April 9, 2020
Klaus Brun & Rainer Kurz

Klaus Brun is the Director of R&D at Elliott Group. He is also the past Chairs of the Board of Directors of the ASME International Gas Turbine Institute and the IGTI Oil & Gas applications committee. Rainer Kurz is the Manager for Systems Analysis at Solar Turbines Incorporated in San Diego, CA. He is an ASME Fellow since 2003 and the chair of the IGTI Oil and Gas Applications Committee.

Testing turbomachinery requires corrections between actual test conditions and those originally specified as the acceptance, operating or design conditions. The factory or field test will usually be performed at different ambient conditions, different process gases, and different pressures and temperatures.

As a result, the industry has developed procedures to determine valid conclusions from tests at different operating conditions. Once the machine has left the factory and operates at different conditions from originally specified, how does one correct from one set of conditions to another?

For centrifugal compressors, this procedure is relatively straight forward. The relevant ASME Power Test Code (PTC 10) provides well thought out and proven methods for data correction. Correction and test methods are based on the concept of finding aerodynamically equivalent operating conditions using the fan laws, and keeping Mach number and Reynolds number deviations within prescribed limits.

However, if the machines have multiple sections, intercooling, or incorporate side-streams, data correction analysis can be difficult.

Similarly, for steam turbines, especially the type with condensing flow back-ends, data correction methods are challenging: There is a phase change and liquid fractions must be assumed or measured.

It is important to note that ASME PTC 10 describes the correction procedure for specific operating points, but not for a complete compressor map. This means that if operating conditions range widely, the entire map must be recreated and there is no single correction factor.

The testing of a gas turbine creates a number of different complications. A gas turbine consists of two or more sets of turbomachinery, i.e., there are at least one compressor and one or more turbine sections.

Between the compressor and the turbine, there is a combustor where fuel is converted to heat. This means any method that uses the concepts of aerodynamic similarity is limited by the fact that it is nearly impossible to have all components at aerodynamic similarity at the same time.

Additionally, the parameters that impact performance, such as clearances, do not follow similarity rules and may be dependent on actual temperature levels. As there are multiple aerodynamically and thermodynamically integrated components in a gas turbine, their performance is not independent. This becomes more complicated if the power turbine is tested separately from the gas generator section of the engine.

Correction curves

An easy way out seems to be the creation of correction curves. These are sets of curves that separately correct for engine inlet temperature, ambient pressure, inlet and exhaust losses, relative humidity, power turbine speed, and others. Curves also exist that correct for part-load operation.

At first sight, this may appear to be a perfectly valid approach. Unfortunately, these correction factors are not independent. If they were independent, the only action needed would be to multiply all individual correction factors to yield the overall correction for power and efficiency at any given site condition. But that is an incorrect assumption since these correction factors are physically interdependent.

For example, the impact of relative humidity depends on the engine inlet temperature. Similarly, for two-shaft generator sets, even though the generator runs at a constant speed, the power turbine driving the generator operates at various distances from optimum, which depends on ambient conditions and load.

Correction curves do not predict the impact of control settings. And in a gas turbine, clearances are not determined by ambient air conditions, but by temperature levels within the engine. Therefore, even if aerodynamic similarity is accomplished, the running clearances may be different.

The accuracy of curve-based corrections is typically enough for simple troubleshooting and condition monitoring purposes. Such curves can also be used for day-to-day operation monitoring by correcting for ambient temperature and elevation.

The engine is tested at the prevailing ambient temperature. The relative difference between the tested power or efficiency, and the value on the correction curve are calculated. The same relative difference is applied to the curve at the originally specified ambient temperature. This yields the engine performance at that ambient temperature within a reasonable accuracy appropriate for condition monitoring purposes.

What can the operator do when higher accuracy is required? One way is to use the cycle performance deck of the manufacturer. Many condition monitoring systems and digital twin approaches do that. The use of the software, which is usually proprietary, then becomes a trust issue. ■