Experimental Quantification of Fan Rotor Effects on Inlet Swirl Using Swirl Distortion Descriptors

Commercial aviation relies on continued technological advancements to safely and efficiently transport passengers and cargo while limiting risk and environmental impact. Until recently, this has been accomplished through incremental gains in subsystem (airframe, propulsion, electrical, controls, etc.) design, performance, and efficiency; however, in 2009, NASA initiated the environmentally responsible aviation project. With aggressive goals for noise (42–71 dB), fuel burn (50–60%), and emissions (70–80%) reductions compared to the current state of the art, the environmentally responsible aviation project has encouraged the commercial aviation industry to rethink subsystem and component level design by focusing on systems level improvements with highly integrated next generation aircraft concepts [1,2].

A leading concept design for a highly integrated flight vehicle that promises to dramatically reduce drag, weight, fuel burn, emissions, and noise is the blended or hybrid wing body aircraft (BWB or HWB, respectively). This architecture combines wing and fuselage components into a smoothly transitioning wing body planform which improves lift-to-drag ratio and requires less thrust compared to equivalent conventional tube and wing architectures [1,3,4]. Additionally, by mounting optimized engine inlets near the lifting surface or embedding the engines within the fuselage, the concept of boundary layer ingestion (BLI) can be exploited to further improve aircraft efficiency. BLI has been shown to reduce fuel burn by more than 16% when compared to a conventionally powered aircraft. Both BLI engine mounting options (podded above wing or embedded subsurface) offer noise reductions by shielding the ground with the airframe and exhausting above or in front of the aircraft trailing edge [5–11].

While the desirable reductions in fuel burn and noise are attractive, BLI poses considerable challenges for conventional turbofan engines that are designed for clean, uniform inlet flow conditions. The low energy boundary layer flow that is ingested by the engines in a BLI configuration, along with vortices generated from the leading edge and body of the airframe, creates severe nonuniformities in flow direction and produces inlet swirl distortions propagating to the fan rotor. Left unabated, this nonuniform inlet flow profile can cause significant propulsive efficiency reductions, induce high aeromechanical stresses, and reduce the overall performance of the engine [7,8,12–16].

Due to the importance of efficiency, survivability, and performance of the engine operating in a nonuniform inlet flow condition, the SAE International S-16 Turbine Engine Inlet Flow Distortion Committee has developed a set of guidelines (AIR5686) to measure and quantify characteristics of inlet swirl distortions. These guidelines result from collaborations between engine designers and inlet designers and are intended to define acceptable limits on the extent and intensity of nonuniform flow entering the turbofan engine.

The committee has classified swirl according to the four categories of bulk swirl, tightly wound vortices, paired swirl, and cross-flow swirl. While multiple types of swirl can exist in a given flow profile, features can be characteristically generalized into these four categories [17]:

• Bulk swirl is a large-scale, global fluid rotation in a single direction; can be either corotating or counter-rotating with fan rotation direction.

• Tightly wound vortices are small-scale, locally intense flow rotation regions where the fluid experiences continuously and rapidly changing direction.

• Paired swirl are regions of well-developed, stable contra-rotating vortices.

• Cross-flow swirl is induced swirl resulting from flow redistribution associated with cross wind inlet conditions.

The S-16 committee has further classified swirl distortion parameters into numerical values of sector swirl (SS), swirl intensity (SI), swirl directivity (SD), and swirl pairs (SP) [17]:

• Sector swirl is a measure of the average positive (fanwise) and average negative (antifanwise) swirl angles along a constant radius measurement ring.

• Swirl intensity is a measure of the average of the unsigned swirl angle along a constant radius measurement ring.

• Swirl directivity classifies the average fluid direction as positive (fanwise) or negative (antifanwise) along a constant radius measurement ring.

• Swirl pairs indicate the number of fluid direction changes along a constant radius measurement ring.

