Process Capability and Statistical process control

1.Statistical Process Control

Organizations use various methods to assess the quality of their products. One of the methods they use is Acceptance Sampling. Acceptance Sampling is the process of inspecting all finished goods before distribution to the customer. It is applied at the output stage and hence if a defect arises during the production of a batch then the entire batch has to be rejected as scrap. Thus there is a need for assessing quality at all stages of production. Thus statistical process control is used. SPC is used to assess the quality of samples of items during the transformation stage at regular intervals of time. If the manager finds that there is any variation from quality, then he must find out the cause for such variation. If the cause of the variation can be traced back to factors such as faulty equipment, the workers fault etc. then it is known as an assignable cause. A process is said to be out of control if there are assignable causes are present in it. The major sources of assignable causes usually are- machine, materials, method and environment. On the other hand, the variations that are caused due to the inherent nature of the process are known as random variations, which are present in every process.

1.1 Control Limits

In a production process, there are no 2 identical products. There is always some variation/ differences between each item. To maintain a standard acceptable level of production, the manufacturer usually specifies certain limits within which a product is considered acceptable. For example, the length of a pen may be specified as (9 + 0.1) and (9 - 0.1) cm. These are known as specification/ tolerance limits. Here 9 cm is the mean and 9.1cm is the upper specification limit (USL) and 8.9cm is the lower specification limit (LSL). But it is also important to take precautionary measures so that these limits are never reached, thus the manufacturer sets control limits within these specification limits. The lower control limit (LCL) is usually set at M-3sigma and the upper control limit (UCL) is set at M+3sigma. When a product exceeds these limits, then the manufacturer starts to look for assignable causes.

1.2 Control Charts

The control limits are a part of the control chart. Control charts are of 2 types- control charts for attributes and control charts for variables. The earlier example where we specified the limits of the pen are an example of control limits for variables. The control charts for variables are X chart and r chart. The control charts for attributes are based on whether the item is defective or not, or in terms of nonconformance with certain criteria (p-chart) or it can be constructed for the number of defects (c-chart).

1.2a Control Charts for variables

1. X-bar Chart This chart is used when the sample means of a process are plotted on the chart. The sample size is 4-5. X-bar chart is based on normal distribution. 2. R-chart We plot the sample ranges to control the variability of the process. These ranges reflect the process variability instead of the tendency toward a mean value. Both R-chart and X-bar chart can be used together, because it is important to keep the process average and the process variability in control.

1.2b Control charts for attributes

1. p-chart A p-chart uses the proportion of defective items in a sample. A p-chart can be used when we want to check for the conformance or non-conformance of a product, not taking into account the number of defects of the product. The sample proportion defective, an estimate of the process average is used to compute the control limits and the sample standard deviation. The sample standard deviation is given by- SQRT{*p*(1-p)+ / n} and the control limits are given byp + 3*SQRT{*p*(1-p)+ / n} where p = average of all sample proportions or (sum of all defectives) / (total sample size)

2. C-chart It is used when it is not possible to compute the number of defectives. It is used to assess the number of non-conformities/ defects per unit. It differs from the p-chart in the sense that it allows for the condition of more than one non-conformity per unit. e.g.- The number of scratches in a duster, where the number of scratches can be counted but the total number of possible scratches is not known so proportion of defectives cannot be used. Here the mean number of defects per item is given by (c) Control Chart Limits for c-Charts:

{(c) + 3*SQRT(c)}

1.3 Comparing control charts for variables and control charts for attributes
1. Variable charts give a quick indication, while attribute charts are slow to respond 2. Attribute charts are preferred to variable charts especially for the inspection of finished goods. The attribute is usually if the product is conforming to the specifications or not. 3. Variable charts are preferred over attributes for finding the process capability ratio. 4. Variable charts take much more inspection time per unit as compared to attribute charts. 5. Variable control charts require a much smaller sample size.

1.4 Benefits of Control Charts

1. They help to diagnose the assignable causes in a process which can then be eliminated using managerial interventions. 2. They help to prevent unnecessary process adjustments. 3. They help to decrease the overall variability once the assignable causes have been eliminated then other aspects of the process can be focused on.

