An Inverse Design Based Methodology for Rapid 3D Multi-Objective/Multidisciplinary Optimization of Axial Turbines

Design of axial turbines, especially LP turbines, poses difficult tradeoffs between requirements of aerodynamic design and structural limitations. In this paper, a methodology is proposed for 3D multi-objective design of axial turbine blades in which a 3D inverse design method is coupled with a multi-objective genetic algorithm. By parameterizing the blade using blade loading parameters, spanwise work distribution and maximum thickness, a large part of the design space can be explored with very few design parameters. Furthermore, the inverse method not only computes the blade shape but also provides accurate 3D inviscid flow information. In the simple multi-disciplinary approach proposed here the different losses in axial turbines such as endwall losses, tip leakage losses and an indication of flow separation are related through well known correlations to the blade surface velocities predicted by the inverse design method. In addition, geometrical features such as throat area, lean angles and airfoil cross sectional area are computed from the blade shape employed during the optimization. Also, centrifugal stresses and bending stresses are related to the blade geometry. The methodology is then applied to the redesign of an LP turbine rotor with the aim of reducing the maximum stresses while maintaining the performance of the rotor. The results are confirmed by using the commercial CFX CFD (Computational Fluid Dynamics) code and Ansys FEA (Finite Element Analysis) codes.

Adjoint-Based Flow Sensitivity Analyis Using Arbitrary Control Surfaces

This article describes how to extend the dsicrete adjoint method to functionals that are evaluated on arbitrary rotational control surfaces that intersect the flow domain at a position specified by the user, e.g. the pressure loss coefficient of a single blade in a multi-stage configuration. The definition and implementation of the mixed-out states on such surfaces is revisited. The calculation of the corresponding right-hand sides in the adjoint system is explained. These techniques can be used to specify functionals that quantify the deviation of the radial distribution of the flow angles, relative mass flow, etc. from a given target distribution. Sensitivity studies using the conventional approach, i.e. by means of finite differences of many steady solutions, are compared to results based on the adjoint method. The applications demonstrate that the agreement between adjoint and conventional sensitivity predictions is excellent, if the exact definition of the surface functionals is taken into account.

Tip Clearance Investigation of a Ducted Fan Used in VTOL UAVS: Part 1—Baseline Experiments and Computational Validation

Ducted fans that are popular choices in vertical take-off and landing (VTOL) unmanned aerial vehicles (UAV) offer a higher static thrust/power ratio for a given diameter than open propellers. Although ducted fans provide high performance in many VTOL applications, there are still unresolved problems associated with these systems. Fan rotor tip leakage flow is a significant source of aerodynamic loss for ducted fan VTOL UAVs and adversely affects the general aerodynamic performance of these vehicles. The present study utilized experimental and computational techniques in a 22″ diameter ducted fan test system that has been custom designed and manufactured. Experimental investigation consisted of total pressure measurements using Kiel total pressure probes and real time six-component force and torque measurements. The computational technique used in this study included a 3D Reynolds-Averaged Navier Stokes (RANS) based CFD model of the ducted fan test system. RANS simulations of the flow around rotor blades and duct geometry in the rotating frame of reference provided a comprehensive description of the tip leakage and passage flow. The experimental and computational analysis performed for various tip clearances were utilized in understanding the effect of the tip leakage flow on aerodynamic performance of ducted fans used in VTOL UAVs. The aerodynamic measurements and results of the RANS simulations showed good agreement especially near the tip region.

