Transportability of measurements

Jan. 20, 2006
In noncontact temperature measurement, accuracy trumps repeatability.

On many production lines, critical product properties must be measured and controlled carefully to achieve the desired product quality. For instance, when producing steel sheet that the auto industry stamps into fenders, doors and other panels, the mill must control many variables. These range from the initial chemistry of the molten steel to the way the final cold-rolled sheet is processed. Failing to control some variable can mean the stamped door panel can’t meet shape specifications. If it’s too thick, has variable thickness or lacks the correct strength or hardness, it won’t perform properly in the stamping die. It’s all about springback -- the tendency of bent or shaped metal to resist mechanical deformation.

There’s a lot more to it, though. Even the steel’s microscopic surface roughness must be controlled because the roughness profile determines how much lubricant the part retains during stamping, which relates to tearing defects and paintability issues.

For any value-added product -- meaning specifications are so tightly controlled that customers pay extra if the product is guaranteed to perform as advertised -- hundreds of measurements often must be made and controlled during processing. Such value-added products are where manufacturers make most of their profit.

Two places at once
But, most plant professionals are unaware of the problem of transportability. It’s a phenomenon that often prevents achieving the guaranteed properties in more than one place at a time. I’ll take an example from the steel industry.

A large steel mill developed high-strength steel that must be subjected to, among other things, a precise time-temperature-thickness reduction cycle to achieve the desired properties. Months of metallurgical studies determined the limits of temperature and other variables that achieve those properties.

But these studies were based on the specific instrumentation that was used to measure the variables. Another steel plant tried to produce the identical product using the “same” measured variables for its production and was surprised to find that its product didn’t meet specs. Plant management found this troubling. From a bottom line vantage point, which is what it’s all about, a single coil of off-spec steel might represent a $50,000 loss. Produce more than a few off-spec coils and people begin to holler. That’s the sound of real money talking.

My contract with the plant involved an investigation that revealed different instruments and settings were being used at the two plants to measure the same critical temperatures. This is important because the noncontact radiation thermometers (pyrometers) being used might be different types (single color versus ratio), might use a different wavelength, might be adjusted (correctly or otherwise) at a different emissivity, might be sighted at a different angle, might exhibit different atmospheric effects, or might have different reflection effects. Any of these factors can alter the measured temperature.

For example, the first plant might determine the desired temperature range to be 1,250°F to 1,350°F, but the second plant might measure it as 1,350°F to 1,450°F. And the difference isn’t necessarily linear. The process criteria developed at the first plant weren’t transportable to the second plant without modification.

This isn’t meant to imply that the measurements at either plant were correct. The second plant might need to run its own metallurgical tests for months to determine what settings it must use for its particular instrumentation and settings to achieve the same results as the first plant. The delay means lost time and money.

A big difference
The plant that developed the process knew the results it achieved using its equipment and settings were repeatable. But repeatable isn’t the same as accurate. For a process to be transportable, every critical measurement must be accurate as well as repeatable.

And that isn’t easy to achieve with different instruments, setups and settings. Think industry-wide rather than locally, which often goes against the grain. Don’t forget the human factor -- a facility’s management doesn’t wish to think they’ve been doing things incorrectly for years.

Transportability isn’t limited to different facilities using different equipment or settings. It can happen on the same operating line after new equipment is installed. For example, several years ago, we consulted for one facility where the personnel complained that the noncontact radiation thermometers I recommended didn’t produce the same readings as the old system. The plant had made the upgrade intelligently, installing the thermal systems in parallel and logging both sets of readings over a period of months. The indicated differences were large enough to justify modifying their operating variables, which they hesitated to do.

The plant engineers failed to grasp the point of the exercise. I had recommended new instrumentation because the readings would be more accurate and more repeatable. This would allow the plant to produce more on-spec product. To this day, plant personnel continue to translate the readings from the new instrumentation into terms of the old instrumentation and they achieved no quality improvement.

Unintended consequences
Changing instruments isn’t always the problem. In another case, for example, I received a panic call from the hot-rolling division of a steel mill. They had been making on-spec product for years, and suddenly most of the product was off-spec. I gathered up my own instrumentation, flew to the plant and started logging my independent data to compare it with what the plant’s computer was logging. Each of the plant’s temperature readings was consistently less than mine.

I chanced to speak with the plant’s IT people, who told me that steam occasionally blocked the pyrometer view and caused the signals to be a bit erratic. The IT department decided to install a signal averaging module to smooth out the readings. This innocent modification was the source of the problem. Instead, the IT people should have opted for peak picking, which always looks for the highest signal.

Digital assistance
Fortunately, in the art and science of noncontact temperature measurement, it’s possible to use computer modeling to predict how any particular instrument will react to given settings and conditions. I’ve made extensive use of such software to predict how new or different instruments, settings and conditions will affect apparent temperature readings. But most plant personnel don’t know how to perform such computer simulations, nor even why they should use them if they could.

My goal is to help plant professionals understand and achieve transportability. At one steel mill, I surveyed the operations about exactly what equipment was being used, design details, settings being used, how often calibrations were performed and details of the calibration routine. This led to an “intelligent” corporate-wide computer program that allowed operating personnel at any plant to determine what the sister plants were doing and whether a proposed change or initiative was a recommended operating practice.

Amazingly, many of the mills never calibrated their equipment, assuming that accuracy was somehow guaranteed for life. This program began to help, but it also created a lot of arguments. Some of the plants were accustomed to doing things their way. Making changes to achieve accuracy instead of repeatability would be disruptive and costly in terms of testing. The plant personnel failed to see any long-term benefits in the program.

This attitude is particularly acute in the field of temperature measurement, but it’s certainly not limited to that arena. For more than 38 years, I’ve encountered it in fields as diverse as measurements of surface topography, lubrication and reduction, to flow and thickness measurements.

Part of the difficulty, sadly, is that manufacturers’ reps don’t always understand the problems involved with their own equipment and sometimes recommend the wrong equipment and settings. Even reputable testing labs can return wildly inaccurate data if they sometimes assign untrained personnel to perform the testing.

Ralph G Rudolph is president of R. Rudolph Consulting LLC. Contact him at [email protected] and (505) 828-3938.

Measures that help achieve transportability
  • Determine what variables must be measured and controlled precisely during production of your product. 
  • Determine which instruments can measure the variables most accurately and how to best engineer their installation and settings. Don’t always believe the first vendor you meet. Or the second. Benchmark against similar facilities and find out why they do things the way they do. Retrain your personnel as necessary using what you learn. And communicate throughout your company.
  • Establish a standardized calibration program for your instruments and determine how often the units need to be calibrated. Then, follow your program rigorously.
  • Remember that accuracy is most important. Mere repeatability can -- and will -- get you into trouble.

Sponsored Recommendations

Reduce engineering time by 50%

March 28, 2024
Learn how smart value chain applications are made possible by moving from manually-intensive CAD-based drafting packages to modern CAE software.

Filter Monitoring with Rittal's Blue e Air Conditioner

March 28, 2024
Steve Sullivan, Training Supervisor for Rittal North America, provides an overview of the filter monitoring capabilities of the Blue e line of industrial air conditioners.

Limitations of MERV Ratings for Dust Collector Filters

Feb. 23, 2024
It can be complicated and confusing to select the safest and most efficient dust collector filters for your facility. For the HVAC industry, MERV ratings are king. But MERV ratings...

The Importance of Air-To-Cloth Ratio when Selecting Dust Collector Filters

Feb. 23, 2024
Selecting the right filter cartridges for your application can be complicated. There are a lot of things to evaluate and air-to-cloth ratio. When your filters ...