sanjai wakhaloo

Thanks Madhavi - for adding a useful resouce-the Handbook by BIS, just to add further for those who are making products for use in the defense forces the applicable standards which are equivalent to IS 2500 are DEF131A ( british std) and MIL-STD-105E ( earlier 105D US std). The defense sector use Normal inspection Level-II , either single or double sampling plans. Special plans with reduced sample sizes are used for destructive tests. As we know most of the quality system requirements like the QMS have their origin in defense stds and similar high end space/avionics/military application of products & systems. The stds and sampling procedures were already in place in such applications many many years ago and have later came to the civil applications. An online calculator of the Accept no & Rej no is available from a link in Wikipedia search for the MIL Std 105E on the page SQC online, if we know the batch size and AQL % for the defect using a certain level of inspection i.e Normal level -II or ant other the Acc & Rej nos are calculated and displayed.This may be useful to those who do not have the tables readily available. regrads : sanjai

Most of the time tolerance level ( acceptable level of defectives) for a type of defect is pre-decided in a product . This helps to set the AQL (acceptable quality level) for this defect in the batch of the product. The AQL level depends on the impact the defect shall have upon the performace , life, reliabilty, serviceablility, safety, etc. Generally critical defects rendering a product unsafe for use are not permitted. Those having very little or no effect on performance are treated as minor such as Color /shade of paint. This can be also derived form the QFD analysis or the FMEA analysis. Of course the customer is always right and even cosmetics/asthetic looks can become important. Another pertinent assumption is that the batches to be inspected have some percentage of defectives, those with more defective are to be identified and segragated by the sampling process. sanjai

Sampling plans when adopted for inspection help to reduce the cost of inspection. But the sampling plan must be chosen with consideration to the operating characteristics of the sampling plan. These are plotted based upon the statistical calculations. It is generally considered a good plan if for a given criteria of acceptance , say batches with 0.25 % defectives or less are acceptable, then a plan would be expected to be a good one if out of 100 such batches at least 95 are found acceptable by adopting this plan. You may refer to the indian std IS- 2500 Parts 1,2 published by the Bureau of Indian stds, new delhi, which deals in some detail about the use of sampling plans for attribute inspection.It also guides you to the selection of a plan. The problem with the plans is that a few batches with good quality may get rejected and similarly a few no of batches with more defective can get accepted. This being called the risk to the producer ( rej good batch) and for the consumer ( acc bad batch). This risk has also been calculated for the plans based upon the assumption of a random sample and probability of occurrence in the drawn sample. sanjai

Can six sigma help in improving food safety and manufacturing. any practical case studies available?

Dear Ashish, We could possibly look at the problem in another way, though a bit in detail: 1) It seems the if the target of prodn is 100 pcs per hr on average then in order that there is no ( zero deviation ) from this target it is essential that the process is consistently able to give 100 jobs on the hour every hour. Since this is not happenning. the degree of consistent behaviour of the packing process is measurable by a sigma level.or the std dev wrt to the target of 100. In this case the sigma level shall be indicated by the deviaton from the target divided by the std dev of the prodn figures obtained over a period of time. Here if it is assumed the std dev is 3 with the spread of 103-100= 3 on one side and 50-100= 50 on the other . which when divided by the std dev 3 i.e 3/3= 1 and one side and 38/3=12 i.e the total spread is about 13.0 std dev, however only 1 std dev is worth considering as above the required prodn of more than 100. All other events would be called a deviation from reqmts of 100 since they produce less than the desired 100. This corresponds to a low acceptance level of ( 1 std dev) 0.5 sigma. 2) Alternatively we could consider that the reqd prodn is not to be less than 100 job per hr , in which case it is a one sided reqmt. The spread is acceptable from 100 and beyond.Assume that the avg/mean is say 120 ( as close to 100 is desirable) . In which case the event of less than 100 /hr would be called a non conformance to reqmts. The consistency of the process is measured by the sigma level i.e the no of std dev that lie between the lower spec i.e 100 and the mean of the prodn rate say 120.considering that the std dev is 5.0 then the spread on this side corresponds to a 120-100/5= 4.0 std deviations. However all event with prodn beyond 100 are acceptable, then the spread beyond 120 is accceptable. Consider that if we were to get a distribution with a avg of 105 and spread of 30 ( varies between 90 to 120) . in which case the events when the prodn has fallen below 100 are called non acceptable events. The probability of such an event is an indicator of the sigma value. Simply in other words (90-100/5)=2 std dev are in non acceptable zone below 100. The Z score for the non conformance is considered as 2.0 which corresponds to acceptance sigma level of 5.0 ( a spread of 10 stdev against a desirable 12 std dev within acceptance zone) . Also the extent or % probability of such events can be predicted once the sigma level/ no of std dev on the side of acceptance are known. This can be done with the help of the normal probability tables in a book on statistics. Sanjai.W

The statistic that describes the capability of a process or a machine to produce jobs within a reasonable portion of the tolerance is called the Process Capability; this is denoted by Cp or PC ratio. The statistic is a measure of the variation in terms of ( ration or % of ) the width of the tolerance ( band) as available from the specification. However a process may have a poor performance due to several reasons. -input component- quality of conformance to requirements - process parameters- operating conditions,speed, feed,etc - machine condition - vibrations, alignments, geometric accuracies,etc - method of inspection/acceptance -external factors eg environmental, ambient conditions, - the person operating the machine ( this is generally very less with modern machines as they reduce the dependance on operator) sanjai

This is a nice way to synergise the kowledge and experience that individuals have gained in their respective work area and projects. One could also have a annual presentation seminar where experts/MBB can present their work for open forum members to witness. The best ones could be acknowledged and felicitated, this could spur the awareness of the benefits and applications. best of luck!! sanjai wakhaloo