Case study: Preventing surge in a medium-size process gas compressor

The following case study highlights a plant's struggle to prevent surge while using a four-impeller, 1.6 MW integrally geared centifugal gas compressor. The problem and three solutions are discussed.

By Amin Almasi April 22, 2014
The problem:
The plant’s compressor increases the process gas pressure from approximately 6 Barg to around 53 Barg. This integrally geared compressor is driven by a high-speed 3,000 rpm constant-speed electric-motor driver. The first and second stage speed is around 32,000 rpm and the third/fourth stage speed is around 45,000 rpm. An intercooler is provided between the second stage and the third stage. The compressor package is around 25% oversized considering 25% unit expansion in future.
The IGV was not provided for capacity control. The capacity control was via a suction control valve (the inlet throttle valve) that provided the pressure drop on the suction line. This control valve was the only tool for the capacity adjustment, which worked based on the pressure signals of the discharge and athe suction immediately upstream of the first stage (the first impeller). On the normal operation (the operation at 75% of the rated compressor), the anti-surge valve was always partially open. Single anti-surge valve was provided for four stages, from the downstream of the fourth stage to the upstream of the first stage (the inlet separator drum). The suction control valve (the inlet throttle valve for the capacity control) was located upstream of the inlet separator drum. 
The following important considerations should be noted for this compressor:

  • This integrally-geared compressor (four-stage machine) could not be operated as a variable-speed machine because of complex dynamic responses. 
  • IGV cannot be provided due to complexity and high costs. 
A portion of this 25% extra capacity was controlled by the anti-surge valve (the bypass valve), which was wasteful and problematic. In addition, the surge controller became active (independent from the inlet throttle valve control loop), and it often interacted with the inlet throttle valve, which caused unstable operation. Two loops acting simultaneously produced an oscillation that is the characteristic of physically-coupled, independently-controlled loops. Instabilities in the operation particularly the discharge pressure oscillations and oscillations in both the anti-surge valve and the inlet throttle valve were reported.
The solution:
The first proposed solution was to move the anti-surge valve return point from the inlet separator to the upstream of the suction control valve (the inlet throttle valve for the capacity control). This solution was not approved because it could result in poor surge prevention and the possibility of destructive surge. This solution—moving the anti-surge valve return point from the inlet separator to the upstream of the inlet throttle valve—could result in poor surge control because the anti-surge valve was a fast acting valve (response time 1-2 s), but the inlet throttle valve was a relatively lazy valve and it reacted slowly. 
This proposed solution recommended a minimum position of the inlet throttle valve to assure the anti-surge flow. The minimum position (50% or similar) of the inlet throttle valve was not be sufficient because in case of surge, the fast reaction and a flow more than the flow allowed by the minimum position would be required. Obviously, this could be a considerable restriction in the anti-surge system. The loops were independent, and there could be cases where a compressor was very close to surge and the anti-surge valve would try to open and feed more flow to a compressor, but the inlet throttle valve (because controlled by another loop) may start to close and restrict the flow. 
Another issue was the failure or the malfunction of the inlet throttle valve. In case of a failure or malfunction in the inlet throttle valve, the anti-surge system could be totally ineffective. The anti-surge return point should be as close to the compressor as possible to better feed the anti-surge flow to compressor, and this proposed change could move the anti-surge return point away from the compressor inlet. In addition, there was a possibility of new instabilities in the system because two valves in the series, which were controlled by two independent loops, might introduce new cases of instability. 
The second proposed solution was to operate a manual globe valve at the suction to provide further pressure drop and possibly further capacity control capability. This solution was not approved because it could change the process conditions at the battery limit of the compressor package, which could result in loss of validity of the compressor package design. Any centrifugal compressor is sensitive to suction condition changes. An integrally-geared centrifugal compressor is more sensitive to these changes. A solution based on the manual suction valve operation could not offer a good controlability. In addition, there were questions about how this solution could offer a better capacity control since if further pressure drop at suction was useful, why this could not be achieved by the inlet throttle valve in a controlled and proper manner. 
There was a risk that an uncontrolled manipulation of this manual valve can result in surge or other dynamic issues. Considering the flow and the discharge pressure were around 75%, which was approximately 95% of the rated flow and the rated discharge pressure respectively, the operating point was near the surge line. As a result, the operation of the anti-surge valve should not be unexpected. The anti-surge valve should be opened to keep the compressor away from the surge zone. Based on the second proposed solution (the operation of the manual valve at the upstream of the inlet throttle valve), the manual valve closing could decrease the suction pressure and flow. 
The third proposed solution was a combined anti-surge and control system. In simple terms, this was a system that measured necessary parameters and operate both the suction control valve and the anti-surge valve. To better explain, the solution was a combined control system with decoupled features rather than two independent control loops. This solution used the anti-surge valve only for the anti-surge application. This was not complex, but this was a combined control system. This eliminated the root-cause of issues, which was the concept behind the two independent loops. The combined control system was implemented, which resulted in a long-term trouble-free operation of the compressor. 
Amin Almasi is a senior rotating equipment consultant in Australia. He is chartered professional engineer of Engineers Australia (MIEAust CPEng – Mechanical) and IMechE (CEng MIMechE) in addition to a M.Sc. and B.Sc. in mechanical engineering and RPEQ (Registered Professional Engineer in Queensland).
Edited by Jessica DuBois-Maahs, associate content manager, CFE Media, jdmaahs@cfemedia.com.