Predictive Maintenance

Centrifugal compressors are a vital component within processing plants. Compressors are widely used upstream, as well as in midstream plants, to accomplish critical missions for overall production goals.

Compressors are often one of the most important assets due to their impact on capital investment and on output loss caused by possible temporary machine unavailability—and related maintenance and repair costs.

From this perspective, it is evident why understanding if a compressor is running and working properly is crucial. The capability to detect early indicators of malfunctions, and to quickly identify and understand their causes and possible remedies, greatly contributes to plant efficiency, minimizing overall operational costs.

Predictive maintenance

This is generally known as predictive maintenance. Today, the predictive approach is well represented across all the involved industries and applied to several kinds of machinery.

For many, or most, common centrifugal compressors, the predictive techniques commonly implemented are connected to the vibrational and structural dynamic aspects of the rotor’s operational conditions.

This approach, being phenomenological—phenomena separate from normal operations—can be applied in the design stage from insight by the original equipment manufacturers (OEMs) to rotor dynamics. But often, it assumes a more practical and empirical form on the operating floor, where, basically, the machine’s vibrational parameters are measured and compared to acceptable limits and used merely for alarm triggering.

Predictive strategies, based on the analysis of performance, are increasing their presence as a tool for diagnostic and evaluation of machine health status during operational time. For centrifugal pumps, for instance, the use of a machine model for the purpose of comparison of measured performance to design is relatively easy and straightforward.

For centrifugal compressors, the same process is more complex because of the dependency of compressor performance from the gas mix composition and operative inlet conditions (inlet pressure and inlet temperature). For centrifugal compressors, therefore, an approach based on performance assessment requires a more complex machine model that should embed and couple several calculation capabilities, from aeromechanics to thermodynamics.

Inlet conditions

In fact, one of the main difficulties of analyzing centrifugal compressors’ operating performance arises from the need to have, ready and available, the compressor performance map adjusted to actual inlet conditions. Usually, expected performances are described in terms of graphs of discharge pressures, discharge temperatures, polytropic heads, efficiencies and absorbed power, related to the design inlet gas conditions.

In general, inlet conditions in the field are different from specification conditions defined in the unit’s data sheets. A somewhat diffused practice consists in trying to reduce the complexity of the problem, considering the compressor head as invariant with inlet gas conditions, and applying simplified machine formulas.

While this method holds for very low pressure ranges and constant gas mixes, it introduces considerable errors as soon as the pressures go up and gas mix variability is introduced.

In these situations, in order to assess compressor performance, it is necessary to adjust the design performance to operative conditions, taking into account the full complexity of the problem, and then compare that to measured values. This is essentially the approach indicated by the American Society of Mechanical Engineers (ASME) Performance Test Code 10 on Compressors and Exhausters (PTC10).

The main purpose of this article is to present a general method and calculation tool for the prediction of centrifugal compressor field performances in off-design conditions. All numerical evaluations executed with CMAP software reported here have been developed using the most-recent thermodynamic theories and machine aeromechanical models and in accordance with PTC10.

Method description

To begin with, we can consider that, for a centrifugal compressor, performance is strictly linked to the inlet gas conditions. This consideration is valid both to design and off-design performance.

The starting point is the availability of a centrifugal compressor performance curve, the relevant gas mix composition and thermodynamic conditions (pressure and temperature). Having this input data available, the software will perform all complex calculations in a fully automated way and produce the expected compressor performances for inlet pressures, inlet temperatures and gas mix compositions, different for design/reference.

Calculation algorithms used are able to predict both machine behavior and thermodynamic real gas properties in off-design conditions. The nearby table shows the gases that can be configured into the calculation.

Gas components

From a fluid dynamic point of view, a strict similarity of flow at each performance point is necessary. For this reason, the non-dimensional parameters head coefficient, flow coefficient and Mach number must be determined. The proposed method needs the following inputs:

Reference/design compressor maps—In general, these maps provided by the OEM give a reliable indication of the machine’s capability and are considered here as the starting point. It’s clear that in case these input data should be affected by errors, these errors shall propagate in the outputs generated by this method. OEMs usually supply two different machine maps related to different moments of the compressor manufacturing process. “Expected” maps are usually issued in the commercial/design stage, as tested maps are issued at the end of the manufacturing process when the compressor is shop or field tested. In general both maps may be used as input for the method although “as tested” maps may be considered preferable.

