2.1 Test vehicle

The Larzac 04 C5 was used as test vehicle (cf. Fig. 1). This engine is the main scientific test engine at this test bed and the Institute of Jet Propulsion has extensive experience with this test vehicle’s behaviour under adverse inflow conditions, as well as under compressor stall. And it is equipped with a highly instrumented low-pressure compressor (LPC) and is thus well-suited for investigations of intake-compressor interactions. The most important engine performance data at ISA conditions are listed in Table 1.

Table 1. Engine performance data at ISA conditions

Figure 1. Schematic overview of the experimental setup and instrumentation of the MEIRD; 3D-CAD photo of the MEIRD from an angle showing the bottom of the intake; boxes with engine station numbers.

2.2 Investigated intake and instrumentation

At the Institute of Jet Propulsion, the MEIRD (Military Engine Intake Research Duct) has been designed and manufactured to generate a combined pressure and swirl distortion inside an engine intake and to investigate the flow interaction between duct and compressor both experimentally and numerically [Reference Rademakers, Haug, Niehuis and Stößel19]. One of the design restrictions was that provoked pressure and swirl distortions at the duct outlet plane (DOP) do not exceed the operating range of the Larzac 04 test vehicle. Figure 1 depicts the entire experimental setup. The traversable measuring rake housing is designed to be as compact as possible such that the DOP and the AIP (aerodynamic interface plane) merge. The axial length of the duct is set to $3 \cdot {D_{DOP}}$ (diameter of the duct outlet plane), as this seems to be a reasonable length of a realistic short highly bent intake duct. The fan entry diameter is $454{\rm{mm}}$ ( $D = 454{\rm{mm}}$ ) and the distance between DOP and fan face is $0.8 \cdot D$ . To achieve the intended strong flow distortion at the AIP the overall inclination of the centreline features a strong curvature. Furthermore, the MEIRD is designed to prevent a direct line of sight on the DOP from the DIP (duct inlet plane), which is an important requirement for low observability military engine systems, in order to reduce the exposure of the highly reflective fan to radar beams. In Fig. 1 information about the basic geometric parameters, as well as the centreline (red) and the positions of the geometry defining cross-sections (CS) are depicted (see also [Reference Haug, Rademakers, Stößel and Niehuis20]).

The cross-sections are aligned along the centreline (S) in their centres of area. The shape of the cross-sections changes from kidney-shaped ( $S = 0{\rm{\% }}$ ) at the DIP to rectangular ( $S = 75{\rm{\% }}$ ) to round at the DOP. The different cross-sectional shapes are shown in Fig. 2. In Fig. 2 additionally in CS8 the extent of the blowing slot and the diameter of the DOP is displayed. In the individual cross-sections, as well as in the symmetry plane of the MEIRD, there are static wall pressure probes installed to generate data for flow characterisation, which are analysed in detail in Section 4. Cross-sections four and six are not depicted in Fig. 2, since the static pressure holes are only placed on the upper wall of the duct. Furthermore, a measuring rake was installed at the DOP/AIP, which consists of 10 kiel probes and 11 five-hole pressure probes, which are equidistantly distributed along the full span of the rake. Recording the total pressure at the AIP the rake was traversed in either ${6^ \circ }$ steps or ${60^ \circ }$ steps in circumferential direction. The measuring rake was only installed during the stationary operation of the engine, but not during the recording of the intake-compressor map. This was mainly due to prevention of damage to the traversing system by compressor surge events.

Figure 2. Main cross-sections of the MEIRD and sketch of the extend of the blowing slot in CS8.

The CFD investigations by Haug et al. [Reference Haug, Rademakers, Stößel and Niehuis20] did already show that the MEIRD features a separation area located at its second bend (cf. Fig. 1 area ‘A’). To investigate this area of flow separation in special detail and to provide the ability to influence it with both active and passive measures, the MEIRD has three exchangeable panels in this region. While panel 1 is placed upstream of the detachment bubble, panel 2 is positioned exactly on it and panel 3 is partly still located at the detachment area and partly downstream of the flow detachment. With the aim of suppressing the separation area and improving the overall performance of the system three different configurations were investigated and compared in this study:

1. - $Baseline$ : MEIRD without flow stabilising measures (Reference case).

