• Home
  • |
  • Blog
  • |
  • Developing Meaningful Product Specifications

Article 14

Read Time 5 Minutes

Here is a hard quiz question for all of you engaged in R&D and New Product Development? 

A Quiz:

  1. How do you develop product specifications?
  2. How do you know they are meaningful?
  3. Where does the needed data come from? When?
  4. What is the underlying science?



In this article I will try to provide conceptual answers to these questions and share the fundamental scientific principles that one needs to learn to understand how meaningful product (or process) specifications can be developed. 

Historically specifications were developed through gut feel, judgment, a not-so-well-understood trial and error process and negotiations between the buyer and the seller, and sometimes by regulatory agencies. Many companies still use the same age old methods.

These methods are inefficient at best and can prove to be quite expensive in the long run, because the manufacturing gets saddled with a bad specification/process/procedure and does not know what ails the system. And since Manufacturing does not know what ails the system, over time they develop a series of short term band aid solutions!

To develop meaningful specifications, one needs a few concepts from Statistics and Quality Management

Normal Distribution

First key concept is the most popular case, a Normal Distribution. It has two parameters:

  1. Mean which represents the Central Tendency. 50% of data fall below the mean, 50% above the mean.
  2. Sigma which is a measure of the noise or variability in the system. 

 

Example - Two Sided Specifications

So for example for a polymer adhesive manufacturing process you are consistently making a polymer with mean tensile strength of 1000 and a standard deviation of 100, you may claim that 99.7% of adhesive batches would have a tensile strength between 700 and 1300. This is an example of a two sided specification where both minimal and maximal performance is desired. However, you may realize that you may not be able to manufacture 99.7% of batches consistently enough and therefore specifications range for the tensile strength for the adhesive batches should include even wider range. In practice you would use prior manufacturing data/experience to initially estimate these specifications, publish them as tentative specifications and then revise them once good data are available. 

Example – One Sided Specification

Sometimes we need single sided specifications. For example product purity needs to be at least 95%; for a nylon rope for mountain climbing, tensile strength should be at least 10,000 psi;  impurity level at most 10 ppm.

Some product attributes, or process parameters may exhibit different a distribution other than the normal distribution, for example log normal, binomial etc. Computations formulas change somewhat.

Motorola and Six Sigma

In many industries selling products to 95%, 97.5% or 99% specification limits was quite common. In early eighties, Motorola engineers were developing sophisticated first generation wireless phones where each phone used hundreds of transistors. They recognized that for these sophisticated phones 99% good transistors is not good enough, so they came up with the ideas of part per million defective and the Six Sigma process.[ref. 2] Airlines for example strive to achieve successful landings at a very high rate, so failure rate is at most a few ppm.

Sometimes specifications are developed/initiated by the supplier(Crosby, meeting customer requirements), sometimes by the customer(Juran, fitness for use), sometimes by regulatory agencies to ensure public safety. Taguchi mentions that for a given specification more consistent the product, higher the customer satisfaction. 

Blending 

When an out of specification batch is produced, most common corrective approach used by Manufacturing is the blending of the batches. While blending of good and bad batches masks the quality problem(Mean value is acceptable) it results in higher variability(sigma) within the batch and long term customer dissatisfaction with the consistency of the product. 

Hand Picking of Lots

A common problem faced by polymer industry is that over time, they develop many similar products differing slightly in molecular weight, viscosity and a few related properties. This product diversification is the result of large volume customers insisting on slightly different specifications for essentially the same product and the supplier agreeing to deliver the same. The net result is that Manufacturing cannot produce and deliver on demand a few batches specific to a particular customer’s need. “Hand Picking” of lots gets started, “save this lot for customer A, save this lot for customer B,” resulting in higher inventory levels and missed shipping dates. Both the customer and the supplier are unhappy. The supplier does not even realize they do not have the capability to produce a quality product on demand. 

Both Blending and Hand Picking of Lots are symptomatic of hidden quality problems. 

Even when manufacturing uses real time process control technology, these problems arise. Use of real time process control is no guarantee of absence of quality problems. To address operational issues, operators/supervisors change the target set points, which result in change in product quality.

As I said in my last article, Part 13 [ref. 3] in Quality arena, the coaches, the Six Sigma Black Belts, the Statisticians and the Quality Managers practice Deming’s motto. 

In many organizations, “quality” is the responsibility of the Quality Manager.  It is not. He only generates reports on quality. 

Quality is everyone’s responsibility.  To improve quality we need to involve everyone.

So crank up your Product Development engines... Let us speedup new product development and growth rates. And let the fun begin!

References:

  1. Introduction to Statistics Using LibreOffice.org/OpenOffice.org
  2. Six Sigma – Wikipedia, the free encyclopedia
  3. Mukul Mehta, Developing Testing Awareness for Quality P 32-33, Chemical News, March 2016 
{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}

Want to learn more?

Check out these articles below

>