Specifications: the bane of buyer and seller.

Specifications: The Bane of Buyer and Seller Problems commonly faced in developing specifications include determining if the manufacturing process is capable of meeting the specifications and verifying quality.

AFS Quality Systems Committee (1-J)

As previously indicated, the supplier is obliged to provide a product to the customer that will meet the customer's specification on a continuing basis. The best way to assure ongoing compliance is to measure and report process capability. Process capability is defined as "the portion of a predictable process output that remains within the specification limits."

Process capability can be used as a measure of the ability to meet an internal shop specification or as a measure of the ability to meet an external customer specification. This discussion focuses on the latter use and presents several common pitfalls often encountered.

Establish Process Capability

To establish process capability, the following criteria must be met: * specifications must be established; * a "predictable" process must be achieved; * process output must be measured and compared to the specification.

If the output of the process falls within the specification, it is considered capable. If the output of the process falls outside the specification limits, there are three choices: * the process must be changed; * the output must be inspected and defective parts sorted out; * the specification must be changed.

The specifications should identify all appropriate performance characteristics and their acceptable ranges as well as, and most importantly, the appraisal unit. The appraisal unit defines to the supplier the smallest quantity that must meet the specification requirements. The lack of, or the improper identification of the appraisal unit, is one of the pitfalls in establishing process capability.

For example, a foundry which identifies the characteristics that a MgFeSi alloy must meet, i.e., size, %Mg, %Ce, etc, must also establish the appraisal unit. In this case, the smallest quantity of alloy required to meet the specification could be one of the following: * the daily 200 ton run at the supplier's crush and screen operation; * the series of 50 ton furnace heats making up the 200 ton run; * the 22 ton shipment of 11 4000-lb bulk containers.

As shown in the example, there can be considerable differences in the product depending upon which appraisal unit is agreed upon. However, once the appraisal unit is established, the supplier and customer have a point of reference from which the process capability may be established.

Next, the supplier must bring his process into statistical control by eliminating special causes of variation. Once special cause variation is eliminated, the process is said to be "predictable," and "in statistical control." The process output now can be compared to the specification to establish its capability.

In practice, the original control chart analysis often will show that the process is not predictable. If a process is not predictable (out of control) and the special causes of variation cannot be eliminated economically, the process capability can be calculated (including the out of control points).

However, it must be understood that the resulting capability limits will be inflated and that the prediction of capability is only an estimate. Though lacking statistical predictability, this estimate is better than an opinion to establish process capability.

Another pitfall of which to be aware is the use of "aggregate" samples to establish process capability. An aggregate is the combination of several individual samples into one. This one sample is then measured and the result recorded or plotted on a control chart as though it were an individual. This practice generally results from the inability to obtain a good representative sample of the process output and is usually associated with bulk items such as sand, coke, coals, ferroalloys, etc.

The practice results in a smoothing effect on the variation that masks or hides the extreme values. When aggregate samples are encountered, the true variation of the population cannot be estimated. This condition is not necessarily bad as long as it is understood that in accepting material based on aggregate or average values, it is an approximation of the population and does not necessarily show the true variation present. This condition should be identified on the specification and defined as part of the appraisal unit.

Following the process capability guidelines will provide a structured approach to verifying specification compliance. This is essential in order to manufacture products "fit for use."

Quality Verification

The final section of a specification addresses the inevitable problem of agreement on verification techniques and solving disagreements. Points to consider in verifying quality and resolving testing discrepancies are:

List Essential Elements--List only those elements or characteristics which positively or negatively influence the product processing, properties or performance. Having nonessential composition specifications will raise the cost of quality with no benefit to the producer or customer.

Establish Units--Establish the requirements agreed to in weight, percent, sums or other units such as maximums/minimums. Avoid any nonspecific phrase such as "balance" or "remainder" and define requirements to be controlled and requirements to be reported for informational purposes only.

Processing Requirements--Specify requirements to the processing stage, i.e., heat analysis, raw material check or finished product analysis.

Identify Procedures--Identify sampling procedures, hardware, increments, etc. Provide ratios of sample weight to final bulk weight, sampling frequency, final units, etc.

Sample Preparation--Identify any special precautions in sample preparation to avoid contamination or loss, as in gas determinations. Establish sample size to provide for the possibility of third party referee analysis. Agree to sample retention time.

Analytical Procedures--Identify standard analytical procedures of options such as ASTM or any other agreed-to methodology. Laboratories do differ in instrumentation and methods. Provide concentration range with instrument testing (X-ray, emission, ICP, DCP, wet chemistry).

Referee Laboratory--When composition characteristics exceed agreed-to values, name a third laboratory, where practical, for final referee determination. Establish conditions that the cost be borne by the lab with the largest deviation from the established values as determined by the referee laboratory.

Confidence Levels--Wherever practical, agree to levels of confidence, error in test method, standard deviation, etc.

The bottom line is to anticipate areas of contention before they occur and develop a mutually acceptable program to resolve differences before they happen. Preliminary efforts in Potential Problem Analysis or Failure Mode Effects Analysis (FMEA) will go a long way in developing a close customer-supplier relationship.

Reporting Requirements

Reasonable reporting requirements will assure a customer that he is receiving a quality product on a continuing basis. If specifications have been established following reasonable guidelines such as those described, they will represent the best possible description of the desired product. However, this does not preclude the bottom line that the product must function in the customer's process. This requirement, while not always written, is part of the customer expectation.

In determining reasonable reporting requirements, it first must be ascertained that the supplier has an effective program in place that will provide a consistent product on a continuing basis. This is accomplished by an in-depth audit of the supplier's operation.

Now that we have a viable specification and have determined that the supplier has an effective quality program in place, the only remaining element is the supplier reporting on incoming shipments. To request control charts on every shipment is not reasonable in most cases. On high volume products such as sand, scrap, resin, coke and alloys, the paperwork would be voluminous and, in all probability, would not effectively be analyzed.

It is better to receive control charts covering specification parameters on a monthly or even quarterly basis. After all, the supplier's audit has assured the customer that a quality product is being provided. Periodic follow-up audits will reaffirm the continuance of the supplier's program.

If a supplier is not to the point where he can successfully sustain an in-depth audit, then the customer has no alternative but to request a certified analysis with each shipment, covering a specified lot size of material. Again, high-volume materials may dictate lot sizes that encompass more than one load.

Remember, the bottom line is performance and fitness for use. The customer must look to his suppliers as part of his team. He must integrate his supplier's expertise into a total program of continually improving product quality and productivity. The ultimate end to be achieved is a continual flow of quality products at all levels.