A numerical study of the effect of wake passing on turbine blade film cooling

_ Time-accurate and steady three-dimensional Turbomachineryflow fields are inherently unsteady viscous turbulentnumerical simulationswere performed due to the relative motion of adjacent blade rows. The to study the effect of upstream blade wake passing passing of wakes from the upstream blade row causes unsteadiness on the performance of film cooling on a periodic fluctuations in both the magnitude and relative downstream axial turbine blade. The simulations direction of the flow velocity in the downstreamblade modeled the blade as spanwise periodic and of infinite row. In addition,secondaryflows are inducedby tuming span. Both aerodynamic and heat transfer quantities of the radiallynon-uniform flowfields enteringupstream were explored. A showerheadfilm cooling arrangement blade rows. Sharma et al. [I] claim that this flow typical of modem gas turbine engines was employed, unsteadiness has an adverse effect on turbomachinery Showerhead cooling was studied because of its efficiency of several points relative to the steady anticipated strong sensitivity to upstream flow time-mean flow. In turbines, this is likely due in large fluctuations. The wake was modeled as a region of zero part to the effect of the flow unsteadiness on turbine axial velocity on the upstream computational boundary bladeboundary layer transition [2]. which translated with each iteration. This model is compatible with a planned companion experiment in A useful figure of merit for aircraft turbine engines which the wakes will be produced by a rotating row of is the specificfuel consumption,or SFC. This is the rate cylindrical rods upstream of an annular turbine cascade, of fuel consumptionper unit thrust of the engine. For a It was determined that a steady solution withappropriate given flight Mach number, the minimum ideal SFC is upstream swirl and stagnation pressure predicted the dependenton turbine inlet temperature. As turbine inlet span-average film effectiveness quite well. The major temperatureincreases,minimum ideal SFC decreases, so difference is a 2 to 3 percent overprediction of there is a continual incentive to increase the turbine inlet span-average film effectivenessby the steady simulation temperature. This incentivehas givenrise to turbineinlet on the pressure surface and in the showerhead region, temperatures several hundred degrees above current Local overpredictions of up to 8 percent were observed blade material operating temperatures. Due to thermal in the showerhead region These differences can be limitationsof turbine materials, the turbine blades must explained by the periodic relative lifting of the boundary often be cooled by passing cooler compressorair through layer and enhancedmixing in the unsteadysimulations, the blades. When internal cooling alone is inadequate, film coolingmust be employed in the first several blade rows [3]. In this case, the coolant air is discharged through small holes in the turbine surface to form a protective film between the turbine blade and the hot combustor discharge gas. These holes are usually *Aerospace Engineer,PropulsionSystemsDivision concentrated on the pressure surface and leading edge region of the blades. To remain effective, the film coolant must remain near the surface and not separate into the free stream. However, mechanical constraints prohibit ejection at very small angles from the blade surface. Thus it is extremely important to know the trajectory of the film coolant under a variety of flow conditions. Until this time, most detailed research of the film associated with wake and shock passing and potential coolant flow has considered the turbine free stream flow interactions,and effects generated by three-dimensional to be steady [4]. Studies of film cooled turbine stages secondary flows in upstream blade rows. These have included unsteadiness [5], but have lacked secondary flows include tip clearance vortices, passage sufficient detail to isolate the important physical vortices, horseshoevortices, and the relative eddy. It is phenomena associated with film coolant flow. To this expectedthat the primary unsteady effecton film coolant end, this study aims to investigate the effect of flow flow in subsonicand moderatelytransonic turbinesis due unsteadiness on turbine film cooling in a more detailed to wake passing, especially near midspan. Further, and fundamental manner. Previous research has secondary flows are highly dependent on the turbine considered showerhead cooling on a flat body with a geometry,and are not easily generalized to a broad range blunt leading edge. For example, Mick and Mayle [6] of problems. Thus it seems proper to concentrateon the conducted steady showerhead film cooling experiments effectof wake passingon the film coolantflow. Thiscan on a circular leadingedge with a fiat afterbodyand found be accomplished using a rotating rod arrangement large gradients in the showerhead region for both film upstream