2. - $VG$ : Only panel 1 is replaced by a panel with vortex generators (cf. Fig. 3).

3. - $Blo$ : Panel 1 and 3 as in the Baseline case. Only panel 2 is replaced by a panel with an integrated Coanda nozzle for active flow stabilisation by injection (cf. Fig. 3).

Figure 3. Vortex generator configuration (top view) with geometry parameters; blowing configuration (cutaway view) with Coanda nozzle.

Both the VG configuration and the Blo configuration have been extensively investigated specifically for the MEIRD in preliminary tests either numerically in a DOE (design of experiment) study [Reference Kächele, Rademakers, Stößel and Niehuis12] or experimentally in a scaled setup [Reference Max, Stößel, Krummenauer and Reinhard21]. This means that they are technically well-developed variants that have already achieved promising results in previous testing and are not first-time developments.

The vortex generator panel with the most relevant geometric parameters, as well as a cutaway view of the blowing configuration panel, is shown in Fig. 3.

The vortex generator height ( ${h_{VG}}$ ) in relation to the reference boundary layer thickness ${\delta _{ref}}$ Footnote ¹ is 1.3. The VG chord length ( $c$ ) referred to reference boundary layer thickness is 9.3. And the y-distance over the reference boundary layer thickness is 2.8. When the blowing configuration was applied to the MEIRD, the blowing mass flow rate has been increased from 1% in 0.5% steps to 2%. The injection mass flow rate was always related to the respective intake inlet mass flow rate. During the blowing experiments the injection mass flow was provided by an external compressor. The extent of the blowing slot in y-direction was kept constant at $0.85 \cdot D$ (cf. Fig. 2 CS8).

3.1 Distortion descriptors

The first parameter described is the pressure recovery which gives an overall impression on the performance of an intake system.

(3) \begin{align}PR = \left( {1 - \frac{{{P_{t,AV - DOP}}}}{{{P_{t,amb}}}}} \right) \cdot 100{\rm{\% }} = \left( {1 - \frac{{{P_{t,FAV}}}}{{{P_{t,amb}}}}} \right) \cdot 100{\rm{\% }}\end{align}

A further well-known distortion coefficient is the DC60 parameter:

(4) \begin{align}DC60 = max_{{{60}^ \circ }}\left( {\frac{{{P_{t,FAV}} - {P_{t,Worst - {{60}^ \circ }}}}}{{{P_{t,FAV}} - {p_{AV}}}}} \right),\end{align}

where ${P_{t,FAV}}$ is the face average total pressure of the AIP. ${P_{t,Worst - {{60}^ \circ }}}$ is the mean of the measurement points in a sector of ${60^ \circ }$ with the lowest total pressure values. And ${p_{AV}}$ is the average static pressure in the AIP. The DC60 distortion coefficient was first defined by Reid [Reference Reid23].

The circumferential distortion intensity (CDI) at the ${k^{{\rm{th}}}}$ radial location (“ring”) is defined as [22]

(5) \begin{align}CD{I_k} = max_{k} \left( {\frac{{{P_{t,AV}} - {P_{t,AV - LOW}}}}{{{P_{t,AV}}}}} \right),\end{align}

where ${P_{t,AV}}$ is the ring averaged total pressure and ${P_{t,AV - LOW}}$ is the subset of only those values that are lower than ${P_{t,AV}}$ ( ${P_{t,AV - LOW}} \lt {P_{t,AV}}$ ). The maximum of k rings indicates the highest circumferential distortion intensity.