Efficient Gradient-Based Algorithms for the Construction of Pareto Fronts

The objective of this study is to develop and assess a gradient-based algorithm that efficiently traverses the Pareto front for multi-objective problems. We use high-fidelity, computationally intensive simulation tools (for eg: Computational Fluid Dynamics (CFD) and Finite Element (FE) structural analysis) for function and gradient evaluations. The use of evolutionary algorithms with these high-fidelity simulation tools results in prohibitive computational costs. Hence, in this study we use an alternate gradient-based approach. We first outline an algorithm that can be proven to recover Pareto fronts. The performance of this algorithm is then tested on three academic problems: a convex front with uniform spacing of Pareto points, a convex front with non-uniform spacing and a concave front. The algorithm is shown to be able to retrieve the Pareto front in all three cases hence overcoming a common deficiency in gradient-based methods that use the idea of scalarization. Then the algorithm is applied to a practical problem in concurrent design for aerodynamic and structural performance of an axial turbine blade. For this problem, with 5 design variables, and for 10 points to approximate the front, the computational cost of the gradient-based method was roughly the same as that of a method that builds the front from a sampling approach. However, as the sampling approach involves building a surrogate model to identify the Pareto front, there is the possibility that validation of this predicted front with CFD and FE analysis results in a different location of the “Pareto” points. This can be avoided with the gradient-based method. Additionally, as the number of design variables increases and/or the number of required points on the Pareto front is reduced, the computational cost favors the gradient-based approach.

Optimal Application of Riblets on Compressor Blades and Their Contamination Behavior

During the last decades, riblets have shown a potential for viscous drag reduction in turbulent boundary layers. Several investigations and measurements of skin-friction in the boundary layer over flat plates and on turbomachinery type blades with ideal riblet geometry have been reported in the literature. The question where riblets must be applied on the surface of a compressor blade is still not sufficiently answered. In a first step, the profile loss reduction by ideal triangular riblets with a trapezoidal groove and a constant geometry along the surface on the suction and pressure side of a compressor blade is investigated. The results show a higher potential on the profile loss reduction by riblets on the suction side. In a second step, the effect of laser-structured ribs on the laminar separation bubble and the influence of these structures on the laminar boundary layer near the leading edge are investigated. After clarifying the best choices where riblets should be applied on the blade surface, a strategy for locally adapted riblets is presented. The suction side of a compressor blade is laser-structured with a segmented riblet-like structure with a constant geometry in each segment. The measured profile loss reduction shows the increasing effect on the profile loss reduction of this locally adapted structure compared to a constant riblet-geometry along the surface. Furthermore, the particle deposition on a riblet-structured compressor blade is investigated and compared to the particle deposition on a smooth surface. Results show a primary particle deposition on the riblet tips followed by an agglomeration. The particle deposition on the smooth surface is stochastic.

Optimization of an Axial Turbine Stage for Aerodynamic Inlet Blockage

This paper presents a practical and effective optimization approach to minimize 3D-related flow losses associated with high aerodynamic inlet blockage by re-stacking the turbine rotor blades. This approach is applied to redesign the rotor of a low speed subsonic single-stage turbine that was designed and tested in DLR, Germany. The optimization is performed at the design point and the objective is to minimize the rotor pressure loss coefficient as well as the maximum von Mises stress while keeping the same design point mass flow rate, and keeping or increasing the rotor blade first natural frequency. A Multi-Objective Genetic Algorithm (MOGA) is coupled with a Response Surface Approximation (RSA) of the Artificial Neural Network (ANN) type. A relatively small set of high fidelity 3D flow simulations and structure analysis are obtained using ANSYS Workbench Mechanical. That set is used to train and to test the ANN models. The stacking line is parametrically represented using a quadratic rational Bezier curve (QRBC). The QRBC parameters are directly related to the design variables, namely the rotor lean and sweep angles and the bowing parameters. Moreover, it results in eliminating infeasible shapes and in reducing the number of design variables to a minimum while providing a wide design space for the blade shape. The aero-structural optimization of the E/TU-3 turbine proved successful, the rotor pressure loss coefficient was reduced by 9.8% and the maximum von Mises stress was reduced by 36.7%. This improvement was accomplished with as low as four design variables, and is attributed to the reduction of 3D-related aerodynamic losses and the redistribution of stresses from the hub trailing edge region to the suction side maximum thickness area. The proposed parametrization is a promising one for 3D blade shape optimization involving several disciplines with a relatively small number of design variables.

