My test is accurate

Published on: 
Turbomachinery Magazine, March/April 2021,

Many of us have to verify compressor and driver performance in the field. We do a test, read the data, and convert it to flow, head, and some other parameters. Voila, we have done a valid test and we believe our data. Not so fast!

The first question is: What data do we actually need besides the obvious such as pressures and temperatures at the compressor suction and discharge, and the flow? Things like the gas composition, the compressor speed, and ambient conditions are also required. Gas composition is often neglected because it requires extra effort to sample gas and send it to the lab or at least have a calibrated on-line gas chromatograph. It obviously must be the gas composition at the time of the test, not some mixture sampled weeks ago. Even in pipelines, gas compositions changes frequently.

Test data is subject to uncertainties. As well as accuracy of the end-device, this includes the entire measurement chain, instrument location, number of instruments, and process stability. There can be systematic or random uncertainties. In field performance tests, systematic uncertainties dominates; averaging a lot of data does not automatically help. Many tests have uncertainties exceeding several percentage points, especially for low pressure ratio compressors. In these cases, the test must be done with accurate and calibrated instrumentation.

Unsteady operating or process conditions are particularly problematic for the accuracy of test data. It is near impossible to correct for unsteady operating conditions from process changes or ambient conditions. Most test codes were written for factory testing and provide allowable limits for unsteadiness. But these are often not easily achievable in a field test. A good test plan will include steps to ensure steady operating conditions for the duration of the test.

"It is possible to obtain valuable and accurate test data in the field. But it requires planning, preparation, careful testing, and the understanding of fundamental principles."


If a test point is to be compared with a compressor map to determine if the measured efficiency meets manufacturer predictions, we need to know compressor head, speed, and flow. These measured values each have associated uncertainties. Compressor speed is usually measured accurately whereas flow and head can have significant uncertainties, depending on instrumentation, operation, and process conditions. Note that flow measurement impacts the comparison of the measured and the predicted efficiency – if the flow is measured incorrectly, the measured efficiency is compared to the wrong efficiency on the compressor map.

It is often a good idea to plan for redundant measurement. For example, one can determine a compressor’s absorbed power from both thermodynamic measurements on the compressor and from data gathered from the driver. For a gas turbine driver, the engine fuel consumption, data from the engine factory test, or (if installed) from a shaft torque meter measurement, allow for direct comparison of driver and driven power to check if the measured compressor performance is reasonable. For electric motor drivers, the required power can usually be obtained from current measurements at the motor terminals. However, gearbox losses and motor efficiencies must be considered.

There are two reasons to test a compressor: to verify performance or to use measurements to verify the power from the driver. For the latter, we don’t need to test compressor efficiency. We only need compressor absorbed power. This also means that it’s not important where on the map we operate as long as the compressor absorbs the driver’s full load. Accuracy of power measurement depends on many variables.

Just to be clear and for all who are now worried about their test results based on this article’s title: It is possible to obtain valuable and accurate test data in the field. But it requires planning, preparation, careful testing, and the understanding of fundamental principles. ■