3.2 Engine parameters

Several parameters are introduced to evaluate the performance and behaviour under distorted inflow of the test vehicle. The reduced engine mass flow is given as [24]:

(6) \begin{align}{\dot m_{red}} = \dot m \cdot \frac{{{p_{ISA}}}}{{{p_{t1}}}} \cdot \sqrt {\frac{{{T_{t1}}}}{{{T_{ISA}}}}} .\end{align}

The reduced spool speed of the low-pressure compressor [24]:

(7) \begin{align}{n_{red}} = n \cdot \sqrt {\frac{{{T_{ISA}}}}{{{T_{t1}}}}} .\end{align}

The intake-compressor pressure ratio

(8) \begin{align}{\rm{\Delta }}{p_{t\left( {1,21} \right)}} = \frac{{{p_{t,21}}}}{{{p_{t1}}}}.\end{align}

The intake-compressor pressure ratio ${\rm{\Delta }}{p_{t\left( {1,21} \right)}}$ is defined as the ratio between the LPC core outlet pressure ${p_{t,21}}$ and the intake inlet total pressure ${p_{t1}}$ . This means that the intake-compressor system is considered as one single unit.

The definition of the surge margin is [Reference Dudgeon25]:

(9) \begin{align}SM = \frac{{{{\rm{\Pi }}_{t,SM}} \cdot {{\dot m}_{OP}}}}{{{{\rm{\Pi }}_{t,OP}} \cdot {{\dot m}_{SM}}}} - 1.\end{align}

Where ${\dot m_{OP}}$ and ${\dot m_{SM}}$ are reduced mass flow rates. The surge margin definition is used for constant reduced spool speed.

3.3 Swirl angles

According to the SAE Standard 5686 [26] swirl can be grouped into four categories: (1) bulk swirl, (2) tightly wound vortices, (3) paired swirl and (4) cross-flow swirl. Examples on the generation of each type of swirl and how the swirl is formed is also given in the SAE standard. The characteristics of each type of swirl will be illustrated in terms of the swirl angle (cf. Fig. 4). The swirl angle $\alpha $ is defined at a point as the circumferential angle of flow from the axial direction as calculated in Equation (10) and shown in Fig. 4.

(10) \begin{align}\alpha = tan^{- 1}\left( {{v_\alpha }/{v_x}} \right)\end{align}

4.1 Static pressure measurements along intake centreline (x-z plane)

First, the intake without flow stabilising measures was investigated. This baseline case serves on the one hand for the flow characterisation of the intake and on the other hand as a reference to evaluate and assess the efficiency of the active and passive flow control measures. For this reason, the static wall pressures at the upper wall centreline (top, continuous lines) and lower wall centreline (bot, dashed lines) in the symmetry plane (grey, MEIRD contour) of the intake were recorded (cf. Fig. 5). The graphs shown in Fig. 5 were evaluated at a constant engine speed of ${n_{red,rel}} = 76{\rm{\% }}$ . Furthermore, the injection mass flow rate for the Blo configuration was set to 1.5%. This turns out to be the optimal injection mass flow rate from a performance point of view, as will be explained in more detail later in Section 4.4. The results and conclusions given at ${n_{red,rel}} = 76{\rm{\% }}$ can generally be applied to the speed range between ${n_{red,rel}} = 54{\rm{\% }}$ and ${n_{red,rel}} = 86{\rm{\% }}$ . Figure 5 further shows the relative x-position of the detachment bubble (A), as well as the ${x_{rel}}$ position of the individual panels. The relative x-axis is referred to the total length L of the intake (cf. Fig. 1). Additionally, the position of the injection (red arrow) is depicted.

Figure 5. Static wall pressure at the zx-symmetry plane at ${n_{red,rel}} = 76{\rm{\% }}$ , ${\dot m_{Blo}} = 1.5{\rm{\% }}$ .

The static pressure curves at the upper wall of the MEIRD are largely identical. In the area between ${x_{rel}} = 0.1$ and ${x_{rel}} = 0.7$ , the static wall pressure of the VG configuration is lower than that of the Baseline and Blo configurations. At ${x_{rel}} = 0.95$ , thus in the region of the DOP, the Blo configuration possesses the highest static pressure, followed by the VG configuration. The Blo and the VG configuration achieve higher static pressure levels at the DOP than at the DIP. While the static pressure of the baseline configuration is lower at the DOP than at the DIP.

