Bitte wählen Sie ihr Lieferland und ihre Kundengruppe
A highly accurate series of surge test data has been gathered and compared to a state-of-the-art transient simulation. For certain conditions, the code accurately predicted the transient shutdown behavior of a centrifugal compression station without specific tuning of the model. The data quality and detail allowed using the transient software without the uncertainties usually involved in modeling a new compression system. The analysis did under-predict the speed decay rate during shutdown, and the predictions generally got worse using the measured rather than predicted speed decay. Some of this difference was likely due to the simple assumptions (fan law) made in the derivation. Transient codes usually require some assumption about the rate of speed decrease, and this can be a major source of inaccuracy. The analytical model accuracy could also be improved if steady-state test data are gathered in order to refine the pressure drops and pipe friction factors throughout the domain. Further refinement could be achieved if actual valve opening times in the installed operating condition could be obtained. Good agreement was obtained between measured and simulated cases for low suction pressure testing without recycle valve modification. Poorer agreement was observed for higher suction pressures and for tests that modified the recycle valve characteristics. TFS did not predict test cases well that passed through a region close to the measured surge line or crossed it. These tests often involved rapidly changing flow rates, which were damped out in the corresponding simulation. Further tests are recommended to evaluate conclusively how effective TFS is at predicting rapidly changing flow characteristics, as are often present during surge or near-surge transient events. Practically speaking, well-designed surge control systems should not cross the actual surge line, which would improve the predictions. Successful modification of recycle valve characteristics resulted in equivalent valve sizes of 1.2 inches to full scale (6 inches) with a variety of travel times. The anti-surge system was thus modified to permit testing near the surge line, including one test that entered full surge. It was found that the recycle valve was significantly over-designed for the conditions tested. Simulated results were sensitive to the rpm decay data used. A small difference in rpm during the first few seconds of the shutdown event produced drastically different results. An rpm decay curve was calculated, and a second rpm decay curve, based upon actual rpm decay data from the experiment, was also used. Contrary to expectations, the actual rpm decay curve produced worse results than the calculated curve. The cause of this unexpected result could be a subject for further investigation.