Paul Galpin

Coupled Algebraic Multigrid for Free Surface Flow Simulations

An accurate, efficient algorithm for solving free surface flows with ANSYS CFX is described. Accuracy is achieved using a compressive advection discretization which maintains a sharp free surface interface representation without relying on a small timestep. Efficiency is obtained using a solution algorithm which implicitly couples velocity, pressure, and volume fractions in the same matrix, and solves these equations using algebraic multigrid. This coupled strategy overcomes difficulties encountered with segregated volume fraction algorithms, where heavy underrelaxation and long solution times are required. The resulting solution algorithm is scalable, leading to solution times which increase linearly with mesh size.Copyright

CFD Analysis of a 15 Stage Axial Compressor: Part I — Methods

The flow field of a 15 stage axial compressor is analyzed using a 3-D Navier-Stokes CFD tool. The compressor under investigation is a prototype first compressor version before optimization of the Siemens V84.3A family. A methodology is described for steady state and transient flow simulations of the entire 15 stages compressor in one computation (not piece by piece). The simulation includes tip gaps, mass bleeds, hub leakage flows, and ranges from single passage to full 360 degrees analysis. The work is divided into two companion papers. This first part, “CFD Analysis of a 15 Stage Axial Compressor Part I: Methods” describes the overall methodology used, based on a middle portion of the compressor R5S9 (Rotor 5 to Stator 9). Various effects are analyzed: mesh style and refinement, boundary conditions, steady or transient, tip clearance, and numerical issues (turbulence model choice, advection model choice, parallel processing performance). A high sensitivity of the predictions to the tip clearance height was found. Excellent design point predictions are obtained with steady-state frame change interface models (Stage average interface), as well as with transient simulations (transient rotor-stator interface).Copyright

The Efficient Computation of Transient Flow in Turbine Blade Rows Using Transformation Methods

Computational predictions of the transient flow in turbine blade rows are considered. Adjacent blade rows typically contain unequal numbers of blades and vanes, requiring a computation over multiple passages per row to permit application of simple periodic boundary conditions. For typical geometries, use of conventional solution methods requires computation over all or a significant portion of the wheel to ensure a time accurate solution.The computational load is significantly reduced by methods which enable a one or two-passage solution to accurately model the full wheel (or part wheel, if applicable) solution. In this work, three methods are used: Profile Transformation, Fourier Transformation and Time Transformation.This paper will concentrate on the evaluation of these methods on two turbine geometries. The first test case is a frozen gust analysis for a high pressure transonic turbine; the geometry includes hub and casing cavities together with a complex tip. The second test case is a low pressure turbine stage run over a range of operating points. Comparisons between the various methods and the equivalent part wheel periodic solution are made to demonstrate the accuracy and computational efficiency of the transformation methods.Copyright

CFD Analysis of a 15 Stage Axial Compressor: Part II — Results

The flow field of a 15 stage axial compressor is analyzed using a 3-D Navier-Stokes CFD tool. The compressor under investigation is a prototype engine, first compressor version before optimization of the Siemens V84.3A family. The paper describes steady state and transient flow simulations of the entire 15 stages compressor in one computation (not piece by piece). The simulation includes tip gaps, mass bleeds, hub leakage flows, and ranges from single passage to full 360 degrees analysis. The work is divided into two companion papers. The second paper, “CFD Analysis of a 15 Stage Axial Compressor Part II: Results” describes the application of the methods in Part I to the entire 15 stage compressor (Belamri et al, 2005). The flow in the compressor is modeled first with one blade passage per component (periodicity assumed, an interface pitch change model employed). Steady state and transient models are compared. In a second series of computations, all blade passages in 360 degrees are modeled, (no periodicity or pitch change assumptions required), for portions of the compressor. The various simulation approaches are compared to each other, and to experimental data. Good agreement between predictions and experimental results, both in the details of the flow field and the integral prediction of operating range of the compressor, were found.© 2005 ASME

Turbulence Resolving Flow Simulations of a Francis Turbine in Part Load using Highly Parallel CFD Simulations