The lower wall static pressure curves show up almost identical pressure values between ${x_{rel}} = 0.1$ and ${x_{rel}} = 0.68$ . Between ${x_{rel}} = 0.76$ and ${x_{rel}} = 0.86$ , the baseline configuration shows an apparent pressure plateau on the lower wall, which can be attributed to the separation area (A) [Reference Haug, Rademakers, Stößel and Niehuis20]. At the VG configuration as well as for the Blo configuration, a pressure plateau is no longer present. Looking at the Blo and VG configuration the static wall pressure increases continuously from ${x_{rel}} = 0.73$ up to the DOP. The baseline inlet separates, whilst the VG and Blo configuration still maintain diffusion to the DOP.

The VG and the Blo configuration reach a higher static pressure at the DOP than at the DIP. Based on the analysis of the static wall pressure characteristics, it is evident that both flow-stabilising measures effectively suppress the separation bubble. According to the static pressures at the AIP the Blo configuration features a higher static pressure level than the VG-configuration and therefore the LPC is exposed to a higher static pressure level and its performance is better (cf. Fig. 12). This is due to the higher and more uniform static and total pressure distribution in front of the LPC. The pressure loss through the intake is consequently lowest using the Blo configuration (compare total pressure in the AIP at Fig. 9(c) and (a)).

4.2 Static pressure variation along x-y plane

The wall pressure data along the centreline ( ${y_{rel}} = 0$ ) provide initial knowledge that the flow stabilisation measures are effectively suppressing the detachment bubble. However, the data refer exclusively to the z-x plane and do not provide any information about the x-y plane. Therefore additional static pressure holes were positioned along the x-y plane of Panel 2 and 3 to investigate the 3D effects of the VG and Blo configuration, shown in Fig. 6. The panels are attached to the aluminium frame integrated into the MEIRD (cf. Fig. 1). The transition between the panels and the MEIRD is aerodynamically smooth. However, due to the aluminium frame, measuring points in the junction between the individual panels are missing, which leads to a static pressure step in the wall pressure distribution displayed in Fig. 6.

Figure 6. Static wall pressure on panel 2 and 3 for the different investigated configurations at ${n_{red,rel}} = 76{\rm{\% }}$ , ${\dot m_{Blo}} = 1.5{\rm{\% }}$ .

Figure 7. Static wall pressure on the different cross-sections at ${n_{red,rel}} = 76{\rm{\% }}$ , ${\dot m_{Blo}} = 1.5{\rm{\% }}$ .

Figure 8. Static wall pressure on the different cross-sections at ${n_{red,rel}} = 76{\rm{\% }}$ , ${\dot m_{Blo}} = 1.5{\rm{\% }}$ .

Figure 6 shows the results of the individual configurations. ${Y_{rel}}$ is related to the intake outlet diameter and the engine speed was kept constant at ${n_{red,rel}} = 76{\rm{\% }}$ . The blowing mass flow rate was 1.5%. Looking at the baseline configuration, the detachment line is clearly visible on panel 2 in a ‘C’ shape. Between ${x_{rel}} = 0.76$ and ${x_{rel}} = 0.78$ the flow detaches. The complete panel 3 is detached ( ${p_{rel}} = 0.82$ cf. Fig. 5 pressure plateau (A)).

The VG configuration shows its impact especially on P2 in the way of a static pressure drop which accelerates the flow. However, the flow is accelerated much stronger at ${y_{rel}} = 0$ than in the outer regions at ${y_{rel}} = \pm 0.35$ . On P3, the static wall pressure continues to increase continuously and homogeneously.

The Blo configuration depicts a strong acceleration on P2 which is clearly visible in the static pressure drop. At ${y_{rel}} = \pm 0.37$ an area with higher static pressure can be observed. In these two areas, the Coanda nozzle was fixed by two struts to prevent the injection slot from widening during the tests. Due to these struts, no injection was possible in this area. P3 shows a massive continuous pressure increase for the injection configuration.

The analysis of the surface plots confirms the statements already made regarding the wall pressure curves in the symmetry plane. The further pressure increase on P3 is more pronounced in the Blo configuration than in the VG configuration ( ${p_{rel,p3,{max},Blo}} = 0.97$ ; ${p_{rel,p3,{max},VG}} = 0.92$ ).

