Hybrid Experimental And Analytical Approach To Reduce Low Frequency Noise And Vibration Of A Large Reciprocating Compressor

This paper describes the experimental and analytical techniques used to identify and correct the noise and vibration concerns in a new family of reciprocating compressors. Operating sound, vibration and pressure pulsation data were acquired simultaneously on a prototype compressor. Analysis of this data led to two areas of concern. The first concern was identified using Operating Deflection Shape (ODS) analysis of the compressor in conjunction with an uncoupled vibro-acoustic model of the suction cavity. The second concern was identified using experimental impact tests on individual components in conjunction with a finite element model. In both cases, experimental data were not only used to diagnose the problem but also to validate models of the baseline configuration. The models were then used to investigate possible countermeasures. EXPERIMENTAL INVESTIGATION Two prototypes of a new family of large reciprocating compressors were used for the noise and vibration troubleshooting: one was intact and was used for the operating tests while one was un-welded and was used for the modal investigation of the interior sub-system. The complete compressor was first evaluated by measuring its sound power level at different conditions of suction and discharge pressure and at 50 and 60 Hz line frequency. The measured sound power spectrum of the compressor at ARI conditions and 60 Hz line frequency is shown in Figure 1. The corresponding overall sound power level was 73.9 dB (A) and the ARI SRN was 7.6. 20 30 40 50 60 70 80 50 63 80 10 0 12 5 16 0 20 0 25 0 31 5 40 0 50 0 63 0 80 0 10 00 12 50 16 00 20 00 25 00 31 50 40 00 50 00 63 00 80 00 10 00 0 12 50 0 16 00 0 20 00 0 To ta l Frequency (Hz) So un d Po w er L ev el (d B ) Linear Corrected A-Weighted Figure 1. Sound Power level at ARI at 60Hz The two 1/3-octave bands at 250 Hz and 500 Hz clearly suggest the presence of troublesome tonal components. These were then investigated by performing line frequency sweep tests to search for system resonant frequencies. The compressor was instrumented with four triaxial accelerometers on the housing (three around the lower part and one at the top), two high pressure microphones at the top and bottom of the housing, as well as dynamic pressure transducers in the suction and discharge lines. Four microphones around the compressor were also used to measure radiated noise, and the motor current signal was acquired as a line frequency reference. A total of twenty-one channels were acquired simultaneously using an Agilent VXI front-end driven by commercially available data acquisition software running on a NT workstation. The line frequency was swept from 65 Hz to 45 Hz in about 60 seconds, while static suction and discharge pressures were manually held constant. The time histories of all data were first stored to disk then processed into sets of frequency spectra that can be shown as 3D plots or Campbell diagrams. The discharge pressure pulse was band passed using a digital FIR filter and used to extract the pump speed information. This pump speed was compared to the line frequency calculated from the current to produce a measure of the motor slip versus time. Figure 2 shows these two RPM versus time curves for one of the speed sweep tests. The difference between the two curves shows the motor slip, which is about 150 RPM at the 60 Hz line frequency. Time Response