes over time. Indicators of a process out of control include a range point outside the limits, an average point outside the limits, seven points going up or moving down, seven points moving up or below the mean, eight points in the middle third (process improvement), or any non random pattern. In case of negative signals, the process must be stopped or inspection conducted for all items produced until the problem has been solved. In case of positive signal, the limits should be recalculated and reason understood, so that the improvement could be retained or replicated. Successful application of SPC requires an understanding of variations and their effects on processes, and continued efforts to control and improve processes (“Introduction to Statistical Process Control,” 2006).
After achieving statistical control in a process, it is desirable to examine the pattern of variation in the process and determine if it within acceptable limits. Capability evaluation involves determination whether the process is able to meet the set specifications. Four processes has been illustrated in figure 6 (see appendix a). Such information allows us to make appropriate decisions, such as which process to use or control. Process calculations have been illustrated in figure 7 (see appendix a). These include process potential, which compares the process spread to the width of the tolerances, and likelihood of producing non conforming product for the process, which compares the distance from the centre to the tolerance limits to the distance from the centre to the limits of the process. It should be noted that likelihood of producing non conforming product should be set equal to process potential, which also means that the process is on target. Taguchi has defined quality as being on target with minimum variation. Thus, when a process is stable, normally distributed and centred, it is possible to predict out of tolerance, and process capability could be improved by reducing common cause variation (see appendix a, figure 8). It is imperative to set the process close to the target value to minimise losses (Shewhart, 1980).
Measurement of Process Capability
Errors in decision making could be caused by variability in measurement systems. Such mistakes could be avoided by characterising the variability in the measurement process and ensuring that it does not interfere with the data significantly. Four types of variation have been illustrated in figure 9 (see appendix a). The difference between the process measurement and the true value is the accuracy. The variation in a set of measurements on a part is known as repeatability. The difference between distributions in the same process is known as reproducibility. The drift of the process of measurement over time is known as stability. Variation caused by repeatability and reproducibility must be kept within acceptable levels. Accuracy is dealt by the use of correction factors. Stability is addressed during process design and setting calibration and maintenance levels. Sources of variation in a measurement system have been illustrated in figure 10 (see appendix a). Variability is introduced in the measurement system at various stages of the measurement process. The sources of variability within the process are easily addressed once identified. A measurement process has been illustrated in appendix b. Common causes of variability, such as those observed between operators is addresses by using the best operator, and special causes are addressed in a similar fashion (Shewhart, 1980).
The relationship between committed expenses and actual expenses has been illustrated in figure 11 (see appendix a). Actual expenditures for different phases have been illustrated in the figure. Actual expenditure ramps up during the development and production phase and the opportunity to make changes decreases, thereby emphasizing the importance of proper design and planning. Thus, there is an understanding of the need for improving up-front effort that results in improving product quality and reduction in expensive product support functions. QFD has thus established itself as a vital process in various stages of the product life cycle (“Quality by Design”, 2006).
Recommendations have been developed for improving competitiveness, based on an improved understanding of customer behaviour and requirements, and assessment of quality techniques.
Recommendation 1: Form a team for QFD (quality, reliability and maintainability). A team comprising of members from marketing, sales, product design and development, and manufacturing should be formed for implementing QFD tools. It is desirable to impart professional training to key individuals in another organisation or by inviting consultants. After learning QFD skills, these professionals would train other professionals within the organisation.
Recommendation 2: Develop a program for QFD. A program for QFD should be developed to suit the specific activities and requirements of the organisation. This includes the development of policies and procedures for meeting quality objectives. Specific tools, such as QFD matrices have to be developed based on customer requirements, organisational capabilities and targets. Specific tools have to be developed for different stages of the product development cycle. The program should include custodians for the different phases and responsibilities assigned to ensure that the goals are met.
Recommendation 3: Develop a SPC program. A SPC program should be developed for specific activities that have been identified to add value to the product development cycle. The program includes procedures to address quality requirements and methods to achieve them. The process could be automated to some extent by the use of software. Roles and responsibilities to ensure that the program objectives are met should be addressed.
Recommendation 4: Develop a program to measure process capability. This is an extension of the SPC program, and the same roles and responsibilities should be assigned for this program. The procedures and tools to be used for the measurement of process capability should be addressed in the program.
Recommendation 5: A pilot program should be conducted to test the effectiveness of the different components of adopted quality techniques, and best practices adopted. This includes running the programs under different conditions and considerations made for practical situations. A report illustrating the benefits achieved should be prepared and presented to the management. This serves to justify initial expenditure to the management for the acquisition of materials, training and changes in processes. Also, continued support from management is necessary for long term success of such initiatives.
Recommendation 6: Training and education. Often changes in processes or adoption of new techniques meet with resistance from line workers. The best strategy to deal with such situations is to arrange for training and educating them about the benefits of including such changes. Line workers should be included in teams whenever possible.
A review of customer requirements and customer satisfaction has been made. Availability of a wide variety of choices to the customer has made it imperative for organisations to be customer focussed, and deploy all possible techniques and tools to satisfy customer requirements. Also, deploying quality management techniques has proved to be of economic benefit to organisations. Techniques for improving quality include QFD (quality, reliability and maintainability) for various stages of the product development cycle. Specific tools such as SPC could be deployed to ensure that processes are under control. Processes capabilities should be measured to ensure reliability and maintainability. Recommendations include development of a cross functional team for QFD, and initial training for select individuals for developing competency. Trained individuals would train the rest of the team. Specific programs for QFD, SPC and measurement of process capabilities have been recommended. Communication and education of line workers are essential elements of program implementation. Training and education of line workers has been recommended. Management must be kept informed of program initiatives and results accomplished. This will enable successful execution and integration of the overall quality management program, as continued support from the management is vital for long term success.