Although these swirl descriptors are classically calculated at the aerodynamic interface plane (AIP), the innovative data analysis technique outlined in the present paper leverages the values at measurement planes surrounding the fan rotor to demonstrate the capability of the component to process and recover from an inlet swirl distortion condition. The Experimental Methods section describe the experimental methodology of utilizing a StreamVane™ swirl distortion device to generate a tailored flow profile that simulates inlet conditions found on a conceptual next generation aircraft, including the procedures utilized to collect and process experimental data. The resulting flow angle profiles, secondary flow streamlines, and swirl distortion descriptors are then presented and discussed in detail, followed by a summary of important findings and experimental outcomes.

Inlet swirl distortion ground tests were conducted in the Turbomachinery and Propulsion Research Laboratory at Virginia Tech. The research platform (Fig. 1) is powered by a Pratt & Whitney Canada JT15D-1 turbofan engine and consists of the engine, an inlet duct, a flow distortion generator, an engine health monitoring suite, and flow diagnostic instrumentation. The P&WC JT15D-1 turbofan engine is well suited for distorted inlet flow research due to the presence of a midspan shroud and a part-span stiffener on the fan blades. Both features stiffen the blade and produce an aeromechanically robust fan design for dynamic blade loading.

Air at atmospheric conditions entered a mass flow rate calibrated bellmouth inlet at Station 0.0 and interacted with a StreamVane distortion device midway along the inlet duct, Station 0.5. For this investigation, the StreamVane inlet plane was located approximately two fan diameters upstream of the fan rotor inlet plane (one fan diameter upstream of the established AIP). The StreamVane is a distortion device that utilizes a custom pattern of flow turning vanes to manipulate a uniform inlet flow into a desired nonuniform inlet swirl distortion profile. This nonuniform flow developed in a straight connecting duct before being ingested by the turbofan. The StreamVane method has been previously proven to accurately generate specific swirl distortion for both wind tunnel [18] and turbofan engine [19–21] testing. To mitigate the risk of stalling/surging the fan compression system under the distorted flow conditions, the fan speed was limited to 65% design maximum, corresponding to an inlet mass flow rate of approximately 45 lbm/s (20 kg/s) and fan inlet Mach number of approximately 0.3.

Two measurement planes surrounded the fan rotor for analysis. The fan inlet plane (Station 1.0) and the fan outlet plane (Station 1.5) were located within one-half of the tip chord length (approximately three inches) of the leading and trailing edges of the fan rotor blades, respectively. At each plane, a five-hole three-dimensional (3D) flow probe was mounted to a radial traverse allowing the sensing area of the probe to plunge into the flow stream and collect flow data along fifteen inlet radial measurement rings and thirteen outlet radial measurement rings. The radial measurement locations span from fan rotor hub to fan rotor tip in the inlet measurement plane; however, due to engine geometry constraints, the two innermost radial measurements could not be repeated in the outlet measurement plane.

The StreamVane swirl distortion device used in this experiment was modeled according to realistic inlet flow patterns generated by computational fluid dynamics models (Fig. 2) based on boundary layer ingestion and airframe wake vortex shedding of a conceptual blended wing body aircraft. This inlet flow profile was considered a worst-case scenario associated with aggressive aircraft maneuvers. It included two primary swirl categories—a large-scale, anticlockwise (fanwise) bulk swirl and a small-scale, clockwise (antifanwise) tightly wound vortex. When viewed forward looking aft, the core of the tightly wound vortex was located at 90 deg fanwise from top-dead-center at approximately 80% tip radius from centerline.

Once designed, the intricate geometry of the engine-scale (21 in (53.34 cm) diameter) StreamVane was fabricated as a single part using additive manufacturing (Fig. 3). The StreamVane was installed in a rotating fixture within the inlet duct of the turbofan engine research platform. The rotating fixture allowed the StreamVane to incrementally rotate to twenty-four circumferential locations, while the radially traversing mount allowed the five-hole three-dimensional flow probe to incrementally plunge to each radial location. A coordinated motion of the two devices enabled full annular measurement coverage at each measurement plane while requiring only one probe penetration through the fan case at top-dead-center, preserving the structural integrity of the component. The complete experimental test matrix is summarized in Table 1.

