Compressor map

In Figure 4 the compressor characteristics are shown for 5 different, undistorted VIGV settings at the same corrected speed. The range of the different VIGV angles approximately reflects the stagger angle variation of the 50:50-pattern, which is later investigated in further detail. The nominal characteristic is depicted in green. Along all characteristics, the two-dimensional flow field of a single stator passage was acquired at 9 different corrected mass flow rates. In Figure 4a, the relative aa-total pressure ratio πt is shown and in Figure 4b the relative isentropic efficiency ηis. The stage inlet SI is used as reference for all flow quantities and the values are normalized with respect to the range of the compressor map. All characteristics are of the same shape, which emphasizes that the flow behaves similarly for this magnitude of VIGV variation and that no additional non-linear flow phenomena occur. This allows for a passage-wise view of the flow field, which is presented later.

Figure 4.

Compressor map. (a) Total pressure ratio and (b) Isentropic efficiency.

Besides of the undistorted characteristics, the aa-results of 360∘-measurements of various asymmetric VIGV patterns are depicted as squares. For comparison, the green symbols indicate the respective aa-results of the symmetric pattern. At the peak efficiency operating point (PE) the mass flow rate is constant due to the above described procedure of setting the patterns and the total pressure ratio is barely changed by the distorted flow (<1%). The measurements concur with the PCM assumption: The total-to-static pressure rise ps,SE/pt,SI at PE (not shown here) is almost identical for all patterns and the circumferential variation of static exit pressure is negligible. Closer to the stability limit all patterns show a slight decrease in pressure ratio at the near stall operating point representative for all patterns (NSAll), indicating additional losses and a reduced operating range compared to the undistorted case. The minimal mass flow rate for each individual pattern (NSPat) depends on the shape of the distortion and can be higher or even lower than for the nominal case. For all VIGV patterns, the isentropic efficiency of the compressor is significantly decreased compared to the undistorted case. For the 50:50-pattern the reduction is >3% at PE and at NSAll.

If the stage exit SE is used as reference plane for all flow quantities and the inlet pressure is taken from the 5HP at the rotor inlet, then the development of the swirl distortions is excluded from the control volume. However, the results (not shown here) are similar to the ones presented above. The parallel shift of the curves in Figure 4 leads to an almost identical range in exit mass flow for all characteristics. At PE, the distortion has again little influence on the global operating point. The operation of downstream stages is therefore only influenced by the spatial variation of the flow at the stage exit. At NSPat all patterns have a similar mass flow rate, in contrast to the described results with the inlet as reference plane.

Operating range

The observed changes in operating range are systematically investigated at the NSPat operating points of all different patterns. The corresponding measurements are carried out with a coarser angular resolution to save measurement time. However, the trends are consistent with the finer resolved 360∘-measurements that were only performed for selected patterns. Figure 5a shows the relative aa-stage exit static pressure at NSPat for a varying count of mis-staggered adjacent vanes and thus different sizes of the area with increased co-swirl. For distortions of small circumferential extent, the pressure decreases linearly with a gradient that depends on the magnitude of the distortion. The medium and small magnitude distortions feature 50% and 25% deviation in VIGV stagger angle compared to the primarily investigated large distortion of the 50:50-pattern. This linear behavior is in agreement with the results for the critical angle in the case of pressure distortions in Figure 1b (Reid, 1969). The angle can be determined to Θcrit≈85∘. For larger angles, the trend also corresponds with Reid’s results: At NSPat the 50:50-pattern and the 25:75-pattern exhibit a similar static pressure at the same minimum mass flow rate. This is somewhat unexpected, as it contradicts the model concept in Figure 1a. For Θ>Θcrit the PCM suggests a reduced loss in maximum static pressure. Possible reasons for this deviation will be discussed later.

Figure 5.

Influence of swirl distortions on global operating range. (a) Relative static pressure and (b) 0D-Distortion Parameters.