2.Process Capability and Process capability measures

Tolerances are not determined from production process; they are imposed by designers of the product and service. Control limits are based on production process, and they reflect process variability. It is possible for a process to be statistically in control according to control charts, yet the process may not meet the design specification. Therefore, the process must be evaluated to meet the product specification before it is initiated. Process capability refers to the natural variation of the process relative to the variation allowed by the design specifications. The three main elements associated with process capability are process variability, the process center and the design specification. One measure of the capability of a process to meet design specifications in the process capability ratio(Cp).It is defined as the ratio of the range of the design specifications to the range of the process variation. If its value is less than 1 the process range is greater than the tolerance range. Cp = Tolerance Range/Process Range = (Upper specification limit Lower specification limit)/ 6 A second measure of process capability is the process capability index (Cpk). It tells whether the process mean has shifted, and if shifted then in which direction. The process capability index measures the capability of the process relative to the upper and lower specifications. Cpk = Minimum [(Process Average Lower Specification)/3, (Upper specification Process Average)/ 3 ]

Process capability

3.Role of SPC in improving the Bottom line (Profitability)

SPC Charts are used for maintaining the quality of the products or the services. There is a cost associated with quality and to improve the operational efficiency, costs have to be brought down. Quality costs fall into two categories:1. Cost of quality assurance The cost of quality management includes prevention costs and appraisal costs. Prevention costs are the costs incurred during product design. This includes costs of planning, design, process, training and information costs. On the other hand appraisal costs include cost of measuring, testing and analyzing. 2. Cost of not conforming - This is also known as cost of poor quality. The cost of poor quality (COPQ) is the difference between what it actually costs to produce a product or deliver a service and what it would cost if there were no defects. Statistical process control helps in reducing the cost of poor quality. Cost of poor quality is divided into two categories 1. Internal failure costs - Internal failure costs are incurred when poor quality products are discovered before they are delivered to the customers. Internal failure costs comprises of scrap

cost, rework cost, process failure cost, process downtime cost, price downgrading cost. Manufacturing has a high internal failure costs as compared to external failure costs. 2. External failure costs External failure costs are incurred after the customer has received a poor quality product and are primarily related to services. These costs include complaints, returns, warranty claims, liability and lost sales. Internal failure costs tend to be low for a service, whereas external failure costs can be quite high. Improving the Process capability improves the efficiency of the process and helps in reducing the costs of quality assurance by reducing the frequency of measuring, testing and analyzing.

4.Applications of SPC
SPC as a quality management tool is used in both manufacturing and service industry. The difference is in the nature of defect being measured and monitored. Control charts for service industry tend to use quality characteristics and measurements such as time, customer satisfaction and order accuracy. Manufacturing industries have a manufacturing design and process tolerances. These attributes and variables have to be strictly followed to maintain the uniformity among the products. SPC charts are nowadays used in measuring the deviations in supply chain departments, shopfloor production process and also in assembly lines to improve the repeatability and maintain the uniformity of the process. Process control charts are the basis of Six sigma manufacturing and helps in determining the next step to be taken for improving the process capability. Examples of various service industries which use SPC are:1. Hospitals: Timeliness and quickness of care, staff response to request, accuracy of test lab, cleanliness, courtesy, speed of admittance, accuracy of paperwork and speed of checkouts. 2. Grocery Stores: Waiting time to check out, quality of food items, frequency of out of stock items, cleanliness, checkout register errors, customer complaints. 3. Airlines: Flight delays, waiting time at ticket counters and check in, lost luggage and luggage handling, agent and flight attendant courtesy, passenger cabin cleanliness, accurate flight information and cabin maintenance. 4. Catalogue-order complaint: Order accuracy, packaging, operator knowledge and courtesy, delivery time, phone order waiting time. 5. Fast-food restaurants: Waiting time for service, cleanliness, food quality, order accuracy, employee courtesy and customer complaints. 6. Insurance companies: Billing accuracy, agent availability, timeliness of claim processing and response time.

5.Limitations of SPC in Short Production Run

Short production runs usually have smaller lot sizes, shorter lead times and less available process data. It is characteristic of job shop processes where the variety of products manufactured is high and the number of units of each product manufactured is small. It becomes difficult to use control charts here, because of the sheer number of control charts that will be needed because there are small lot sizes of several types of products. Other methods can be used for short production runs. One such method is DNOM, Deviation from Nominal Control Chart. In this method, the manufacturer assigns the target value for each product group. For example, if there are 3 products being made on one machine each of them will have their own target values, i.e. product A, B, C can have target values 10mm, 20mm and 30mm respectively. Then he calculates the deviations of the actual variable from the target value for each item. These deviations are then used to calculate the X-bar and R values which are then used in the the X-bar and R charts to assess the variability. One important assumption of using this method is that the sample size is constant for all groups/parts and the standard deviation is the same for all parts.

