Tuesday, April 26, 2011

Testing Discrepencies

As any quality technician will tell you, there are discrepancies between test results on a daily basis.  It is nearly impossible for two test results to have identical results, even if it is a split of the exact same sample.  Just like anything else in life, there are hundreds of variables that play into testing a sample.  Going back to my high school science classes I know that the three main types of error are human, mechanical, and environmental, all of which can significantly change test results in an aggregate and asphalt testing laboratory. 

On the human side of things your technician could read the results wrong, calculate an equation wrong, miss a step in the test method... any number of problems can results from human error and it is usually the largest source of error in the laboratory.  These types of errors are often very difficult to find after the test has been completed and as such, it is extremely important to keep a safe environment for your technicians in which they can admit when they have made a mistake and not be punished for it.  Without this type of open communication your technicians will not tell you of errors they made and may pass a bad test result off to you under the guise of a correct test result. 

Mechanical errors are usually much more obvious.  This could be anything wrong with the equipment used in the test method such as a worn screen, a broken fan, a dead heating element, etc.  Although you will not be able to eliminate these errors completely, the most important thing to do is to keep up with your preventative maintenance and be sure to calibrate your equipment on the required intervals.  Up to date calibrations are there to let you know that your equipment is operating correctly and can be the difference between doing a test correctly the first time and having to repeat it, costing the laboratory both time and money.

Environmental errors are the rarest of the three sources but should not be ignored by any means.  Small changes in temperature, airflow, air quality, etc. can all make significant swings in test results.  These sources of error are easy to forget about and are hard to quantify but the most effective solution is to be aware of them.  Know what environmental factors can affect your test and be aware of their status before and during your test. 

Even with all of the information available to us from test methods and training, we still have differences in test results.  Yesterday we had a technician that alerted us that we were getting very different test results for a Sand Equivalent test at two different labs.  Neither of the labs' results were very close to each other, showing that there was at least one sort of error occurring at possibly both of the laboratories.  Before diving in and starting a war between labs, the first thing to determine was whether or not the difference between results was enough to raise concern over.  Test Methods are commonly tested for and adjusted to control a variety of variables before they are completed and distributed to the public.  The most respected organization for standards in the United States is the American Society of Testing and Materials (ASTM).  ASTM takes extreme care in writing test methods and will go through many revisions until they determine that the test is repeatable within a specific deviation.  They test this by collecting samples from various locations across the United States and testing them at multiple laboratories with multiple technicians.  They determine a standard range within which your test results should remain for a single sample.  In their test methods they refer to this as "Precision" and there is also a section called "Bias".  The bias section is intended to state possible deviations due to human error of the test.  The sections of the ASTM methods have been invaluable to me in determining if we have problems in our laboratories.  Not only can it help to show if we should redo a test or not but it can also help in dispute resolution with outside laboratories like Caltrans.  Knowing how much a test result varies can not only make you look more knowledgeable but gives credibility to your laboratory's test results. 

Which brings me to my next point.  Knowing the precision range of a test result can help you in conversations with your plant operators.  When reporting an out-of-spec test result to a plant operator you often get a lot of finger pointing.  No one likes hearing that something they made was of poor quality and often times, instead of looking for a solution, you will see people try to place blame for who's fault it is.  I can't count how many times I've been told that our test result is wrong and that our material was in spec.  By showing a plant operator a small range of deviation between your test results vs. the allowed range by the test you have a higher chance of convincing them that there is truly a problem with the material and not the laboratory and can start looking for a solution that much faster.  However, on the reverse side of things, knowing the precision range of a test can also let you know when your laboratory is having problems so make sure that you are using this tool fairly and make sure to look at the laboratory if you see an out-of-precision set of test results. 

But back to my original problem, the differing test results between labs.  Our differing test results were indeed outside of the ASTM recommended precision range.  The first step in dispute resolution is ALWAYS look at yourself first.  Not only can this eliminate a lot of headache and embarrassment if you were wrong, but it can save your relationship with other people, people you may need someday in the future.  Our technician did just that and retested his sample and found that his test result was repeatable in his lab.  His next step was to alert everyone involved of the discrepancy.  This is what I would call the second step of dispute resolution, communication.  He asked that all of the parties involved re-examine their testing procedure and determine if there had been an error.  After examining the equipment at the other lab it was found that their solution bottle was expired and this had been the source of discrepancy in their test results.  Which brings me to the third step of dispute resolution, resolution.  Our technician did not decide that his job was done when he proved that his test result was the correct one.  Instead, he took the next step and asked himself how they would avoid this problem in the future.  He asked that all parties involved get together to review the test procedure together to be sure that both labs were performing the test the same and with the right equipment.  You see, the trick about quality control is that it is all about doing things consistently.  Even if you follow the test method to the tee there is still a lot of grey area where a technician can take a liberty that can change the entire test result.  The secret to assuring that your test results match those of another technician, laboratory, or government entity is communication and making sure that you are all doing the test exactly the same.  If our technician would have immediately accused the other lab of doing the test wrong they would have most likely gotten defensive and even if they HAD found something wrong with their testing they may have kept it to themselves out of spite.  With a bit of prep work and good communication, we are able to all sit down together with a solution for the future. 

Welcome to Quality Control, the epicenter of communication.      


No comments:

Post a Comment