Collected five-hole three-dimensional flow probe pressures were processed and correlated to discrete radial flow angle, tangential flow angle, total pressure, and static pressure results through the use of calibration curves defined in Ref. [22]. The resulting data allow for further calculations of three-dimensional fluid velocity components as well as swirl descriptors defined in Ref. [17]. Uniform (nondistorted) flow conditions were measured and removed from the distorted data set. This technique of background subtraction eliminates probe installation errors at both measurement planes as the probe remains in place during StreamVane installation and removal. This technique also removes the overwhelming fanwise flow direction at the fan rotor outlet measurement plane; therefore, only the relative effects of the swirl distortion remain.

Data from five test replications were averaged, resulting in annular radial flow angle profiles, tangential flow angle profiles, and in-plane secondary flow streamline profiles at the two measurement planes. The results are presented forward looking aft according to the illustration in Fig. 4, where circumferential angles begin at top-dead-center and progress in the fanwise direction (anticlockwise). All discussion points utilize this reference frame as well as the station schematic in Fig. 1. The StreamVane design computational distortion profile (Station 0.5), the fan inlet experimental results (Station 1.0), and the fan outlet experimental results (Station 1.5) are presented left-to-right in each of the following figures.

Comparisons of radial flow angles (Fig. 5) and tangential flow angles (Fig. 6) at the three data planes show approximately 50% reductions in absolute magnitudes from Station 0.5 to Station 1.0. This result is due to viscous dissipation, bulk swirl convection, and natural distortion development in the connecting duct between the StreamVane exit plane and the fan inlet measurement plane. Further reductions of approximately 30–40% in absolute flow angle magnitudes are apparent at Station 1.5. This result indicates partial recovery of nominal flow conditions through the fan rotor. While the fan was found to reduce the distortion magnitude, complete recovery was never achieved. Consequently, continued off-design conditions propagate to downstream engine components.

In-plane, secondary flow streamline plots (Fig. 7) derived from the three data planes clearly indicate that the tightly wound vortex feature convects fanwise from plane to plane but passes nearly unchanged through the fan rotor. The extent of the distortion feature also appears to have remained constant from Station 0.5 to Station 1.0 before being slightly reduced at Station 1.5. The bulk swirl region of the distortion appears to dissipate considerably through the fan rotor; however, significant swirl distortion is found in the outermost measurement radii.

Swirl descriptors were calculated at the three data planes according to Eqs. (1)–(5). Because the experimental data planes were not located at the AIP, the analysis technique employed at Station 1.0 and Station 1.5 deviated from the classical methods outlined in Ref. [17]; however, this novel application clearly demonstrates the development of distortion parameters throughout the investigation domain. Furthermore, the calculated values assist in quantifying the ability of the fan rotor to process and recover from the inlet swirl distortion.

Figures 8–11 present the resulting sector swirl, swirl intensity, swirl directivity, and swirl pairs distortion descriptors, respectively. The plotted values correspond to each radial ring of data in accordance with Eqs. (1)–(5). Due to the relatively benign distortion levels near the centerline at Station 0.5, the descriptors are only provided at the experimental radial measurement locations permitted by engine geometry constraints at Station 1.0 and Station 1.5. Additionally, engine core and bypass regions are clearly designated.

The resulting positive sector swirl trends (Fig. 8, top) show a continuous decrease in magnitude along the flow path measurement stations at all radii. The decrease of approximately 50% from Station 0.5 to Station 1.0 is caused by viscous dissipation and vortex development in the connecting duct between the StreamVane exit plane and the fan rotor inlet plane. Because the in-plane secondary flow magnitude is only a fraction of the axial flow velocity, the dominant axial flow interacts with the bulk swirl and tightly wound vortex, reducing in-plane angular magnitudes. The positive sector swirl magnitude continues to decay an additional 25% through the fan rotor. This reduction results from the corotational direction of the positive sector swirl; blade loading is decreased in this situation, enhanced flow attachment and turning is achieved by the fan rotor blades, and distorted outlet flow conditions nearly recover to nondistorted conditions.