of a turbine cascade [7]. Such an arrangement effectiveness and heat transfer coefficient. Detailed produces a periodic wake pattem which impinges on the experimental results for showerhead film cooling with turbine blades. The rods are sized to match the trailing representative blade geometries, particularly in the edge diameter of an appropriate inlet guide vane for the unsteadyenvironment,are lacking in the literature, test turbine. There is some debate as to the similarity of cylinder wakes to blade wakes [8]. However cylinder Research is continuing throughout the industry to wakes have been shown [9] to accurately represent the develop new turbine blade materials, both metallic and relativevelocity vector diagram and mean wake velocity non-metallic, which are able to withstand higher profilesofanactualinletguidevane. temperatures. However, even if materials are developed in the future which safely withstandcurrent turbine inlet The NASA Lewis Rotor-Wake Heat Transfer Rig temperatures, the incentive will remain to achieve even (Figure I) was chosen as an appropriate facility to higher turbine inlet temperatures through cooling of the investigate wake induced unsteadiness effects. The new material. Thus it seems that turbine film cooling computation described in this report attempts to model will remain a valid research concern well into the 21st the geometry and flow conditions to be studied in the century, future companion experiment. In this experiment, a rotating set of 0.32 cm diameter cylindrical rods are pRORLEM DESCRIPTION placed upstream of an annular turbine cascade consisting of 23 blades [10]. A maximum of 24 rods may be used In theory, the unsteadyflow field in a turbinecan be in the rotor. The rotor speed is adjustable up to 7000 modeled using current three-dimensional viscous rpm. The cascade is representative of a lightly loaded unsteady flow codes. Such codes can even model film turbine rotor. The blades have 67 degrees of turningand cooled blades through flow injection at selected grid one blade is film-cooled and instrumented. The points [5]. However,the computationaltimerequiredfor instrumentation will consist of nickel thin film gauges such a computation in sufficientdetail for this problem capable of resolving both chordwise and spanwise would be enormous,making it impractical to perform a variationsin unsteady temperature. This informationwill parametric study of the importantvariables. In addition, be used todetermineunsteady film effectivenessand heat it is not guaranteed that an improved understandingof transfer coefficient profiles on the blade surface for the flow physics would be obtained from such an effort, comparisonwith the computationalresultsof this study. Therefore this problem is studied through a combination of physical and computational experiments. These The film cooling scheme consists of five staggered experiments are flexible enough to allow application to rows of showerhead film holes. The pitch-to-diameter an appropriate matrix of operating conditions and ratio in both the spanwise and streamwise directions is detailed enough to enable accurate interpretationof the 4.0. The holes are angled 30 degrees to the blade in the underlying flow physics. Through this effort, it is spanwise direction, and 90 degrees in the streamwise anticipated that improved physics-based unsteady film direction. Showerhead cooling was chosen because of coolant flow models will be formulatedand incorporated the morepressing need for film cooling [5] and thelarger into the steady design codes, temporalfluctuationsin static pressure [11] in the leading edgeregion. The unsteady effects in turbines can generally be divided into two classes: two-dimensional effects IINSTEADYCOMPlITATIONDESCRIPTION in the spanwise direction. This angle produces a 2:1 aspect ratio ellipse at the blade surface. Figure 3 shows A three-dimensional viscous turbulent calculation the surfacegrid with all holepoints removed. Becauseof was performed for the experimental geometry using the the discontinuitybetween wall and hole, wall and hole code RVC3D [12]. The code solves the thin-layer boundary conditions were smoothly interpolated for Navier-Stokes equations with an explicit points straddlingthe boundaries. It was determinedthat finite-difference technique. The Baldwin-Lomax this method was preferred over attempting to distort the turbulence model was employed. The calculation surface grid to conform to the elliptical holes. Such modeled the flowfield as a linearcascade with spanwise distortions were found to produce spatial spikes of periodicity based on a unit cell of the film hole pattern, increased entropy upon close inspection of earlier Although the experiment is not precisely periodicdue to calculations. the annular geometry and endwall effects, this simplification greatly reduces the number of grid points The unsteady calculation was performed on the required to resolve the flow field. This is especially NASA Lewis Cray Y-MP supercomputer.