6.Process capability and Six Sigma

Six sigma methodology employed CP as the accepted standard quality measure. Six sigma was developed for reducing the complexities of the product and for observing the failure of the products in order to predict the performance of manufacturing and services. Same as six sigma methodology, in a process capability, we use sigma units for denoting the number of standard deviations between the process mean and the nearest specification limits. Sigma quality level is used to express the capability of the process that means how well the process is performing with respect to its control limits. Nowadays most of the manufacturing companies are required to produce a product from a process having a desired Cpk value. As the market is becoming more and more competitive it is getting tougher for the organizations to reduce cost and the price of their products. They are under a tremendous pressure to improve and sustain the specified Cpk value or quality level. Higher value of process capability index (Cpk) indicates better process. Process capability that is Cpk = 1.0 is equal to a three sigma capability. At this level the process is capable of producing 99.73% of good products. Which is not acceptable in todays world of high competition. Many industries have started working at four sigma levels that is at Cpk = 1.33. At this level the process has the capability of producing 99.9937% good products. At six sigma capability level, a process will produce very few or negligible defects. This level has a Cpk value of 2.0 which is known as six sigma capability.

7.Illustration of Cost benefits by using SPC

Fig. 1 Given Data: 1. 2. 3. 4. Each workstation (A, B, C and the quality station) has a cycle time of 1 Min. Input cost of the material is Rs 1000. Total cost of manufacturing the product is Rs 1000. Each station is adding some cost to the manufacturing process.

Consider there are two companies, Company A and Company B. Company A has not implemented SPC whereas Company B uses this tool for its process. Company A: There is quality defect detected at Quality station. Operator stops the line for inspection to find out which station has deviated from its ideal or the standard process. Now in doing this he has lost production equal to the cycle time of each station. Assumption here is that he is able to find out the root cause in that time period. In doing so he has lost on production volumes, has incurred scrap or rework lost for the material in the process which is already defective and also will now incur maintenance cost and further shutdown cost. Company B: This Company has been following SPC tool and is able to detect the process deviation before there is a possibility of a line stoppage. This helps in detecting the process deviation at each workstation. By looking at the process chart operators can also make out the potential root cause which helps in reducing the time further. It also helps in eliminating scrap which is the sunk cost and reduce the rework which has a rework cost. It also helps in avoiding the order loss which may finally have an adverse effect on sales and customer satisfaction.

S. No 1 2 3 4 5 6

Costs Involved Cost of scrap at quality station Cost of material which is defective and in process Cost of Production lost to check the work stations Time consumed in finding the fault Maintenance or resetting cost Improvement of the process

Company A Rs 2000 Rs 6000 Rs 3000 High High High Investment (Separate Manpower Required) Nil Nil Nil

Company B

Less Low Operator can do it

8.Boeing: A Case Study

In 1994, the Boeing Companys Military Aircraft and Missile Systems Group was unhappy with its Acceptance Sampling processes. It adopted a new Periodic Validation Assessment (PVA) in a partnership with its suppliers where it also conducted periodic monitoring of vendors processes. PVA helped Boeing in process improvements. However in the last decade, reductions in the defence expenditure has resulted in smaller lot sizes. The traditional SPC charts cannot be used for smaller lot sizes because the lot size units are usually less than the required sample size units. Aircrafts include components which are manufactured in small lot sizes, with a diverse quality characteristics. Boeing could not use short run charts too, because they found that it was difficult to ensure that each quality characteristic had the same standard deviation. They used another methodology called percent tolerance precontrol charting (PTPCC) for all its suppliers processes to control the quality of the small parts. For the PTPCC charts the specification limits are used when the process 6 spread is 70% of the difference between the USL and LSL, i.e. (6/ USL-LSL)= 0.7
The process capability is given byCp = (USL-LSL)/ 6 = 1/0.7 = 1.429 i.e. (USL-LSL) = 1.429 * 6 The specification limits are at: +4.285 and -4.285 . The region between the specification limits is divided into 2 equal parts- yellow region and green region. The region beyond the specification limits is the red region. The control of the process works along the following lines. Five pieces from the process are chosen and the variable is measured, if it falls in the green region, we implement the pre-control chart. We measure the variable for 2 successive units from the production. If both fall in the green region or if one falls in the green region and the other in the yellow region then no action is required. If both are in the yellow region or if one or both parts fall in the red region then adjustment is needed.