A Dual Flow High Mach Number Centrifugal Compressor for a Small Gas Turbine APU

Small gas turbine auxiliary power units (APU’s) of conventional load compressor type wherein the gas generator or core module directly drives a separate centrifugal load compressor are installed in aircraft and helicopters to supply both compressor air for main engine starting and air conditioning combined with shaft power to drive an electric generator. This paper describes the test development of a dual flow centrifugal compressor (DFC) where the impeller flow was split into two streams, the inner (hub) stream supplying compressed air to the gas generator core module, and the outer (DFB) bleed stream delivering a compressed air to the aircraft pneumatic power system. DFC development rig testing revealed that the hub or core stream satisfied compressor design requirements but that the DFB stream flowpath demonstrated unstable characteristics with decreasing efficiency as test speeds were increased. At the time of the development program in the early 1990’s convergence difficulties were encountered with CFD attempts to corroborate the test results, and thus pinpoint plausible explanations, as a consequence a renewed upgraded 2010 CFD analysis of the dual flow compressor is presented herein confirming the test performance characteristics of both flow streams and the fundamental reason for poor DFB performance as excessive diffusion at high relative Mach numbers.

Aerodynamic Characterization of Transonic Turbine Vanes Optimized to Attenuate Rotor Forcing

A multi-objective optimization procedure is applied to the 3D design of a transonic turbine vane row, considering efficiency and stator outlet pressure distortion, which is directly related to induced rotor forcing. The characteristic features that define different individuals along the Pareto Front are described, analyzing the differences between high efficiency airfoils and low interaction. Pressure distortion is assessed by means of a model that requires only of the computation the steady flow field in the domain of the stator. The reduction of aerodynamic rotor forcing is checked via unsteady multistage aerodynamic computations. A well known loss prediction method is used to drive the efficiency of one optimization run, while CFD analysis is used for another, in order to assess the reliability of both methods. In both cases, the decomposition of total losses is performed to quantify the influence on efficiency of reducing rotor forcing. Results show that when striving for efficiency, the rotor is affected by few, but intense shocks. On the other hand, when the objective is the minimization of distortion, multiple shocks will appear.

Multiscale Partially Averaged Navier Stokes Approach for the Prediction of Flow in Linear Compressor Cascade With Moving Casing

In this paper we present an innovative Partially Averaged Navier Stokes (PANS) approach for the simulation of turbomachinery flows. The elliptic relaxation k-ε-ζ-f model was used as baseline Unsteady Reynolds Averaged Navier Stokes (URANS) model for the derivation of the PANS formulation. The well established T-FlowS unstructured finite volume in-house code was used for the computations. A preliminary assessment of the developed formulation was carried out on a 2D hill flow that represents a very demanding test case for turbulence models. The turbomachinery flow here investigated reproduces the experimental campaign carried out at Virginia Tech on a linear compressor cascade with tip leakage. Their measurements were used for comparisons with numerical results. The predictive capabilities of the model were assessed through the analysis of the flow field. Then an investigation of the blade passage, where experiments were not available, was carried out to detect the main loss sources.

Application of RANS and LES to the Prediction of Flows in High Pressure Turbine Components

This paper investigates the capability of numerical simulations to estimate unsteady flows and wall heat fluxes in turbine components with both structured and unstructured flow solvers. Different numerical approaches are assessed, from steady-state methods based on the Reynolds Averaged Navier-Stokes (RANS) equations to more sophisticated methods such as the Large Eddy Simulation (LES) technique. Three test cases are investigated: the vortex shedding induced by a turbine guide vane, the wall heat transfer in another turbine guide vane and a separated flow phenomenon in an internal turbine cooling channel. Steady flow simulations usually fail to predict the mean effects of unsteady flows (such as vortex shedding) and wall heat transfer, mainly because laminar-to turbulent transition and the inlet turbulent intensity are not correctly taken into account. Actually, only the LES (partially) succeeds to accurately estimate unsteady flows and wall heat fluxes in complex configurations. The results presented in this paper indicate that this method considerably improves the level of physical description (including boundary layer transition). However, the LES still requires developments and validations for such complex flows. This study also points out the dependency of results to parameters such as the freestream turbulence intensity. When feasible solutions obtained with both structured and unstructured flow solvers are compared to experimental data.