Integrally Geared Centrifugal Air Compressors (IGCAC) are becoming popular in many industrial applications. Development of such compressors requires in depth Rotordynamic Design and Analysis. To facilitate this, an in-house computer program based on transfer matrix method was developed using MATLAB® software. This computer program is capable of computing rotordynamic parameters such as static deflection, critical speed and interference diagram, and can output critical speed map, mode shape, unbalance response, orbit, for lateral direction.
This software was used to analyze a two stage IGCAC with two impellers on a simply supported rotor running above second critical speed, driven by a two pole induction motor through a step-up gearbox. Undamped critical speed map, an output from the program was used to predict intended bearing stiffness for design. Using the above data and commercially available software DyRoBeS© a suitable bearing was designed. The speed dependent bearing characteristics, an output from DyRoBeS©, were used to determine damped unbalance response plot for a given residual unbalance. Corresponding to a maximum peak in unbalance response the damped critical speed and amplification factors (AF) were found out. The results from the newly developed software were compared with prediction from DyRoBeS©. It was found that critical speed was within 5% and AF was of the same order. Results from in-house software were comparable to that from DyRoBeS©. Based on the guidelines from API 684, the AF and separation margins were determined.
A prototype IGCAC compressor as described above was built and tested. The testing included the collection of steady state, coast-up and coast-down data. Using the coast-up, coast-down data, a Bode plot was created. From this the critical speeds and AF’s were determined and compared with results from in-house software. It was found there was an error of less than 5% for the critical speed and around 5% for AF from the predicted results. For the same compressor a study on the potential excitation frequencies due to unbalance, impeller-diffuser and impeller-scroll tongue interactions were calculated. FFT of the steady state vibration data was deduced. It was found that the calculated frequency and measured frequency at maximum amplitude were aligning. Further noise measurements were recorded based on sound intensity as per guidelines in ISO 9614. The impeller-tongue interaction frequencies for stages were seen in the processed noise data. It was found that the predictions were in good agreement with the test results.