Counter-rotating, negative sector swirl (Fig. 8, bottom) remains essentially constant at each axial measurement plane along each radial measurement ring. Two factors have strong effects on negative sector swirl. First, the negative swirl angles are only associated with the tightly wound vortex feature of this distortion profile. This strong, relatively small feature is less influenced by the dominant axial flow direction and remains coherent at each measurement plane. Second, the counter-rotating direction of the tightly wound vortex increases the local fan rotor blade loading, reduces the turning effectiveness of the component, and results in strong negative sector swirl at the outlet of the fan rotor.

In both directions, the unique blade geometry features appear to have little to no effect on sector swirl magnitudes; however, the negative sector swirl results in the core flow region are slightly amplified at Station 1.5. This result is likely due to the aggressive turning near the hub radii associated with the fan rotor blade design found on this particular engine model.

The swirl intensity is discovered to decrease from one axial measurement plane to the next in the streamwise direction. From Station 0.5 to Station 1.0, the reduction of swirl intensity is more substantial in the core flow radii nearer to the engine centerline. Further along the span of the fan rotor blade, the resulting swirl intensity remains nearly constant between the first two axial planes. This result is due to the relatively strong, tightly wound vortex feature located at these radial positions. The tightly wound vortex is shown to exhibit less dissipation when compared to the bulk swirl flow field.

The fan rotor has a much more significant effect on swirl intensity than dissipation in the inlet duct. As shown, the swirl intensity levels exiting the fan rotor are reduced to less than half of the initial values. This result indicates that the fan rotor interacted with the distorted inlet flow profile and was capable of partially recovering flow turning effectiveness. Again, the data at radii associated with the tightly wound vortex feature indicate sustained swirl distortion through the fan rotor; however, reductions in magnitude are achieved.

The resulting swirl directivity along each radial measurement ring is summarized in Fig. 10. Bulk swirl at hub radii contributes to sustained, purely corotating swirl. Beyond approximately 75% tip radius, the effects of the counter-rotating tightly wound vortex partially balance the corotating bulk swirl, driving the average swirl directivity toward zero. Flow mixing in the inlet duct as well as fan rotor interactions cause a weakening of the effects of the tightly wound vortex, resulting in additional corotating swirl across the entire measurement domain. Scatter in the fan outlet swirl directivity data is attributable to core flow irregularities and fan rotor blade geometry. The swirl directivity of the flow at Station 1.5 within the core region indicates large amounts of flow mixing and little coherence in any particular direction.

Figure 11 shows the resulting swirl pairs values at each measurement location. At Station 0.5, a single bulk swirl feature accounts for the flow profile from centerline to approximately 75% tip radius. At greater radial locations, the swirl pairs indicator trends toward unity, specifying the existence of, at minimum, one vortex pair (i.e., the corotating bulk swirl and the counter-rotating tightly wound vortex). The appearance of the single vortex pair diminishes at Station 1.0 where the counter-rotating tightly wound vortex feature dominates. This is also evident at Station 1.5, which indicates only the tightly wound vortex feature survives the fan rotor interactions and persists downstream.

Overall, distortion effects were decreased as the inlet swirl distortion propagated downstream from the StreamVane device to the fan inlet and continued through the fan rotor. Calculated swirl distortion parameters at Station 1.5 indicated that the fan was incapable of completely recovering from nonuniform inlet flow conditions.

To better understand the capabilities of conventional turbofan engines operating under distorted inlet flow conditions, an experiment using a StreamVane swirl distortion generator was conducted with the primary goal of collecting flow properties at two planes surrounding and isolating the fan rotor. The measurements yielded insight to fan tolerance of widely, continuously varying incidence angle, asymmetric distortion profiles, and multiple types of distortion within a single profile.

Initial planar experimental data indicated that the fan rotor partially processed the distortion and reduced absolute flow angle magnitudes across the entire domain; however, flow angles exiting the fan rotor remained off-design and confirmed that the component was incapable of complete recovery to nominal flow directional values. Furthermore, evidence of a coherent vortex structure in the secondary flow streamlines at the fan rotor exit plane revealed that a high intensity tightly wound vortex distortion feature passed nearly unaltered through the fan rotor.

