The Materials Analyst, Part 35:Choosing performance over specificationsThe Materials Analyst, Part 35:Choosing performance over specifications

This series of articles is designed to help molders understandhow a few analytical tools can help diagnose a part failure problem.Michael Sepe is our analyst and author. He is the technical directorat Dickten & Masch Mfg., a molder of thermoset and thermoplasticmaterials in Nashotah, WI. Mike has provided analytical servicesto material suppliers, molders, and end users for 15-plus years.

We live in anindustrial world that is infatuated with standards and documentation.This is epitomized by the headlong rush into ISO 9000 certificationsand the industry-specific spin-offs like QS 9000. Literally hundredsof billions of dollars have been and will be spent creating, maintaining,and supporting the burgeoning paperwork machine that results fromthese efforts.

The implicit assumption in all this activity is that it makesus better at what we do. Few would argue with the premise thata periodic examination of methods and procedures is a good thing.Ideally, it causes us to question the things that we think weknow and leads us to streamline our processes.

But there is another side to this trend. The bureaucracy associatedwith the creation of standards and practices tends to take ona life of its own. Left to itself, it can become the objectiverather than the tool. Already we have the new version of ISO 9000for the year 2000. And in case you thought you were done whenthat has been satisfied, here comes ISO 14000. Perhaps the bestexample of how this becomes an institution was revealed in anarticle in the Wall Street Journal from mid-December of last year.The piece discussed a sudden trip an automotive engineer had tomake to area clothing stores in search of pink women's underwear.

Why? Because, according to the article, the specificationsfor crash testing state that all of the crash test dummies mustbe wearing pink women's underwear. (This sounds incredible, butthis is the Wall Street Journal we're talking about.) The writerof the article approached the government office responsible forwriting and overseeing such specifications. Rather than admitthe stupidity of the stipulation, the official interviewed borroweda page from the language of statistics and replied, "Anythingthat we can do to minimize variation."

Uniform Testing
This is the tyranny of standardization disguised as science. Oncesomething is written down it becomes enshrined, and it can takethe proverbial act of Congress to undo it-which brings us to thismonth's topic. Part of the documentation process revolves aroundmaterial properties. Over a period of time, the large automotivecompanies have come to realize that it makes more sense to developa list of performance-based properties for an application andthen allow any material that meets or exceeds this profile tobe used in the application.

Other industries have picked up on this approach. This givesthe processor more latitude in what it purchases and the increasednumber of options leads to competition for the business. And thatmeans cost reduction. The critical minimum properties are thencodified into a specification that becomes the benchmark for anymaterial considered for the application.

The selection process is not quite as wide open as this systemmay make it seem. Specifications are written around a particularmaterial family. Polypropylenes can be considered for replacementby other polypropylenes, polycarbonates compete with other polycarbonates,and so on. The system makes more sense if we look at how the specificationsare developed. Existing applications were the first ones shepherdedinto this new approach to material selection. In these cases theend user took a physical property data sheet for the materialalready in use, multiplied the important properties by 80 percent(.8), and used the results as minimum values for any given lotof the current material. Any material being considered as an alternatehad to meet this benchmark.

As time went on, many of these standards were raised by taking90 percent of the standard properties on the data sheet. Thisis known as a quality enhancement program. The point here is thatthe specifications were developed from supplier literature withoutany actual testing to determine the standard deviation for a givenproperty on a lot-to-lot basis.

It is not uncommon in our industry for certain grades of rawmaterial to become obsolete. This can cause significant problemsfor processors where detailed specifications have been writtensince the new materials must undergo a review to make certainthat they meet the spec. While the old material may have beengiven a pass based on a summary review of the data sheet, manyend users want new candidates to go through the actual testingprocess. And this is where it gets interesting.

As hard as everyone is working to get to a system where everyoneprepares test specimens in the same way, tests the specimens inthe same way, and reports the results in the same units, somelarge companies have a long history of doing things their way.They may talk a good game about international standards, but ifthey have a cherished method for evaluating a certain product,woe to the person who suggests that it might not be the best wayto go about it. The particular case we will look at here involvesthe qualification of new polyethylene materials.

Material Consolidation
Polyethylene suppliers have undergone significant consolidationover the last few years. A company that once was four companiessuddenly finds itself with an almost unmanageable list of products,and in order to gain efficiencies it begins to look for duplicationof efforts. If three of the firms all made a material with a meltindex of 8 and a density of .953 g/cu cm that was targeted tothe durable containers industry, it will pick the one being madeon the most modern equipment and the other two are phased out.