The operation of Francis turbines in part load conditions causes high fluctuations and dynamic loads in the turbine and especially in the draft tube. At the hub of the runner outlet a rotating vortex rope within a low pressure zone arises and propagates into the draft tube cone. The investigated part load operating point is at about 72% discharge of best efficiency. To reduce the possible influence of boundary conditions on the solution, a flow simulation of a complete Francis turbine is conducted consisting of spiral case, stay and guide vanes, runner and draft tube. As the flow has a strong swirling component for the chosen operating point, it is very challenging to accurately predict the flow and in particular the flow losses in the diffusor. The goal of this study is to reach significantly better numerical prediction of this flow type. This is achieved by an improved resolution of small turbulent structures. Therefore, the Scale Adaptive Simulation SAS-SST turbulence model - a scale resolving turbulence model - is applied and compared to the widely used RANS-SST turbulence model. The largest mesh contains 300 million elements, which achieves LES-like resolution throughout much of the computational domain. The simulations are evaluated in terms of the hydraulic losses in the machine, evaluation of the velocity field, pressure oscillations in the draft tube and visual comparisons of turbulent flow structures. A pre-release version of ANSYS CFX 17.0 is used in this paper, as this CFD solver has a parallel performance up to several thousands of cores for this application which includes a transient rotor-stator interface to support the relative motion between the runner and the stationary portions of the water turbine.

Computational Fluid Dynamics Modeling of Low Pressure Steam Turbine Radial Diffuser Flow by Using a Novel Multiple Mixing Plane Based Coupling—Simulation and Validation

The diffuser and exhaust of low pressure steam turbines show significant impact on the overall turbine performance. The amount of recovered enthalpy leads to a considerable increase of the turbine power output, and therefore a continuous focus of turbine manufacturers is put on this component. On the one hand, the abilities to aerodynamically design such components are improved, but on the other hand a huge effort is required to properly predict the resulting performance and to enable an accurate modeling of the overall steam turbine and therewith plant heat rate. A wide range of approaches is used to compute the diffuser and exhaust flow, with a wide range of quality. Today, it is well known and understood that there is a strong interaction of rear stage and diffuser flow, and the accuracy of the overall diffuser performance prediction strongly depends on a proper coupling of both domains. The most accurate, but also most expensive method is currently seen in a full annulus and transient coupling. However, for a standard industrial application of diffuser design in a standard development schedule, such a coupling is not feasible and more simplified methods have to be developed. The paper below presents a computational fluid dynamics (CFD) modeling of low pressure steam turbine diffusers and exhausts based on a direct coupling of the rear stage and diffuser using a novel multiple mixing plane (MMP). It is shown that the approach enables a fast diffuser design process and is still able to accurately predict the flow field and hence the exhaust performance. The method is validated against several turbine designs measured in a scaled low pressure turbine model test rig using steam. The results show a very good agreement of the presented CFD modeling against the measurements.

Efficient Time Resolved Multistage CFD Analysis Applied to Axial Compressors

Unsteady computations are necessary if blade row interactions effects are relevant, for example for detailed optimization of a compressor at off-design conditions towards the aerodynamic stability limit, or for structural mechanical tuning of the blades. Modeling time accurate transient multistage flow is expensive both in terms of computer time and memory. Recently the implicit time-resolved Time Transformation method (based on Giles’ time inclining) has been shown to be computationally efficient and a good alternative for modeling transient flow in a single stage (one pitch ratio) turbomachinery configuration. A further advantage of this time resolved method is its ability to capture not only blade passing frequencies but also self-excited frequencies such as in wakes and tip vortex shedding.In this work, an extension of the Time Transformation method (TT) to multistage modeling has been employed to assess the method’s ability in predicting modern multistage compressor performance speedline curve, as well as its ability in capturing dominant machine frequencies. The multistage TT method is verified on a two and a half stage modified Hannover compressor, followed by an industrial validation on a Siemens Energy half scale six stage axial compressor based on the last stages of the Siemens Platform Compressor (PCO). Reference transient solutions on reduced portions of the compressor and/or modified blade count solutions are obtained and compared directly to single passage multistage Time Transformation predictions for the Hannover compressor. The method is then applied directly to the full six stage Siemens compressor employing the true blade counts for all six stages.The first goal of this work is to investigate the ability and accuracy of the multistage TT method to capture all relevant blades passing frequencies, including the impact of different degrees of pitch change between components. The second goal of this work is to explore how best to apply the method for the prediction of a compressor map, up to the surge line. Solutions are compared to experimental test rig data. Physical explanations of the key flow features observed in the experiment, as well as of the differences between the predictions and experimental data, are given.Copyright