Reference input conditions— Reference/design maps are linked to specific inlet conditions such as inlet pressure, inlet temperature and gas mix composition. This set of data shall be stated in order to proceed to off-design calculations.

Off-design input conditions—These are the conditions at the compressor inlet (pressure, temperature and gas mix composition) from which the new performance should be obtained. Off-design conditions may be some alternative inlet conditions to be considered in the compressor design stage, or may be the actual inlet condition in some specific time during compressor operation.

Starting with the above-described inputs, the method proceeds to calculate the compressor performances in off-design inlet conditions. Performance obtained as output from the described method is referenced below as off-design conditions, i.e., design performance adjusted to off-design operative conditions.

When reference is made to off-design performance in off-design operative conditions during operational time, this performance shall be indicated also as “actual” performance. Calculation of off-design performance requires the capability to extract the invariant information that describes the compressor behavior and to use this information to rebuild the performance in new conditions.

These calculations are intimately coupled with the thermodynamics of gas compression and the thermodynamics of real gas mixtures, so that a variation in each component of the gas mixture is potentially able to manifest its effects, as well as a change of the inlet pressure and temperature. The connection among these different modeling areas allows the method to provide an accurate prediction of the compressor performance.

The method does not require information about the internal parts of the machine, and in this sense it should not be considered as a design tool but more correctly as an analysis tool that starts just after the completion of the machine design stage.

How Cmap works

Consider two case studies of performance prediction in off-design conditions that present the method applied, using the Cmap software tool. The compressors studied were running under off-design inlet conditions. The analysis developed with Cmap obtained the performances in these off-design conditions and a comparison to measured field values.

In these cases, the compressor performances map, in design conditions, was available for both. Cmap software allowed calculation, at the actual flow, of the values of expected pressure, temperature, head and efficiency in the actual (off-design) conditions, and then compared them to the measured ones.

With reference to the field values, the nearby tables compare the pressures and temperatures as read from transducers to the value predicted by Cmap software.

The analysis indicated that Compressor 1 was running with operative performance aligned with design expectations. Compressor 2, differently, was running with performance not aligned with design expectations. This comparison gives compressor analysts important quantitative indications on the machines’ health status and should drive the analysis towards an understanding of possible causes that may justify the observed difference.

The analysis developed using Cmap also allows evaluations of the efficiency deviation—difference between the actual compressor efficiency and the expected efficiency in the actual operative conditions. Time trends of calculation results provided a useful analytical basis for compressor maintenance decisions.

The method has been profitably used to predict compressor performances and support planning of machinery maintenance activities

Cmap allows compressor performance predictions using different equations of state (EOS), depending on gas mix considered in the calculations. For hydrocarbon gas mixtures, the Lee- Kesler method of determining saturated vapor pressure, or Peng–Robinson equation of state, can be used.

For Freon R134a the Benedict-Webb- Rubin (MBWR) EOS should be selected to determine the thermodynamic properties of the operating fluid.

Conclusions

Experience with real machinery showed that compressor performance predictions obtained with Cmap software are in alignment with OEM predictions and field measurement for machines in good conditions. Also, experience showed that in most cases the deviation of some parameters, such as efficiency, indicates some problems.

The proposed method may be used in a fully automated way, and it could provide substantial benefits, especially for those machines that work in high pressure ranges and under rapidly time-varying process conditions.

Automated application of CMap gives the possibility to provide a continuous monitoring of performance and therefore provides an automated surveillance and diagnostic. Also, compressor protection from surges can be automatically and continuously updated to actual inlet conditions, overcoming limitations of actual systems (see the IPC Surge Protection System).

Methods proposed and described here can allow operators to:

• Predict the performances of a centrifugal compressor in off-design conditions. The prediction of compressor performances is accurate even at high pressures, where the ideal gas theory commonly used introduces considerable errors;
• Predict the modification of surge points in actual operative conditions, with different inlet pressure and temperatures, different operative gas and to implement advanced protection from surge;
• Have useful indications on the health of the compressor (diagnostics), based on the capability to analyze the performances and efficiency of the machine in a simple and immediate way; and
• Support decisions and planning of predictive maintenance and activities.

___________________________________________________________________________________________________________________________________________

Massimiliano Di Febo is an operations manager and Pasquale Paganini is a technical manager for IPC.