Statistical Process Control PDF

STATISTICAL PROCESS CONTROL  SAMPLING The development of the food industry in the world and in our country with technological innovations has brought with it a great production potential. Large-capacity enterprises have gradually replaced small units. However, the...

STATISTICAL PROCESS CONTROL  SAMPLING The development of the food industry in the world and in our country with technological innovations has brought with it a great production potential. Large-capacity enterprises have gradually replaced small units. However, the necessity of ensuring the continuity of production compliance with the determined specifications has led food manufacturers to implement an effective quality control system.  SAMPLING The most important phase of an effective quality control system is sampling. Sampling is used in quality control, input, inspection and output control studies of enterprises. A complete count of the main mass is very difficult and sometimes impossible. For this reason, sampling is used because of the destructive nature of a full count and its cost.  SAMPLING The point to be considered in sampling is; taking the sample that will best represent the whole, taking the most appropriate sample for the analysis to be applied.  SAMPLING When the amount of sample to be taken from a whole, and the amount to be used for analysis from this sample, is considered to be very small compared to the whole, the importance of a sample that will represent the whole is better understood. Therefore, it is very important that the sampling process is carried out by people trained in this subject and that the sample taken is stored without any changes until analysis. If the sampling process is faulty, it will cause wrong decisions to be made about the quality of the whole batch and will also bring sampling risks. For this reason, the concept of sampling is of great importance in food quality control and therefore in food analysis.  Purpose of control FACTORS Structure of material to be tested AFFECTIN G Test methods SAMPLING Nature of the lot  FACTORS AFFECTING SAMPLING PURPOSE OF CONTROL Not all controls may be performed for the same purpose. Therefore, a suitable sampling plan should be selected for the intended purpose. In some cases, only the batch may be accepted or rejected for sales purposes. For this purpose, the control of qualities is used quite frequently. However, in cases where there is no chance of rejecting the batch, the aim may be to evaluate the entire quality level on a previously prepared score scale or it is necessary to control the quality of the product during production. In such cases, the control of variables is appropriate.  FACTORS AFFECTING SAMPLING STRUCTURE OF MATERIAL Homogenity of the material: If a sample is taken from a homogeneous material (e.g. water), one sample is sufficient. If a sample is taken from a heterogeneous material, the sample amount should be increased and homogeneity should be ensured before sampling. Sample size: When taking samples from liquid and semi-liquid materials, a sample unit should be determined, such as the container volume or the probe volume to be used. As the unit amount of material increases (e.g. corn cob), one unit can be accepted as the sample size.  FACTORS AFFECTING SAMPLING Source of the material: If it is a reliable source, the sample size can be reduced, but if it is a questionable source, the sample size can be increased. Cost of the material: If the material is expensive, the sample quantity can be reduced.  FACTORS AFFECTING SAMPLING TEST METHODS The importance of the test: The sample size should be large for the control of critical quality characteristics that will directly affect the health of the consumer, while the sample size to be used for the control of other factors of economic importance can be smaller. Whether it is detrimental to the sample: If test methods that damage the sample, such as chemical methods, are to be used, the sample size should be kept smaller, and for methods that do not damage the sample, such as sensory methods, large amounts of sampling are recommended. Time and equipment: In cases where the test takes a long time and the equipment cost is high, the sample size should be reduced.  FACTORS AFFECTING SAMPLING NATURE OF THE LOT The size of the lot, its characteristics such as being packaged or in bulk, affect the size of the sample. If the material is packaged, it should be determined how many units will be taken in the sample. , In addition, if the pile is stacked randomly, the sample should be taken by chance. N: lot size n: number of tested samples c: number of accepted samples P: percentage of defective  ADVANTAGES OF SAMPLING It is economical because only a part of the product is checked Minimum damage is done to the product during the check Less number of inspectors are required, education and training problems are simplified Test methods that measure product quality quantitatively can be applied even if they are destructive to the product. Return of products due to insignificant defects, rejection of the entire batch in retail sales provide strong motivation for development  SAMPLING RISKS Rejecting good batch or accepting bad batch with sampling  -PRODUCER RISK: Probability of rejecting good batch with sampling plan. n is decreased, c is increased to reduce producer risk β -CONSUMER RISK: Probability of accepting bad batch with sampling plan. n is increased, c is decreased to reduce consumer risk  PURPOSE OF SAMPLING Generally, the sampling process is done for two purposes; ACCEPTANCE SAMPLING Acceptance or rejection of the final product. Acceptance sampling is done for the purpose of separating defective units. CONTROL SAMPLING Taking corrective action during the process when necessary. Control sampling is to find and eliminate the source of the error,  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL BRAINSTORMING Brainstorming is a tool used to generalize all possible causes of a selected problem. In this technique, the process is random and no judgment of ideas is made. At the end of the process, ideas are exchanged to evaluate the possible causes of any selected problem.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL PARETO CHARTS Pareto charts are a method of evaluating problems by a committee based on their importance. In this technique, data is organized into “vital few” and “useful many”  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL PROCESS FLOW CHART Process flow diagramming is a technique of creating a diagram showing the flow of a process in order to discuss and evaluate the various stages of the process.  