Wet gas compressor performance evaluation

Abstract

The demand for oil and gas from the maturing Norwegian continental shelf (NCS) remains high. Even though the concern for climate change is shifting energy production towards renewables, production on the NCS is expected to continue for decades to come. If carbon capture and storage is managed and combined with technologies, such as conversion from hydrocarbon to hydrogen, production could be extended even further.

Wet gas compression, especially for subsea installations, is a promising technology for both new and existing fields. It can enable the tie-in of remote reservoirs to the existing infrastructure and allow for simplified process solutions thus possibly reducing both investment and operational costs. Furthermore, by locating compressor stations near the subsea well head increased recovery can be achieved because the well can be produced at a lower pressure compared to topside compression.

A prerequisite for the successful installation of subsea wet gas compression is exceptionally reliable solutions, due to the excessive cost of any intervention. It is thus necessary to increase knowledge on key aspects of the system, such as rotor dynamics, transient behaviour, performance and fouling.

The focus of this thesis has been to investigate how wet gas affects the performance of centrifugal compressors, how performance parameters should be established and to investigate if models can be built that allow for correction between different wet operating conditions. Experimental results were partly obtained from the test rig at the Norwegian University of Science and Technology (NTNU) and partly from Equinor’s test facility at K-Lab. The NTNU open-loop test rig has a single impeller centrifugal compressor and runs on water and air under ambient conditions. Equinor’s test facility at K-Lab has two full-scale test loops that operate at high pressures and temperatures and utilize mixtures of hydrocarbon gas, condensate and water.

As regards wet gas compressors, there is currently limited knowledge on how the performance will shift for changing inlet conditions. Therefore, the vendors are unable to estimate performance for a specific wet inlet condition a priori. Furthermore, no international standard has been established that specifies how to evaluate wet performance. Indeed, much work is required to sufficiently describe the performance and test procedure for wet gas performance. The current work aims to illuminate key areas of wet gas performance for the design, test, and operational phases.

When it comes to the design and testing of wet gas compressors, it is not clear how to compare results between different operating conditions, such as test and real operation. This problem was analysed first by reviewing the dry gas similarity theory and then expanding these concepts to include wet gas. Finally, the appropriateness of such expansion from dry to wet gas was discussed. This work documents that Type2 testing (large difference between test and specified condition) can be difficult to achieve for wet gas compressors.

A two-impeller wet gas centrifugal compressor was tested at K-Lab. The performance test results were analysed and key factors affecting performance parameters were identified. Furthermore, a model for wet gas performance based on these parameters was created. The method used to address this issue was a combination of regression and minimization of an objective function. The proposed model showed the capability of collecting the data points obtained under different operating conditions onto a “performance surface” with good precision.

Limited knowledge exists about how uncertainties propagate through a complex wet gas test loop, such as the VGII loop at K-Lab, and eventually result in uncertainties of wet gas performance parameters in general. To address this issue the Monte Carlo method was used in combination with a sensitivity analysis. In this way, the uncertainties could be propagated through the entire system. Thus, standard uncertainties and coverage intervals for the performance parameters could be established, as well as the identification of key input parameters that affect these uncertainties. Somewhat elevated uncertainties were found for the wet compared to dry conditions, but still within reasonable limits. Furthermore, many of the performance parameters are highly sensitive to inlet and discharge temperature, especially those relying on enthalpies. By including a torque meter and a gas density meter in the wet gas test facility, these uncertainties could be significantly reduced.

The detrimental effect of fouling on wet gas compressor performance was observed during testing at K-Lab. To further investigate this effect, a fouling test under controlled conditions was conducted at NTNU. The experimental results show that wet performance characteristics are heavily affected by flow path fouling. Furthermore, to quantify how the wet performance is affected by the fouling, a model was developed to correct all the efficiency curves to the dry clean curve. The model was able to collect the spread between clean and fouled curves for the same GMF.