Dr. Dan's Tech Notes

 

  Technical Notes by Dr. Dan Barberree, 

Vice President, Research and Development

Each month, AccuTru publishes a newsletter for our distributors.  One of the features of the newsletter is a technical article by Dan Barberree.

Below, is his series on thermocouple drift.

 

Decalibration and Drift

Thermocouple Decalibration and Drift – Part 1

All thermocouples are subject to calibration drift with use, it is just a matter of how much, and how fast this may happen. Thermocouple performance is critically dependent upon absolute uniformity of both physical and chemical properties along the entire length of the wires in the circuit. This is because the thermoelectric emf produced by the thermocouple is a combination of the emf produced at every point along its length.

Thermoelement materials are carefully produced to assure that uniformity (or homogeneity) is achieved. However, when placed in service, different parts of the thermoelements experience different conditions of heat, chemical exposure etc, and as a result those parts actually do become physically and chemically "different" over time.

Because the thermoelectric emf (and therefore the temperature reading) depends on the chemical and metallurgical properties of the wire along it’s entire length, the total emf produced by a used probe can be different from an otherwise identical new one under the same conditions. Fortunately in many applications the changes over time are small. But under adverse conditions, large drifts at rapid rates can and do occur and the temperature reading can be far from the true temperature in the process.

Some processes can tolerate small errors in the thermocouple measurements from the true temperature without being adversely affected. However in many processes the temperature measurements are critical to process safety, yield, energy consumption, equipment life and environmental compliance and even small deviations in the readings from true can have significant economic impact.

To achieve long and reliable thermocouple life, the usual strategy is to operate the device comfortably under its maximum temperature, and provide it with the cleanest possible environment in which to work. Protective sheaths, tubes, and thermowells are often employed to try to control the conditions that surround the thermoelements themselves.

 Thermocouple Decalibration and Drift – Part 2

Since all thermocouples are subject to calibration drift with use, it is just a matter of how much, and how fast this may happen.

Protective sheaths, tubes, and thermowells are employed to try to reduce damage to the thermoelements but drift still occurs.

This is a serious problem for many users. So what do users do?

Here are the traditional ways users try to compensate for thermocouple drift and some problems associated with each:

Scheduled Replacement – Simply remove and replace them on some regular frequency not knowing whether they have de-calibrated or not.

·       This can be costly both in sensor costs and in process losses if they are not replaced frequently enoug

Redundant Sensors – Backups, Multi-points or Bundles of Sensors

·       Which one do you believe?

·       May use sophisticated “voting” systems

·       Like sensors may drift off together

·       Systems can become costly and you’re still not sure

Cross-checking against other Sensors in the process - Using Material and Energy balances or a combination to try to decide if the measurement seems correct.

·       Requires computer modeling and good process models

·       Balances are often hard to close in real time

·       Other sensors also de-calibrate, plug and wear out

·       Comes down to deciding which of your measurements is most likely wrong

Calibrations and Re-calibrations – Comparison with Traceable Standards

·       Cannot prevent de-calibration

·       Read the “Fine Print” on your certificate*

·       Insertion depth effects normally cannot be reproduced in Cal Lab

·       Can be very labor intensive, time consuming and costly

*"Tolerances indicated in this table are not necessarily an indication of the accuracy of temperature measurement in use after initial heating of the materials.”

Any of the above situations is a dead give-away that the User has drift problems and knows it.

Thermocouple Decalibration and Drift – Part 3

 What causes Thermocouple Drift?

Thermocouple performance is critically dependent upon absolute uniformity of both physical and chemical properties along the entire length of the wires in the circuit.

Why is this so important? It is because the thermoelectric emf produced by the thermocouple is a summation of the emf produced at every point along its length. This fact is difficult to explain theoretically but has been proven experimentally. If you think about it for a minute, this also makes sense logically. Without a temperature difference, there should be no emf produced. Note: Many thermocouple users are not aware of this phenomenon.

The result is that the emf produced by a thermocouple is generated in the temperature gradient. In other words, it is generated in the section of the thermocouple where the temperature changes. This can be anywhere along it’s length as it transitions from the temperature at the tip to the temperature at the “cold junction” or measuring end.

Therefore, anything that affects the uniformity of the thermocouple elements along their length can cause drift.

Next time we’ll discuss some causes of non-uniformity (also called inhomogeneity).

 

Question to ponder: If both ends of a thermocouple are at the same temperature will there be any emf produced?

Thermocouple Decalibration and Drift – Part 4

Why worry about Non-uniformity (also called inhomogeneity)?

Last time we posed the question: If both ends of a thermocouple are at the same temperature will there be any emf produced?

The answer is: No, there shouldn’t be, but there can be.

Actually it’s a little more complicated than that. emf, or electromotive force, is another term for voltage. A thermocouple acts like a little battery producing a voltage across the leads if the remote end is at a different temperature than the measuring end. emf is produced by a thermocouple anywhere a temperature gradient exists.

 

1.      If the entire sensor is at the same temperature along it’s length, there will be no emf produced anywhere and no net emf or voltage signal.

 

2.      If both ends of the sensor are at the same temperature, but somewhere along it’s length the temperature is higher or lower, emf will be produced, but it will cancel out and no net emf will be produced IF the wires in the sensor are uniform or homogeneous along their entire length.

 

3.      If the wires are NOT homogeneous, temperature gradients can produce emf contributions that do not cancel out, and there can be a net emf or voltage signal produced. This is of course an erroneous signal in our example caused by what we call “inhomogeneity”.

In practice, inhomogeneity in thermocouple wires routinely causes erroneous signals. Often they are small, however there are documented cases of huge errors of as much as 50% of the signal. The real problem is that in conventional thermocouples you don’t know if it’s there or not. If you just want to know if it’s hot or not, this might not make a difference to you. But for most industrial applications choosing the right temperature sensor for the job and taking care to install and maintain it correctly is very important.