TECHNIQUES USED IN STATISTICAL PROCESS CAUSE AND EFFECT DIAGRAM CONTROL Cause-and-effect diagrams are also commonly referred to as “CEDAC” or “fishbone” diagrams. These diagrams are used to determine the causes of problems in processes.  TECHNIQUES USED IN STATISTICAL PROCESS FREQUENCY DISTRIBUTIONS, CONTROL HISTOGRAMS, AND SPECIFICATIONS Frequency distributions, histograms, and specifications are used to gather and interpret the data collected after problems have been identified. Frequency distributions Frequency distributions provide the information needed to show the variability or distributional characteristics of the data.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL Histogram A histogram is a tool that shows variations in a process and indicates whether the data is within normal, specifications, and process control limits. Calculations of standard deviation, mean, and specification limits help the operator determine if changes are needed during the process.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL Specifications Specifications are tools that provide an estimate of the probability that a product will be accepted or rejected within specified tolerances.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL CONTROL CHARTS Control charts are tools used to examine changes in a repetitive process. These charts are used to determine whether upper and lower control limits are exceeded and whether the process is under control. Specification limits are values ​developed to show that the product or process is under control, and are also referred to as allowable tolerance limits and/or limits.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL CONTROL CHARTS ACTION ZONE UCL WARNING ZONE STABLE ZONE STABLE ZONE WARNING ZONE LCL ACTION ZONE  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL CONTROL CHARTS FOR MEASURABLE VARIABLES - VARIABILITY CHARTS This group, also called variability charts, is classified as control charts x and R charts. In order to benefit from all information in the most effective way in serial measurements, charts showing average (x) and range (R) values ​are preferred. For example, these charts are xused in the evaluation of in-process and final product data. Valuesx​of the chart show variations originating from the central line and caused by factors such as material, method, machine, human or environment. R charts are an indicator of uniformity. Variations in uniformity can be caused by factors such as the distraction of the operator and the variability of the parts.  TECHNIQUES USED IN STATISTICAL PROCESS CONTROL CONTROL CHARTS FOR UNMEASURABLE ATTRIBUTES - ATTRIBUTE CHARTS In the graphs, also called quality charts, observation is required for only one characteristic. These graphs are generally used in the food industry to evaluate data on raw materials or additives and are classified as defect rate (p), number of defective (np), number of defects (c) and defect per unit (u) graphs.  INTERPRETING CONTROL CHARTS A point below or above the control limits means that there is an error in the process due to special reasons and corrective measures must be taken (Figures a and b).  INTERPRETING CONTROL CHARTS Observation of a series of five or six points above or below The two-out-of-three rule the mean indicates that a applies when two out of three trend has set in. In this case, consecutive points are too the control limits and the close to the control limits. This mean should be reconsidered is not natural and the reasons (Figure c). must be investigated (figure d).  INTERPRETING CONTROL CHARTS It is not natural for there to be If the limits are exceeded very upward and downward trends. frequently, the limits should be Limits must be reconsidered (figure narrowed further (Figure f). e).  INTERPRETING CONTROL CHARTS If there is a periodic change, Most of the points should be separate control limits should be randomly distributed around the established for each period (figure mean line. Some points may g). approach the control limits, but they should not fall outside them (figure h). CORRELATION OR SCATTER DIAGRAMS Correlation diagram, also known as scatter diagram, is a tool used to show the relationships between different variables. These graphs are used to analyze possible cause-effect relationships. The control variable (cause) should form the base or x-axis of the graph, and the measured variable (result) should form the vertical or y-axis on the left of the diagram.  COMPARISON OF THE OLD SYSTEM AND THE NEW SYSTEM APPLIED IN QUALITY ASSURANCE ACTIVITIES THE NEW WAY (TOTAL QUALITY MANAGEMENT SYSTEM) THE OLD WAY Accept all material as delivered Establish specifications on all incoming materials and accept only those in conformance Employ “grab” sample system Use statistical sampling Control of product quality by QC personnel All production personnel trained in QC practices and the application and use of SQC Separation of bad products from good products after Poor products are not produced because of prevention processing methods used in the manufacturing process Make the product. Check it. Decide what to do? Measure the process while making the product, control Concentrate on sample plans for future “control” the process, predict the outcome, study the process to improve it Criticizing of employees for poor product quality Recognition of employees for any improvements in product quality  COMPARISON OF THE OLD SYSTEM AND THE NEW SYSTEM APPLIED IN QUALITY ASSURANCE ACTIVITIES THE NEW WAY (TOTAL QUALITY MANAGEMENT SYSTEM) THE OLD WAY Employees have no training; no say in decision All employees working in the system are trained; process; do what they can under the circumstances have voice in the decision process; do their job right because they have direction, tools, knowledge Problems are solved by hit or miss practice Problems are solved by Pareto principles and use of cause and effect charts Company receives complaints Company receive compliments Products are manufactured by“close enough” Products are manufactured according to syndrome specifications and in conformance to label requirements Quality is controlled subjectively Quality is controlled objectively Managements is not seen or available Management is walking around Quality is unpredictable“what you see is what you Quality is planned and predictable “what you see is get” what to expect at all times” Process of Detection Process of Prevention