Published on: 


In the past few years, it has become commonplace to install centrifugal compressors in parallel with reciprocating compressors. Among other concerns related to this system layout, centrifugal compressors cannot efficiently operate with high-amplitude pressure fluctuations (pulsations) in the attached piping system. As reciprocating compressors create higher-amplitude pulsations than centrifugal compressors, it is important to determine and reduce, if necessary, the pulsations created when the reciprocating compressors are operating in parallel with centrifugal compressors.

In areas where pulsations are a concern, system mechanics can also be an issue. Centrifugal compressor piping systems are typically installed with little thought of vibrations and piping dynamics. Therefore, restraints are sometimes designed only for static loads, which can lead to high vibrations.

The Gwinville Compressor Station in Jefferson Davis, MS, for example, experienced severe 30-inch piping vibration problems in early 2009. At this location, parallel operation of a new centrifugal compressor and eight reciprocating compressors was planned for a 1,000 psia, 60°F suction piping system (Figure 1). Seven of the eight reciprocating compressors operate over a speed range of 250 to 300 rpm. Unit 8 operates over a speed range of 200 to 250 rpm. Pipe routing consisted of approximately 300-feet of piping between the new centrifugal compressor and the closest reciprocating compressor.

Severe piping vibrations were noticed near the centrifugal compressor shortly after startup. Even relatively small amplitude pulsations can cause concern regarding efficiency and operation closer to surge than would be predictable with no pulsations.

Field data acquired by Southern Natural Gas (SNG) was used to guide a pulsation analysis and mechanical piping review performed by Southwest Research Institute (SwRI). Vibration amplitude of 10 mils peak-to-peak (pk-pk) at 8.3 Hz was measured on 30-inch piping at the 90° elbow just outside the centrifugal compressor building (Figure 2).

Vibration amplitude of 10 mils pk-pk at 3.4 Hz was measured at the pipe support near the surge valve (Figure 3). All field-measured vibrations were within the “design” region or lower when plotted (red stars on Figure 4). However, vibrations measured during the field study were much lower than those experienced earlier. The maximum pulsation amplitude was 3.7 psi pk-pk at 8.3 Hz in the piping near the compressor inlet.

That magnitude could create a shaking force of approximately 2,400 lbf. As the measured surge valve vibration frequency did not correspond with any significant pulsation frequency, it is likely that the 3.4 Hz vibrations were not driven by system pulsations. However, the measured pulsation frequency did correspond with the measured 30-inch suction piping elbow vibration frequency, indicating that the 8.3 Hz vibrations were being driven by 8.3 Hz pulsations.

In addition to the field study, a pulsation analysis covered each of the eight reciprocating compressors and the centrifugal compressor. Pulsation data were predicted at various locations. During field testing, Unit 8 was operating at 250 rpm during pulsation measurements, and the second order of excitation associated with a running speed of 250 rpm is 8.3 Hz. When Unit 8 was the excitation source in the pulsation model, a dominant resonance was identified at approximately 8 Hz (Figure 5).

The predicted resonance provided reasonable correlation with field data. After model correlation was accomplished, system modifications could be evaluated. Both side branch absorbers (SBAs) and orifices were simulated to determine which type of system modification would be most effective. Modeling determined that an SBA was more effective and does not result in as much pressure loss as that of a typical orifice.

Therefore, an SBA was sized and recommended. As can be seen when comparing the green curve (SBA installed) with the red curve (no SBA) in Figure 5, it was predicted that pulsation amplitudes would decrease by approximately 83% at the 8 Hz resonance. When applying that amount of pulsation amplitude reduction to the field measured pulsation amplitude, it was predicted that the field measured 3.7 psi pk-pk amplitude would reduce to approximately 0.6 psi pk-pk.

A mechanical piping review (MPR) evaluated the piping system’s dynamic mechanical characteristics. Vibration amplitudes were not predicted during the MPR, but the mechanical natural frequencies (MNFs) of the piping system were estimated. If during the MPR, the MNFs were estimated to be relatively low or close to the excitation frequencies (pulsations), recommendations were made in an effort to increase the MNF of each suspect span of piping.


Figure 6 includes notations that identify where clamp type restraints, U-bolts, and piers existed prior to the MPR. Additional restraints that were recommended as a result of the MPR are also indicated on Figure 6.

Recommendations were made as a result of the pulsation analysis and the mechanical piping review in an effort to reduce the pulsation energy (or the driving forces) and to increase the MNFs of the piping system such that the MNFs were above any residual pulsation energy. Reducing pulsation energy and increasing piping system MNFs generally results in the overall goal of the analyses and the owner company: reducing vibrations. At the Gwinville Compressor Station, the recommended modifications resulted in a noticeable reduction in vibrations.

Lessons learned

It can be concluded that it may have been possible to avoid the vibration problem if analyses were performed during the design phase. Considering the fact that low-amplitude pulsations (generally in the range of 1-2 psi pk-pk) are required in the piping system near centrifugal compressors, it is important to evaluate the need for pre-installation analyses of acoustics and mechanics in the new centrifugal compressor piping system.

It should also be noted that restraints are typically minimized during the design phase in an effort to stay within the maximum allowable static load on the compressor suction or discharge connection. Restraints are then added after construction if vibration or pulsation problems exist.

Most compressor suppliers are applying factors, such as 5.55 and 3.00 to the allowable load guidelines in NEMA SM23 and API 617, respectively. These safety factors are negotiable, and that which may first appear as unacceptable overloading may be deemed structurally acceptable upon request.

This methodology (discussions and negotiations with the OEM) may help bridge the gap between conflicting restraint strategies that exist when addressing vibration control and allowable compressor nozzle loads.


1. Brun, K. and Kurz, R., “Analysis of the Effects of Pulsations on the Operational Stability of Centrifugal Compressors in Mixed Reciprocating and Centrifugal Compressor Station,” ASME GT2008-50540.

2. Brun, K. and Kurz, R., “Compressors are not pipe anchors,” Turbomachinery International, September/October 2010.


Eugene Broerman is a Senior Research Engineer in the Fluid Machinery Systems Section at Southwest Research Institute (SwRI).

Klaus Brun is Machinery Program Director at SwRI.

Kirk Mueller, P.E. is a Program Manager at Kinder-Morgan.