To describe inlet distortions and their influence on the stability limit, typically 0D-parameters are used. For pressure distortions, the DCΘ-coefficient was found to correlate with the loss in static pressure at the stability limit (Hynes and Greitzer, 1987). For the present investigation, the static exit pressure is no ideal descriptor as the minimum mass flow rate also varies. Therefore, the Stability Margin SM (Cumpsty, 2004),

which uses the corrected exit mass flow rate m˙corr,SE, and therefore considers both quantities, is used. Multiple swirl descriptors have been proposed to describe not only the intensity but also the shape of the distortion. For the presented data, a good correlation was found using the descriptor Swirl Intensity SInt from SAE Aerospace (2010). It quantifies the effective magnitude of the flow angle α with its integral over the whole circumference. In Figure 5b, a linear trend between both normalized quantities is observed. The SM of the 50:50- and the 25:75-patterns are similar, which is consistent with the previous observations. However, the respective values of the Swirl Intensity differ, as the definition of SInt neglects the critical angle. Additionally, deviations from the linear trend occur for patterns consisting of multiple segments (gray area). This might be attributed to a combination of the blade response time constant, which quantifies the sensitivity of the compressor blade to inlet distortions (Cousins, 2004), and the duration each blade spends in the distorted regions. For the segmented patterns, the time in the distorted sectors is apparently too small to affect the SM significantly.

Other descriptors as the Swirl Coefficient (Guo and Seddon, 1983) provide similar results. An improvement of the correlation by including the sensitivities to other distortion descriptors as proposed by SAE Aerospace (2010) was not achieved. The DCΘ-coefficient of all patterns has a maximum value of 0.07, which is considered to be low (Pazur and Fottner, 1990). Therefore, no large influence of non-linear coupling between pressure and swirl distortion is expected. This is confirmed by the fact, that no strong correlation between SM and DCΘ is observed.

Rotor Inlet (RI)

The 360∘-flow field downstream of the VIGV is measured with the 5HP. The circumferential variation in aa-total pressure for the single passages is in the order of ±1%. The minimum value is located downstream of the diverging VIGV passage (see marker “A” in Figure 1a). Its magnitude does not change with the count of mis-staggered adjacent vanes. The absolute circumferential flow angle α varies strongly, i.e. ±50% for the 50:50-pattern. For an area of mis-staggered vanes greater than Θ=22.5∘, the flow deflection is fully developed and the difference in α equals the difference in VIGV angle. Therefore, neither the variations in total pressure nor in flow angle can explain the linear trend in Figure 5b or why the 50:50-pattern has the same SM as the 25:75-pattern.

For tip critical rotors, compressor instability originates at the blade tip. The relative circumferential flow angle β and the relative Mach number Marel influence the shock topology. Their passage-wise aa-values can be calculated from the 5HP data. In Figure 6a both quantities are depicted for the same undistorted characteristics as in Figure 4. All near stall values of β are in the same order of magnitude, defining the critical incidence of the airfoil, and Marel varies with the respective VIGV- and throttle setting. The individual passages of the 50:50-pattern at NSAll are shown as well. A large variation in relative flow angle can be observed. As expected, the compressor locally operates at different operating points at different characteristics. However, the local operating points vary more than would be suggested from the VIGV pattern with only two distinct vane angles (compare Figure 1a). This is due to the developing mass flow redistribution. Single passages already operate beyond the critical incidence of the symmetric configurations. The maximum incidence is observed downstream of converging VIGV passage (see red markers “1” in all following figures). The minimum incidence develops downstream of the diverging VIGV passage and leads to operation close to the PE mass flow rate of the nominal configuration (see green markers “2”). In Figure 6b, the circumferential distribution of β at the rotor tip is depicted for 360∘-measurements of different counts of mis-staggered adjacent vanes. The measured flow angles at NSAll are extrapolated to the values at the respective last stable operating points NSPat. This is possible due to the relation between mass flow rate and relative flow angle (not shown here), which is strictly linear over the whole characteristic, and the small differences in mass flow rate. It is evident, that all different patterns locally operate beyond the maximum incidence of the symmetric configuration. While the minimum mass flow rates of the patterns change, that respective portion (see rectangular markers) of the circumference remains similar in size. Its circumferential extent at the stability limit is about 105∘, hence close to the determined critical angle. This is remarkable, as it cannot be derived from literature. It also explains the linear trend in Figure 5b, e.g. that for smaller distorted areas smaller mass flow rates and thus larger static pressure can be achieved until this circumferential extent, hence instability, is reached. However, both the 50:50- and 25:75-pattern, exhibit a distorted area of this size at NSPat at an identical mass flow rate. This is due to the local variation of mass flow at the rotor tip region, which influences the local value of β. An increasing count of mis-staggered vanes leads to a greater circumferential and radial redistribution of mass flow. Only for these two patterns this variation affects the entire circumference. In rotational direction, the mass flow rate continuously decreases downstream of the closed vanes and increases downstream of the opened vanes. This results in identical amplitudes of the local mass flow variation for both patterns and thus to an identical SM.