This investigation also included a first of its kind analysis of fan rotor effects on swirl descriptors. Previously, swirl descriptors have been used to classify and describe inlet swirl distortion profiles at the AIP within the engine inlet duct. By extending the definitions to fan outlet measurements, direct analysis of fan rotor capabilities was achieved. In general, the magnitudes of the various swirl descriptors were reduced by the fan, further indicating that the fan rotor was able to process the distorted flow and partially recover to nominal flow conditions.

Comparison of swirl intensity at the fan inlet and outlet measurement planes revealed a reduction of approximately 50% at all radial measurement locations. The direction of the swirl relative to the fan rotation direction was determined to have a significant effect on the sector swirl results. Corotating swirl caused negative incidence, reduced fan rotor blade loading, resulted in nearly nominal flow turning, and drove positive sector swirl values toward zero. Counter-rotating swirl caused positive incidence, increased fan rotor blade loading, decreased turning effectiveness, and resulted in little to no effect on negative sector swirl.

Corotating bulk swirl dominated the directivity of the swirl distortion profile. The relatively small extent and weakened intensity of the tightly wound vortex reduced the impact of the local feature on the overall in-plane fluid direction as the distortion propagated through the inlet duct and fan rotor. This result was apparent in the diminished swirl pairs descriptor which collapsed toward an indication of zero coherent swirl pairs at both the rotor inlet and outlet measurement planes.

While the swirl descriptors indicated relative intensity and directivity of the inlet swirl distortion profile as it progressed from initiation, to fan inlet, and finally to fan outlet, the descriptors failed to illustrate specifics about the flow field. Small-scale vortex features were easily dominated in the descriptor values by large-scale bulk swirl. Additionally, radial flow information was lost through the use of swirl descriptors alone. The tangential direction of co- and counter-rotating flow is known to have the greatest impact on blade loading, flow separation, rotor stall/surge, and compression efficiency; however, radial flow directions can indicate flow redistribution and component geometry effects (i.e., midspan shrouds) and should not be ignored.

This investigation supports the use of classical swirl descriptors to quantify fan rotor effects on inlet swirl distortions. Results demonstrated that the fan rotor interacted with the inlet swirl distortion and was capable of partially recovering outlet flow conditions to more closely match design values. Using swirl descriptors alone did result in suppressing local small-scale features. It is, therefore, advised that swirl descriptors be used to quantify relative levels of inlet swirl distortion, while full annulus profile maps continue to be used to highlight particular flow features within the distortion.

The authors would like to acknowledge the National Institute of Aerospace (NIA) and NASA Langley Research Center for funding this work in association with NASA's Environmentally Responsible Aviation Project, project managers Fay Collier (LaRC), Hamilton Fernandez (LaRC), Greg Gatlin (LaRC), and Bo Walkley (NIA). The authors would like to additionally thank the Boeing Research and Technology Company for assistance with the BLI inlet flow profiles (POCs: John Bonet and Ron Kawai).

• NASA's Environmentally Responsible Aviation Project (NIA Cooperative Agreement No. RD-2917), Langley Research Center (RD-2917).

$αr$ =

radial flow angle

$r$ =

radial measurement location

$rtip$ =

fan blade tip radius

$SDi$ =

swirl directivity of the ith radial measurement ring

$SIi$ =

swirl intensity of the ith radial measurement ring

$SPi$ =

swirl pairs of the ith radial measurement ring

$SSi+$ =

positive sector swirl of the ith radial measurement ring

$SSi−$ =

negative sector swirl of the ith radial measurement ring

$αt$ =

tangential flow angle (swirl angle)

$θ$ =

circumferential measurement location

$θi+$ =

circumferential extent of positive swirl angle for the ith radial measurement ring

$θi−$ =

circumferential extent of negative swirl angle for the ith radial measurement ring