I would like to invite your comments on the point that the specification limits, LSL and USL are really not the best way to specify what the customer really wants.

For example, if the upper specification limit is 10. A product or service that has a value of 9.999 will be termed good (in-spec) while a value of 10.001 will be termed as bad (defect). In reality, both these products will probably operate very similarly or will probably be indistinguishable by the customer.

According to Taguchi, it is better to shoot for the optimal value and consider any deviation from the optimal value as a loss to society. Do you think specification limits drive the wrong behaviour of considering all parts and services that are within LSL & USL okay - rather than shooting for the optimal value? What do you think?

SJ

Hi SJ sir,

        You've indeed brought up a very thought provoking question on choice between Specification limits & Optimal value while manufacturing & delivering a product/service per customer expectation.

      The pertinent point to note here is how does one define the 'Optimal value' of the product/service that is to be delivered, as that again arrives at a fixed value, unlike  specification limits which have a range. Keeping the customers specification / expectation in mind each manufacturer/producer/supllier of a particular product / service would have a different definition of the 'Optimal value' per his/her perspective of customer satisfaction, unlike specification limits. This again may lead to conflict in business interest(s) unless a standardised 'Optimal value' can be arrived at through a regulatory / governing body, which again becomes difficult as it may not find wide acceptance in business circles.

    Having/setting specification limits does facilitate more operational convenience for the business as against optimal value, but it does compromise on the product/service quality (causing rejection) even it not noticeable by the customer. So, it may be said (if not wrong) that choosing between the two (Specification limit & Optimal value) is essentially a conflict between business interest & customer interest as the former gives more operational convenience as mentioned above & the latter requires adhering to guidelines ensuring better quality, but without any scope for deviation/flexibility increasing the chances of rejection (till proficiency is achieved).

   Ofcourse, without doubt a business/process which operates per optimal value always has better control over itself as compared to its operation per specification limits. However, till a solution can be arrived at, it may be prudent to arrive at optimal value limits (like specification limits) if possible based on customer expectations, for functioning to ensure better efficiency as it will certainly reduce the range/limits of business/process operation compared to the usual specification limits, leading to better business/process control & consequently less variation & defects.  

Other thoughts from one & all.....

Thanks & rgds.

Dear SJ Sir,

Though I agree with your point, I would like to emphasise that most of the organisations still work on specification limits. Only highly matured organisations are looking for targetting at center value or optimum value. In reality, when a product or service performance is approaching specification limits ( Either lower / upper or away from spec for one sided spec ), it indicates that is is almost nearing the deviation.

In my opinion working with control limits (If possible) will be the better approach than working with specification limits. When a product or service performance deviates from control limits, we can initiate necessary corrective action to eliminate the assignable cause and bring back the performance to normal . In such cases the performance will never become defect as it will never cross specification limits.

Please correct me if I am wrong.

Regards

KVS Suresh

Dear SJ

Hi

Generally the spec and limits should be same digits.

eg If the spec is 10 then limits should be 9.5 to 10.4

eg If the spec is 10.0 then limits should be 9.95 to 10.04

eg If the spec is 10.00 then limits should be 9.995 to 10.004

If the limit is 10.0000 Then the desired limits be 9.99995 to 10.00004

I hope the concept of rounding of rules also to be considered

Hi SJ,

I am new to six sigma, just posted my thouhgts on the this question, please provide your comments on the same:-

A product or service that has a value of 9.999 will be termed good (in-spec) while a value of 10.001 will be termed as bad (defect). In reality, both these products will probably operate very similarly or will probably be indistinguishable by the customer.

It is better to have our control limits fixed so that we may take an action when we are going beyond the control limit rather than taking a call on the spec limit as it may be a very thin line.

Regards,

LR

Dear LR,

It is okay to take action when points go beyond the control limits (especially when there is a special cause which causes the points to fall out of the control limits).

In order to compute process capability, we compare the process performance to specification limits. The control limits are an indication of process performance. When we talk about specification limits, we implicitly use LSL or USL or both. However, when we calculate DPMO or Sigma level using this approach, we assume any parts that fall outside LSL/USL as defects - which causes some problems as you have correctly identified. 

It may be better to use Taguchi loss function to characterize quality / cost rather than use specification limits.

SJ

Dear SJ,

Could you explain "Taguchi loss function to characterize quality / cost"  . briefly and where I can find more information on the above

Regards,

K.Maharajan

The values  which are very close or nearing to the USL OR LSL  does not give a proper indication of the desired result as it will only points to the nonconformance. The Optimum Limit  can give a better conformance

The main idea is that losses do not start suddenly at specification limit, They gradually increase as you move away from the target. This can be visualised for a broadband connection (or for a manufactured car with compnents which do not meet the target - shall have a break down on the road probably in few months of driving if not immediately and may damage some other components or balance as well). His concept was we should try to be as close to target as feasible.  

On the lighter side, losses do happen just beyond specification limit when you have penalty's defined by client based on the spec limits. 

However, if you see the bigger picture with the end user perspective (for both sided spec limits) the more you match your promise and his expectation, the better it is for everyone in the longer run.

Here is Taguchi Loss Function explained with a visual. In this visual you can note that as the variation reduces, loss reduces. http://elsmar.com/Taguchi.html 

Many a times, the specification limits are not rightly set. This is because of lesser customer awareness on the subject. Therefore, the specification limits are always met despite a wide gap between what is achievable and what are the current specification limits

One of the example could be newly transitioned processes from an onshore location where the customer does not have enough empirical data to set the LSL & USL.

A point in context here is the specification limits are ever changing. In the cited example, once customer knows that, we are meeting the expectations he redraws the targets and at times, these are beyond comprehension.

Regupathi Krishnan.