4.3 Static pressure variation around wall

To further understand the static pressure rises and drops within the MEIRD, the static wall pressure curves in the individual cross-sections were evaluated. From cross-section 1 to 7, the baseline configuration and the Blo configuration are almost identical. At CS5, at y/d = 0.8 the blowing configuration shows a slightly higher static pressure than the baseline configuration does.

It is to note, that the VG configuration in cross sections 1-6 has an overall lower static pressure level than the other two configurations. This effect was already observed in Fig. 5 when analysing the centreline pressure curves. The vortex generators provide a slight acceleration of the flow in CS 1-6. In CS7 just before the detachment bubble, all three configurations are almost identical. In CS8, the baseline configuration shows the familiar pressure plateau of the detachment bubble. The pressure curve of the blowing configuration could not be recorded completely due to the Coanda nozzle slot placed in this area. In CS9, the Blo configuration shows a significant acceleration on the lower wall area, while reaching already a higher static pressure level than the other two configurations at the upper wall of the MEIRD. The VG configuration shows an asymmetry at the bottom in CS9, which can also be observed in the analyses of the AIP plots. As expected, the Blo configuration shows the highest static pressure recovery in CS10.

4.4 LPC inlet flow field

The results of the total pressure measured with the Kiel probes of the measuring rake are displayed in Fig. 9. Thereby, an immense total pressure distortion at the ${180^ \circ }$ position occurs in the baseline configuration (Fig. 9(a)). This total pressure distortion is caused by the detachment bubble. Furthermore, two total pressure disturbances are slightly visible at the ${60^ \circ }$ and ${300^ \circ }$ positions. These disturbances are caused by the vortex system which arises in the first bend of the MEIRD [Reference Haug, Rademakers, Stößel and Niehuis20]. The distortion parameters for the baseline configuration and the other configurations are listed in Table 2. At an engine speed of ${n_{red,rel}} = 76{\rm{\% }}$ , the DC60 parameter of the baseline configuration is 0.9, which is too high for most engine manufacturers. An injection mass flow rate of 1%, reduces the extent of the total pressure disturbance (cf. Fig. 9(b)). However, the dominant core of the distortion at ${180^ \circ }$ is still present. Nevertheless, the DC60 is reduced to 0.3. As only the lower wall of the MEIRD is equipped with flow stabilising measures the pair of vortices at ${60^ \circ }$ and ${300^ \circ }$ are prominent in all configurations. With an injection mass flow of 1.5%, the total pressure disturbance at ${180^ \circ }$ is completely suppressed (Fig. 9(c)). A further increase of the mass flow to 2% causes no improvement (Fig. 9(d)). In contrast, at 2% injection mass flow, ${p_{t,rel}}$ is above one at the ${180^ \circ }$ position, which again can be considered as a distortion onto the compressor system. Due to the limited time at which the injection mass flow of the external compressor is available, the blowing AIP plots are very coarsely resolved. The buffered injection air from the external compressor is not sufficient to finely traverse the measuring rake including sufficient measuring time. Therefore, the corresponding distortion parameters are also shown in brackets in Table 2. Nevertheless, a qualitative statement can be made. The total pressure plot of the VG configuration is shown in Fig. 9(e)). The total pressure disturbance at ${180^ \circ }$ is weakened and stretched in width. However, the total pressure disturbance is best suppressed in the injection configuration with 1.5% blowing mass flow.

Figure 9. Total pressure within the AIP at ${n_{red,rel}} = 76{\rm{\% }}$ , (a) baseline configuration, (b) Blo configuration with ${\dot m_{Blo}} = 1{\rm{\% }}$ , (c) Blo configuration with ${\dot m_{Blo}} = 1.5{\rm{\% }}$ , (d) Blo configuration with ${\dot m_{Blo}} = 2{\rm{\% }}$ , (e) vortex generator (VG) configuration.

In addition to the Kiel probes, the five-hole pressure probes in the measuring rake have also been evaluated [Reference Bohn and Simon27]. The results in circumferential direction ( $\alpha $ -angle) are shown in Fig. 10.

Table 2. Distortion coefficients and surge margin

Figure 10. Swirl in circumferential direction in the AIP at ${n_{red,rel}} = 76{\rm{\% }}$ ; (a) baseline configuration, (b) VG configuration, (c) Blo configuration with ${\dot m_{Blo}} = 1.5{\rm{\% }}$ (positive rotation: counterclockwise direction).

