“THE COST OF EVERYTHING, THE VALUE OF NOTHING” - DOES YOUR DATA ADD COST OR INCREASE VALUE?

Data costs money, of that there is no doubt. Often a prime driver in acquiring data is the bottom-line cost. What is rarely asked is the value of data. It never ceases to amaze me how often test programmes are performed on the cheap, and then we get a call to explain why our sensors or instruments gave the “wrong” results (answer – never, with too many anecdotes to list here!)

What should be obvious is that the only difference between a well-defined and implemented test and a poor one is not a significant cost differential, but a big value difference. Poor or missing data has an infinite cost per quantity, whereas quality data has very high value.

Repeatable tests that allow engineers to evaluate each data set in real time pays dividends because the test parameters can be modified and problems diagnosed before continuing. Performing pre-tests, or intermittent data evaluation, is essential. I had an irate customer complaining that our instrumentation system had produced faulty (noisy) data, with an entire week’s worth of testing resulting in no usable information. A simple check revealed a loose earth strap on their shaker table – a turn of a bolt and the data was now virtually noise-free.

One-off tests such as destructive impact testing are not repeatable and require significant levels of planning and evaluation before the button is pressed. I personally witnessed an impact test where the first low-level impact caused a small unplanned issue which propagated on the second of three runs to cause a huge, almost explosive, failure. In hindsight the small anomaly could have been corrected, but the time and cost constraints caused the engineers to push ahead regardless. Months of planning resulted in little usable data.

It is common during test planning to list multiple sensors for critical areas, which gets pared back to fit within the budget. On more than one occasion, after attending a customer’s test programme, I’ve overheard it said that they wish they had a few more sensors. Now they must consider repeating the test programme to get the data that was missed! If a test costs £100k, why risk saving £1000 on the sensors?

It’s not always necessary to create a complex test providing you ask the right questions - a “quick and dirty” test can sometimes be the best solution. Investing in a quick evaluation by, say, bonding one or two strain gauges can reap dividends by evaluating whether a test is valid. I fondly remember a company who spent many thousands on a high-speed vision system which failed to diagnose a jamming mechanism, but with two strain gauges we found that one of two levers was not contacting anything during deployment. Occam’s razor can be applied here – simple solutions are sometimes the most effective.

Prepare to be flexible. Test programmes do not always go to plan, and it’s better to delay than to push ahead in the face of uncertainty. I’ve seen entire weeks wasted for want of a simple sanity check. Delays in manufacturing seem to eat into testing time because a test date is set in stone, leading to rushed last-minute installations; requesting a 24-hour turn-around when a component requires 3 days in an oven is surprisingly common. Gauges cannot be installed until the hardware has been received. They cannot be placed until an engineering drawing with clearly-identified locations and orientations, and leadwire routing, has been provided. Rushed installations have often resulted in questionable data at the end of a test programme.

The lesson that I learned over the last 30 years is that before embarking on a test programme the right questions must be asked.

What do we want to test?

What measurement quantities do we need (load/pressure/strain/stress/temperature/displacement/acceleration)?

Does the test match the real-world environment?

How much time does it require to mount the sensors?

Do we have the right instrumentation?

What analysis tools do we need?

What are the errors and uncertainties?

What are the associated risks?

Using quality products from well-recognised suppliers and manufacturers can add significant value. VPG Micro-Measurements have long been known as the premium manufacturer of strain gauges, with our Advanced Sensors Technology strain gauges further improving our historically-excellent accuracy, stability and reliability. StrainSmart data systems are well-recognised, and Pacific Instruments bring another layer of capability and performance to our instrumentation range. Our reputation is further supported by our own expertise in providing training and support from the initial enquiry all the way to the acquired data. You can contact one of us at any stage in the process and get expert advice.

Asking the correct questions, choosing appropriate sensors types, defining locations and orientations, planning the test programme carefully and above all being flexible can help with obtaining high-quality, high-value data. Choosing the right supplier can add significant value too; VPG Micro-Measurements are the #1 choice!

achittey's picture

Anton Chittey

United Kingdom