Figure 6.

Relative flow angle at the Rotor Inlet (RI). (a) Compressor map with aa-Values and (b) Circumferential distribution at the blade tip at NSPat.

Rotor Tip (RT)

To confirm the aa-results from measurement plane RI, the 360∘-data from the axial WPT-array at the rotor tip is used to analyze the developing shock topology. In Figure 7 (right) the ensemble averaged flow field at the circumferential position with minimum incidence is shown for the 50:50-pattern. The shock position (black line) is similar to the undistorted characteristic with the same mass flow rate (dashed green line). The shock is already detached from the blade, its angle γ and distance x to the leading edge are indicated on the left side. For the undistorted characteristics and all passages of the 50:50-pattern, these derived quantities are plotted in Figure 7 (center). A variation of the operating point along and beyond the different characteristics is observed. Its amplitude matches the respective changes in mass flow and local VIGV characteristic, which confirms that the aa-5HP results are valid. The compressor can locally operate beyond the undistorted stability limit if a large enough portion of the annulus remains in stable operation. The passage with maximum incidence exhibits the largest shock distance (see red markers “1”). However, most operation beyond the stability limit is due to comparably small shock angles (see blue markers “3,” compare flow field in Figure 7 (left)). In contrast to Reid’s results this behavior not only occurs for Θ<Θcrit. Consequently, the PCM assumption of the global stability limit, i.e. maximum static pressure for the co-swirl characteristic in Figure 1a, is not valid. This is supported by the observation of mass flow redistribution at measurement plane RI and might explain the constant static pressure for large distorted areas from Figure 5a.

Figure 7.

Static pressure field (outer parts) and shock quantities (center) at the Rotor Tip (RT).

Rotor Exit (RE) and Stage Exit (SE)

At measurement plane SE, data is acquired via a 5HP- and two Kiel rakes. Downstream of the stator, the absolute flow angle α locally varies by only ±1% around its mean value, which is similar to the mean value at the rotor inlet. This is a significant reduction in amplitude compared to the rotor inlet measurement plane (RI) and shows the damping effect of upstream vanes also described by SAE Aerospace (2010). Instead, the variations in total pressure and temperature increase to ≈±3% at RE and SE. In Figure 8 (left) the measured 360∘-flow field of πt,RI−SE for the 50:50-pattern is shown. All values are normalized with the global aa-mean value. The areas of opened and closed vanes are clearly distinguishable. As expected, the aa-pressure ratio per passage is lower for the closed area. However, the local maximum values are also located in this region at the blade tip. In general, larger radial gradients develop downstream of the closed VIGV, which indicates local operation close to the stability limit (Karl et al., 2023). The corresponding operating points for all individual passages are shown in Figure 8 (right). The respective mass flow rate is determined from the 5HP data at the rotor inlet. The matching between local mass flow and pressure at measurement plane SE was achieved via preliminary measurements of the phase difference in pressure and temperature between the two measurement planes. It is evident, that the local operating points follow the shape of the symmetric characteristics and fit the results from the measurement planes at the rotor inlet (RI) and the rotor tip (RT), especially for the passages downstream of the closed VIGV. The drop close to the stability limit in the opened section (see blue markers “3”) points toward additional losses and could explain the reduced pressure ratio of the 50:50-pattern in Figure 4a. It is mainly caused by the flow at the rotor tip, see 2D-field data in Figure 8 (left), and potentially related to the local shock topology: The data from measurement plane RT indicate the smallest shock angles at the respective circumferential position. A reduced efficiency resulting from this pronounced off-design operation is likely.

Figure 8.

Flow field and compressor map per passage at the stage exit (SE).

Due to the attenuation of the swirl variation, the difference between the characteristics of the local operating points is smaller compared to measurement plane RI. However, the variation in mass flow rate is similar to what can be inferred from the WPT data. Even at the stage exit, parts of the circumference are beyond the stability limit of the undistorted compressor. This is mainly the case for the area downstream of the closed VIGV, which explains the large radial gradients in Figure 8 (left).