Investigation of Time/Frequency Domain CFD Methods to Predict Turbomachinery Blade Aerodynamic Damping

Accurate and efficient prediction of blade aerodynamic damping is critical for the design of turbomachines such as gas and steam turbines. Traditional unsteady time-marching CFD methods used in aerodynamic damping calculations are expensive because they require simulation of many or all blade passages in a given blade-row. In order to reduce computational cost, one can use a pitch-change method and reduce the problem to a small sector of the geometry (one or two blades). Even still, the time-marching method is expensive as many vibration cycles must be simulated to reach a quasi-steady periodic state. To further reduce computational cost, a time/frequency solution method is required.This paper uses an implicit pressure-based time/frequency solution method in combination with a Fourier transformation (FT) pitch-change method and validates its implementation in ANSYS CFX solver. Three cases are investigated, including Standard Configuration 11 (subsonic and transonic), and NASA Rotor 67 transonic fan. Predictions of unsteady pressure coefficient are compared against the experimental data and reference full wheel simulations, over a range of nodal diameters. Computational resources (CPU time) required by the time/frequency method are compared to time transient simulations and discussed in detail.© 2016 ASME

CFD Modeling of Low Pressure Steam Turbine Radial Diffuser Flow by Using a Novel Multiple Mixing Plane Based Coupling: Simulation and Validation

The diffuser and exhaust of low pressure steam turbines shows significant impact on the overall turbine performance. The amount of recovered enthalpy leads to a considerable increase of the turbine power output, and therefore a continuous focus of turbine manufacturers is put on this component. On the one hand, the abilities to aerodynamically design such components is improved, but on the other hand a huge effort is required to properly predict the resulting performance and to enable an accurate modeling of the overall steam turbine and therewith plant heat rate. A wide range of approaches is used to compute the diffuser and exhaust flow, with a wide range of quality. Today it is well known and understood, that there is a strong interaction of rear stage and diffuser flow, and the accuracy of the overall diffuser performance prediction strongly depends on a proper coupling of both domains. The most accurate, but also most expensive method is currently seen in a full annulus and transient coupling. However, for a standard industrial application of diffuser design in a standard development schedule, such a coupling is not feasible and more simplified methods have to be developed.The paper below presents a CFD modeling of low pressure steam turbine diffusers and exhausts based on a direct coupling of the rear stage and diffuser using a novel multiple mixing plane. It is shown that the approach enables a fast diffuser design process and is still able to accurately predict the flow field and hence the exhaust performance. The method is validated against several turbine designs measured in a scaled low pressure turbine model test rig using steam. The results show a very good agreement of the presented CFD modeling against the measurements.Copyright

Time Transformation Simulation of 1.5 Stage Transonic Compressor

The accurate prediction of the aerodynamic and aeromechanical performance in a modern transonic compressor often exceeds the capability of traditional steady state mixing plane simulation methods. Time accurate transient blade row simulation approaches are required when there is a close coupling of the flow between the blade rows, and for fundamentally transient flow phenomena such as aeromechanical analysis including blade flutter and forced response, aerothermodynamic analysis and aero-acoustic analysis.Transient blade row simulations can be computationally impractical when all of the blade passages must be modeled to account for the unequal pitch between the blade rows. Most turbomachines consist of multiple stages, further exacerbating the computational challenge. In order to reduce the computational cost, time accurate pitch-change methods are utilized so that only a sector of the turbomachine (one or few passages per row) is modeled. The extension of the time-transformation pitch-change method to multistage machines has recently shown good promise in predicting both aerodynamic performance and resolving dominant blade passing frequencies for a subsonic compressor, while keeping the computational cost affordable.In this work, a modified one and a half stage Purdue transonic compressor (modified for unequal pitch for all three blade rows) is examined. The goal is to assess the ability of the multistage time-transformation method to accurately predict the aerodynamic performance and transient flow details in the presence of transonic blade row interactions. The results from the multistage time-transformation simulation are compared in detail with a transient full-wheel simulation, a profile transformation simulation, as well as to a steady-state mixing-plane model. Flow details are examined including an FFT analysis of select signals, and the onset of stall is compared between all methods. The relative computational effort is compared between all of the analysis methods.Copyright