Reliability Direct February 2003 Newsletter

For too long thermographers, after finding problems, have been stymied about how to accurately and repeatably assess the severity of their findings. Because appropriate standards have yet to be developed for most major applications, many thermographers have simply opted for using "rules of thumb," often with less than optimum results.

Nowhere is the need greater than with electrical inspections. Here, we often use temperature based systems of prioritizing the problems found. Unfortunately none of these systems have been scientifically proven to be reliable. They are also subject to dangerous misuse. The factors influencing temperature are many and varied, including, among others, load, wind (for outdoor utility inspections), emissivity, and the thermal conductivity between the source of the heat and the surface we view. Of course, the criticality of the component and ambient conditions also influence the repair priority a finding should receive. How can all these factors-and others-be included in our analysis?

One possibility is to try to quantify all of them systematically and use the results as feedback to teach the system to correct itself. While such a system would benefit from being computerized in the long run, initially it can be designed and tested using a simple paper based format. And, because it can be self correcting, you don't need to wait to "get it perfect" before implementing such a system. Just give it your best shot and get started testing it.

How can this be done? Begin by determining the factors that are important in your decision making process. We suggest you involve a team composed of all the key players, including technical, safety, management, and accounting personnel. Factors often used include safety issues, costs of failure, availability of spare parts and maintenance personnel, history, and loading conditions. Each factor is assigned a value of one to ten based on an assessment of the problem.

An example of a weighted assessment matrix:

Score =	2	4	6	8	10	x=	Total
Is the component critical?					**	2.5
Is safety or risk high?					**	3.0
Is emmitance low?						1.5
Is IFOV within spec?						1.0
Will load increase?						2.0
Will ambient increase?						1.5
Is convection high?						2.0
Is history poor?						2.5
Is heat source hidden?						1.5
Is temperature increasing?						2.5
Total						Total

For total scores of:

180+ = repair immediately
120-179 = schedule repair
75-119 = repair next available
<75 = monitor
** repair immediately, regardless of total score

* This example should be modified before using to suit your particular needs.

Next, assign a "weighted value multiplier" to each of the key factors. Safety, for instance, may be so important that its assigned value is tripled before being added in with the sum of the other factors. Weighted values enable you to assign priority to factors you consider to be more important than others. In fact, some factors may not merit a multiplier.

As can be seen in the sample matrix, all that remains to be done is to add the values of each factor to find a total score for the finding. The total score of a problem will then suggest a repair priority based on all important, interrelated factors. It is possible, however, that this method of scoring could result in a serious problem having an total score that is low. To prevent this you can require problems having high scores in certain factors to be considered serious regardless of their total score. For instance, you may determine that if a safety factor score is five or above, the problem will automatically be a first priority, even if other scores are low.

For this system to accurately predict the priority of a finding, the predicted outcome must meet the actual outcome. It is probable this will not be the case 100% of the time. But accuracies approaching 100% are achievable if a feedback component is added. For all major findings, it is vital to compare the actual conditions found during repairs to those that were predicted. If results equal predictions, keep doing what you are doing! When they do not match, appropriately adjust the method of scoring so that prediction and actual results do, in fact, match. In short order your systems will be fine tuned to produce better correlations. Obviously careful record keeping and cross training of people with whom you interface will also be essential to success.

Click here for a look at our Infrared Imaging Camera systems.