All of the customers using the older, obsolete grades get anice letter from the corporate vp of marketing assuring them thatthe move to streamline the product offerings is being made withthe best interests of the customer in mind and the newest technologyis being used to make the replacement material, which, by theway, will run just like the old material. Where it gets troublesomeis when the rationalization eliminates materials that are approximatelythe same. In polymers there is rarely such a thing as identical.End users have become aware of this and insist that the new gradebe tested to their specification to be certain that performancein some key area is not being compromised.

The problem arises from the initial blind faith in the datasheet that accompanied the birth of the specification. The datasheet may have come from an era when nonstandard test methodswere being applied to generate certain properties. There are alot of games that can be played to boost a property value andmost of them are perfectly legal within the framework of an ASTMor even an ISO standard.

In polyethylene materials the test condition that causes thegreatest controversy is that of strain rate in the tensile andflexural tests. The properties of all polymers are dependent tosome extent upon the rate of loading. But materials like polyethyleneand polypropylene are particularly sensitive to this effect. Asthe strain rate increases, the strength and stiffness of the materialincrease and the elongation, which is a relative measure of toughness,decreases.
Figure 1 (p. 48) shows tensile stress-strain curves for a high-densitypolyethylene tested at three strain rates, .2, 2, and 20 in/min.Notice that the stress peaks at a higher point, rises faster asa function of a given strain, but breaks at a lower extensionwhen the strain rate is high. In other words, the faster we applya load to a material the stronger and more brittle it will appearto be. This occurs without any change in temperature. We couldalso produce this effect by keeping the strain rate the same andreducing the temperature of the material, but that's a story foranother day. Table 1 gives the actual numbers for peak tensilestrength and modulus at these strain rates.
Now here is the problem. In soft materials like PE the permissiblestrain rates can be interpreted to allow either the 2 in/min orthe 20 in/min rate. For the specification we were being askedto review, the data sheet for the first material used to developthe specification used the rate of 20 in/min. Consequently thestrength and modulus of the material were optimized. Based onthe performance of the original material, which was never actuallytested, the minimum tensile strength was set at 21 mPa (3045 psi)and the minimum modulus was set at 1040 mPa (151,000 psi).
Along comes a rationalization of the product line, and suddenlythere is a great deal of activity focused on coming up with anew material. When the materials are submitted for test, a three-inch-thickdocument accompanies the bag of raw material. Sometimes it's hardto tell which package is heavier. This document is the often-reveredbut seldom-read material specification. In this case we dutifullyreviewed the specifications and performed a variety of tests designedto produce an approved alternate. But all three alternate materialsfailed the tensile tests.

Blind Specification
A close review of the specification revealed the cause of thefailures. For some unknown reason the end user had set up itstensile test to operate at 10 mm/min (.394 in/min). You can referback to the table and interpolate where this would put the tensilestrength and modulus of the material. All of the materials submittedfor substitution produced tensile strength and modulus valuesthat fell short of the minimum by approximately 5 percent. Werethese defective materials? Of course not. If the strain rate werechanged to the ASTM recognized value of 50.8 mm/min (2 in/min),the materials all passed, although just barely. When tested atthe highest strain rate, the one used to generate the propertiesfor the data sheet, all of the materials passed easily.

Now here is the good part, and this is where the specificationceases to be useful. With inventories running low and a quickdecision needed, the molder submitted the resin that had beensuccessfully used for years to the test protocol outlined by theend user. It also failed by the same amount as all of the prospectivesubstitutes.

You would think this would get someone's attention in the specificationdepartment, but you would be wrong. The search continued for thecorrect material, again using property charts that employ strainrates that are faster than those specified by the end user. Theirony is that by the time a material is found that gives "good"numbers at the slower strain rate, it will represent a fundamentallydifferent material. It will be significantly stronger and stifferand more brittle than the material that has been in use all theseyears. It may fail for these reasons. At the very least it willprobably shrink and warp more. But it will meet the spec.

This is a cautionary tale for the writers and users of testspecifications. As we move away from the regional protocols ofDIN and ASTM and into ISO universality, some of the key test methodsare being rewritten. As an example, ISO 527, which is the counterpartto ASTM D 638 for tensile strength, has mandated one strain ratefor tensile strength determinations but a second, much slowerstrain rate for modulus measurements. Many materials will nowproduce a much lower modulus value than they once did becauseof this reduced strain rate. But it is important to remember thatit is not the material that has changed. There will likely bepanic in the industry as the new test methods cause cherishedproperty values to fall. Hopefully reason will prevail in placeof a single-minded focus on pink underwear.