A positive $\alpha $ -angle indicates a counterclockwise flow and vice versa. Figure 10(a) depicts the $\alpha $ -angle in the baseline case. The two opposite flow directions in the right and left halves of the AIP are evident. The maximum flow angles reach $ \pm {30^ \circ }$ . At the ${180^ \circ }$ position the two vortices meet. At ${200^ \circ }$ and respectively at ${175^ \circ }$ two additional pairs of vortices running in opposite directions occur. The flow stabilising measures in the MEIRD reduce the flow angles (cf. Fig. 10(b), (c)). In the case of the VG configuration the maximal flow angles are reduced to $ \pm {15^ \circ }$ (Fig. 10(c)). During injection, the flow angles are further reduced and the circumferential swirl in the AIP decreases (Fig. 10(b)).

Considering the radial flow direction ( $\gamma $ -angle cf. Fig. 11), a positive $\gamma $ -angle means a flow towards the centre of the AIP.

Figure 11. Swirl in radial direction in the AIP at ${n_{red,rel}} = 76{\rm{\% }}$ ; (a) baseline configuration, (b) VG configuration, (c) Blo configuration with ${\dot m_{Blo}} = 1.5{\rm{\% }}$ (positive: to the centre point).

Figure 12. Intake-compressor map for the different investigated configurations ( ${\dot m_{Blo}} = 1.5{\rm{\% }}$ ).

In all investigated configurations, it can be seen that the flow in the upper half of the AIP moves outwards and in the lower half towards the centre. With installed flow-stabilising measures, the maximal flow angles in the radial direction are reduced. To sum up, it might be possible to derive the flow direction within the AIP from the Kiel total pressure measurements data. The fluid moves in the circumferential direction from the higher pressure level to the lower pressure level. This physical aspect is confirmed by the results of the five-hole pressure probes. The effects of the duct onto the fan and the mitigation of the flow separation due to the investigated measures can be summarised in the following steps:

1. 1. When coupled to the baseline inlet, the fan ingests low total pressure at ${180^ \circ }.$

2. 2. Fan responds by reducing inlet static pressure at ${180^ \circ }$ (Fig. 5 at $x/L = 1.0$ ).

3. 3. Static pressure gradients drive flow from upper to lower half of intake.

4. 4. Flow migration generates swirl and radial flow distortion (Figs 10(a) and 11(a)).

5. 5. Vortex generator and injector methods reduce total pressure distortion, which in turn weakens the fan response and top-to-bottom static pressure gradients [Reference Wakelam, Hynes, Hodson, Evans and Chanez4, Reference Carnevale, Wang, Green and Di Mare5].

6. 6. This reduces inlet swirl and radial flow distortion entering the fan.

7. 7. Fan performance rises.

4.5 Intake-compressor map

To assess the effects of the intake disturbances on the compressor system, the intake-low-pressure compressor map has been recorded (Fig. 12). From a reduced relative speed of 54% up to the maximum speed of around 86%, the flow-stabilising measures expand the map and increase the safe operating range compared to the baseline case. This increases the surge margin as listed in Table 2. Thereby, the injection configuration features the largest surge margin from a reduced relative speed of 62% up to the maximal measured spool speed (cf. Fig. 12). Table 2 demonstrates that a reduction of the distortion parameters increases the surge margin. With a DC60 value of 0.05, the Blo configuration exhibits a surge margin of 18.49 at ${n_{red,rel}} = 76{\rm{\% }}$ . Corresponding to an increase of 48% in the surge margin compared to the baseline configuration. For the same operating point, the VG configuration improves the surge margin by 10% compared to the baseline configuration. Both flow stabilisation measures, reduce the PR and the CDI parameters significantly. Whereby the Blo configuration achieves even lower distortion parameter values compared to the VG configuration. For instance, the DC60 parameter at ${n_{red,rel}} = 76{\rm{\% }}$ is reduced from 0.9 at the baseline configuration to 0.35 with the VG configuration to 0.05 using the Blo configuration.