This paper is part of a two paper series on optimization methods for film cooling which seek to address the limitations of experimental optimization by utilizing advances in RANS based CFD to quickly optimize film cooling hole geometries. In the companion paper [1] on parametric optimization the optimum hole was experimentally demonstrated to have > 40% improvement in spatially averaged effectiveness compared to a baseline 7-7-7 hole, and was developed by leveraging RANS as a proxy for experimental data. In this paper adjoint based optimization was used to develop unique film cooling hole geometries. Adjoint optimization moves beyond using RANS as a proxy for experimental data instead utilizing the derivatives available in RANS to fully optimize the geometry of a shaped film cooling hole. The resulting geometry was experimentally validated to further increase performance by over 80% compared to the baseline 7-7-7 shaped hole. The study also show that further increases in performance are predicted when expanding the optimization target region. Furthermore, these new optimized geometries are readily manufactured by Additive Manufacturing (AM) processes and significantly less time consuming to generate than an equivalent parametrically optimized hole shape. These methods provide the tools necessary to fully utilize the large design space offered by AM and will be dramatically shift the future of film cooling hole design.

In this study, a combination of computational simulation and experimental testing was used to evaluate a broad range of forward and lateral expansion angles for a turbine film cooling shaped holes. The study demonstrates the utilizing of RANS based CFD to quickly screen potential optimized geometries, followed by experimental determination of true performance characteristics. As a baseline, the performance of all film cooling holes was evaluated using an internal coolant channel cross-flow. Also, all hole geometries incorporated a filleted inlet-plenum interface, which presumes use of additive manufacturing to construct the turbine components. Experimental validation confirmed that the computational simulations predicted the correct relative performance of various hole geometries, even though actual performance levels were not predicted well. This investigation showed that the performance of laidback, fan shaped holes was much more sensitive to the lateral expansion angle than the forward expansion angle. The optimum shaped hole configuration was found to be a hole with a 15° lateral expansion angle and a 1° forward expansion angle (15-15-1 configuration), which had a maximum average adiabatic effectiveness 40% greater than the baseline 7-7-7 open literature hole. This study also showed that the shaped hole diffuser performance is primarily a function only three parameters: the coolant jet velocity ratio, VR, the shaped hole area ratio, AR, and the hole exit width relative to the pitch between holes, t/P.

Film cooling is used to protect turbine components from the extreme temperatures by ejecting coolant through arrays of holes to create an air buffer from the hot combustion gases. Limitations in traditional machining meant film cooling holes universally have sharp inlets which create separation regions at the hole entrance. The present study uses experimental and computational data to show that these inlet separation are a major cause of performance variation in crossflow fed film cooling holes. Three hole designs were experimentally tested by independently varying the coolant velocity ratio (VR) and the coolant channel velocitty ratio (VR c ) to isolate the effects of crossflow on hole performance. Leveraging additive manufacturing technologies, the addition of a 0.25D radius fillet to the inlet of a 7-7-7 shaped hole is shown to significantly improve diffuser usage and significantly reduce variation in performance with VR c . A second AM design used a very large radius of curvature inlet to reduce biasing caused by the inlet crossflow. Experiments showed that this “swept” hole design did minimize biasing of coolant flow to one side of the shaped hole and it significantly reduced variations due to varying VR c . RANS simulations at six VR and three VR c conditions were made for each geometry to better understand how the new geometries changed the velocity field within the hole. The sharp and rounded inlets were seen to have very similar tangential velocity fields and jet biasing. Both AM inlets created more uniform, slower velocity fields entering the diffuser. The results of this paper indicate large improvements in film cooling performance can be found by leveraging AM technology.

Film cooling is a very common technique used in cooling turbine engine components, and hence has been studied extensively experimentally and computationally. Computational studies of film cooling range from more simplistic RANS predictions to high fidelity LES predictions. Generally the accuracy of computational predictions of film cooling is evaluated based on the adiabatic effectiveness measured and predicted downstream of the hole. For this study, a RANS computational prediction was used, but the evaluation of the accuracy of the prediction was based on measured thermal and velocity fields within the coolant hole and immediately downstream. We chose a relatively complex film cooling configuration consisting of a row of 7-7-7 shaped holes fed by an internal channel flow with a range of internal crossflow velocities and coolant jet velocities. Previous experimental studies using this configuration showed that at various inlet velocity ratios, the coolant jet becomes biased to one side of the diffusing section of the hole, which degrades performance and can cause ingestion into the hole. For this study we wanted to determine the capability of a RANS computation to correctly predict the flow structures, coolant jet biasing, and film effectiveness for this configuration. Computational results were compared to thermal field measurements made with a micro-thermocouple probe, velocity field measurements made with a PIV, and film effectiveness measurements made with an IR camera. These measurements were made within the coolant hole, at the downstream edge of the hole, and at 5 D downstream of the hole. Results from this study show that the RANS computations accurately predicted the bulk flow and thermal fields within and at the downstream edge of the hole, but failed to predict the evolutions of the thermal field, secondary flows, and film effectiveness downstream of the hole.

Internal crossflow, or internal flow that is perpendicular to the overflowing mainstream, reduces film cooling effectiveness by disrupting the diffusion of coolant at the exit of axial shaped holes. Previous experimental investigations have shown that internal crossflow causes the coolant to bias toward one side of the diffuser and that the severity of the biasing scales with the inlet velocity ratio, VR i , or the ratio of crossflow velocity to the jet velocity in the metering section of the hole. It has been hypothesized and computationally predicted that internal crossflow produces an asymmetric swirling flow within the hole that causes the coolant to bias in the diffuser and that biasing contributes to ingestion of hot mainstream gas into the hole, which is undesirable. However, there are no experimental measurements as of yet to confirm these predictions. In the present study, in and near-hole flow field and thermal field measurements were performed to investigate the flow structures and mainstream ingestion for a standard axial shaped hole fed by internal crossflow. Three different inlet velocity ratios of VR i = 0.24, 0.36, and 0.71 were tested at varying injection rate. Measurements were made in planes normal to the nominal direction of coolant flow at the outlet plane of the hole and at two downstream locations — x/d = 0 and 5. The predicted swirling structure was observed for the highest inlet velocity ratio and flow within the hole was shown to scale with VR i . Ingestion within the diffuser was significant and also scaled with VR i . Downstream flow and thermal fields showed that increased biasing contributed to more severe jet detachment and coolant dispersion away from the surface.

Most studies of turbine airfoil film cooling in laboratory test facilities have used relatively large plenums to feed flow into the coolant holes. However, a more realistic inlet condition for the film cooling holes is a relatively small channel. Previous studies have shown that the film cooling performance is significantly degraded when fed by perpendicular internal crossflow in a smooth channel. In this study, angled rib turbulators were installed in two geometric configurations inside the internal crossflow channel, at 45° and 135°, to assess the impact on film cooling effectiveness. Film cooling hole inlets were positioned in both pre-rib and post-rib locations to test the effect of hole inlet position on film cooling performance. A test was performed independently varying channel velocity ratio and jet to mainstream velocity ratio. These results were compared to the film cooling performance of previously measured shaped holes fed by a smooth internal channel. The film cooling hole discharge coefficients and channel friction factors were also measured for both rib configurations with varying channel and inlet velocity ratios. Spatially-averaged film cooling effectiveness is largely similar to the holes fed by the smooth internal crossflow channel, but hole-to-hole variation due to inlet position was observed.