This article discusses the performance testing of transonic rotors at the Turbopropulsion Laboratory at the Naval Postgraduate School. The Mach number is one of the most important parameters in the case of high-speed compressors. In order to limit power consumption in a test machine, the simplest change is to scale down the machine. A second concept to reduce the power consumption of the machine once it has been scaled down is to throttle the flow before the rotor rather than after it. As a high-speed rotor compresses the incoming air by around 1.4–1.6 times, the air leaving it is appreciably denser than that coming in. If one throttles upstream of the rotor, the exhaust air leaves the machine at atmospheric pressure, which means that the incoming air is below atmospheric pressure. With upstream throttling, care has to be taken to provide long enough ducting ahead of the test compressor to present as uniform as possible flow after the flow rate measuring nozzle.
A driving force behind the improvement found in high-speed compressor fans has been the advent of computer simulations. Of course there is still a need to test these fans and in some cases be able to evaluate the computer codes against accurate experimental data. Here I hope to share some of the challenges and solutions found in the testing these compressor fans.
At the Turbopropulsion laboratory (TPL) at the Naval Postgraduate School (NPS) the research and testing of high-speed compressors also affords students the opportunity to work with these machines. With these types of machines supersonic flows are present and there is no longer the ability to make the assumption of incompressible flow as with most fan test rigs in educational environments. While education is a driving force of NPS the test compressors are used for current research with students completing masters theses using data collected within the laboratory. The TPL test facility focused on here, the transonic compressor rig (TCR) shown in Figure 1 is capable of delivering 337kW (450 hp) at 30 000 rpm. What is remarkable is that this power is absorbed by an 279 mm (11”) rotor weighing around 4.5 kg (10 lbs). Of course when involved in testing of such highly stressed machines, on occasion failure does occur; Figure 2.
In an ideal situation experimental work is best done on full scale devices but obviously cost and practical constraints mean this is not usually possible. Instead one needs to isolate the most influential variables and try to design experiments that capture these. In the case of high-speed compressors the Mach number is one of the most important parameters. In order to limit power consumption in a test machine the simplest change is simply to scale down the machine. If the tip speed of the blades is to be kept the same to ensure Mach number matching with operational compressors the rotational speed needs to be increased proportionally. Typically our test compressors run with tip Mach numbers of approximately 1.5.
A second concept to reduce the power consumption of the machine once you have scaled it down is to throttle the flow before the rotor rather than after it. As a high-speed rotor compresses the incoming air by around 1.4-1.6 times the air leaving it is appreciably denser than that coming in. If one throttles upstream of the rotor the exhaust air leaves the machine at atmospheric pressure which means that the incoming air is below atmospheric pressure. Throttling after the machine means one is sucking in atmospheric air and the mass-flow through the machine is approximately 30% higher with a near proportional increase in the power consumption. In the case of aviation engines this reduction in density is similar to flying at high altitude. For sea level based gas-turbines such as in ships or power stations the flow change from laminar to turbulent over the blade will occurs sooner than at altitude, however these effects can be accounted for empirically. With upstream throttling care has to be taken to provide long enough ducting ahead of the test compressor to present as uniform as possible flow after the flow rate measuring nozzle.
Another challenge is bearing choice and the proper functioning of the associated lubrication systems. Rotating test rigs are by there nature unique devices and so operation of them is a continuous learning process. In the last decade the emergence on affordable ceramic ball bearings has been a major advantage. They allow operation at much higher rotating speeds and operating temperatures. Bearing cooling and lubrication is still needed and this is performed using an oil mist cooling system. In the TPL rig there are four bearing sets and four oil mist coolers with redundancy built in by allowing each oil mist cooler to service two bearing sets (Figure 3). In this way if one oil mister fails no bearing are left without lubrication. A bearing arrangement similar to a milling machine spindle is used with the rear bearing pair held in place by flexure arms that are instrumented to measure the net force forward or rearwards on the shaft while the front bearing pair is allowed to float. The reason for this is that the shaft lengthens during operation due to temperature increase and load and thus one bearing pair needs to float. The axial forward pull of the test rotor is counteracted by a pneumatic balance piston at the rear of the shaft and the force adjusted until the net axial force on the shaft is close to zero. The most amazing aspect of this rig is that it was designed on the late 60's by the late Professor Mike Vavra and the rig is still state-of-the-art today.
There was some concern about the ability of ceramic ball bearings to handle the shock loads of compressor stall and surge. Here the compressor applies large cyclic forces to the rig via the rig as it stall and un-stalls during a surge cycle. During an experiment it is routine to drive the compressor into surge and observe when surge occurs and what is required to recover from a surge. Our experience has shown that the bearings can withstand about a dozen surge cycles.
A further operational consideration is the need to operate with very small tip clearances. A scaled down compressor operating at high-speed tends to grow radially and rubbing between the casing and blade tips is normal. Figure 4 shows the abradable material which allows operation at very small tip gaps without damaging the rotor. This material is baked into a trench above the rotor and needs to abrade without sanding the tips of the rotor. We have settled with a machinable rubberized material and grooved the bottom of the trench to improve the adhesion to the casing.
Of course simply having an operating rig is only half the challenge, once in place useful data has to be taken from the test article. The scaling down of the test article does reduce power but makes the positioning of probes more difficult. Steady-state performance measurement such as pressure ratio and efficiency testing is fairly standard and requires the measuring of the pressure and temperatures upstream and downstream of the compressor. Care has to be taken to distribute the probes evenly around the compressor as clustering them all in one place can results in local flow blockage and cause premature stall. We calculate efficiency from probe measurements since an on the shaft torque transducer is both too costly and complicated at these speeds. To make accurate measurement we use a conservative mass averaged approach using 20 pressure probes and 10 temperature probes (Figure 5).
The valuable data is in trying to see what the flow structure within the flow passages is like. The high-speed data acquisition capability was first pioneered by Emeritus Professor Ray Shreeve and researchers to develop their DPDS (Dual-probe dualsampling) procedure to measure the instantaneous velocity vector field downstream of the rotor in the 1970's. The current TPL high-speed pressure transducers are embedded in the wall of the compressor and are each continuously sampled at 196 kHz. At 100 % speed this results in nearly 20 measurements from blade-to-blade and the continuous nature of our sampling allows stalls to be captured. This is important as the change from un-stalled to stalled operation can be as little as 5-10 revolutions or about 1/100th of a second. Understanding the flow just before stall occurs will hopefully allow for designs that are more resistant to stall and surge while maintaining high efficiency.
Our latest rig improvements aim at making the installation of new rotors simpler with a more modular rig layout. Here the use of computer aided design makes design decisions much simpler as it is possible to ’assemble’ the entire rig virtually before cutting material. Future rig modifications will include a closed loop system for testing process gas compressors and the potential to operate the entire rig at lower pressures to be able to test two-stage compressors while still remaining within our power limitations and of course better simulating high altitude operation.