I am currently a Director & Co-Founder at Turbostream Ltd, where I lead an outstanding team of engineers applying advanced simulation technologies to aerothermal problems in turbomachinery. I am the main developer of the Turbostream flow solver that underpins our work.
Prior to founding Turbostream Ltd, I was a Peterhouse Research Fellow at the Whittle Laboratory. I have a PhD from the University of Cambridge under the supervision of Graham Pullan.
Turbostream started as a final year undergraduate project during my MEng degree. My supervisor Graham Pullan had the idea that it might be possible to port a simple 2D Euler solver written in Fortran to run on GPUs. This was during the early days of GPGPU computing when doing HPC on GPUs was not straightforward. Thankfully, there were already frameworks such as BrookGPU that simplified both texture management and shader generation with a stream computing model.
Our first results were published in the IMechE journal in 2007 (running on an ATI X1950). The emergence of CUDA made it possible to extend this solver to 3D flows and results obtained on an NVIDIA 8800GTX were published at AIAA 2008. The first complete version of the solver suitable for turbomachinery designers (based on the TBLOCK code from John Denton and using multi-block structured grids) was presented at the ASME Turbo Expo in 2009, now running on a small cluster with 16 GPUs.
The latest version of Turbostream is today used by many organisations in industry and academia to predict the aerothermal behaviour of turbomachinery. The solver supports unstructured grids, steady and unsteady flows, multiple turbulence models, conjugate heat transfer, fluid-structure interaction, relative blade motion and real gases. The most prominent application of Turbostream in the literature is the use of unsteady simulations to predict and explain the onset of rotating instabilities in both axial and centrifugal compressors (see Pullan et al. and Cao et al. for examples).
A key feature of Turbostream is its use of automatic source-to-source compilation to transform high-level Python definitions of unstructured stencil operations into efficient implementations for modern GPUs (using CUDA) or CPUs (using ISPC). Coupled with state-of-the-art linear solvers (such as AMGX), this implementation approach allows us to perform simulations of multi-stage compressors and turbines with 500+ million cells as a matter of routine.
E. J. Gunn, T. Brandvik, M. J. Wilson and R. Maxwell
ASME Turbo Expo 2021
DOI
Abstract: This paper considers the impact of a damaged leading edge on the stall margin and stall inception mechanisms of a transonic, low pressure ratio fan. The damage takes the form of a squared-off leading edge over the upper half of the blade.
Full-annulus, unsteady CFD simulations are used to explain the stall inception mechanisms for the fan at low- and high-speed operating points. A combination of steady and unsteady simulations show that the fan is predicted to be sensitive to leading edge damage at low speed, but insensitive at high speed. This blind prediction aligns well with rig test data.
The difference in response is shown to be caused by the change between subsonic and supersonic flow regimes at the leading edge. Where the inlet relative flow is subsonic, rotating stall is initiated by growth and propagation of a subsonic leading edge flow separation. This separation is shown to be triggered at higher mass flow rates when the leading edge is damaged, reducing the stable flow range. Where the inlet relative flow is supersonic, the flow undergoes a supersonic expansion around the leading edge, creating a supersonic flow patch terminated by a shock on the suction surface. Rotating stall is triggered by growth of this separation, which is insensitive to leading edge shape. This creates a marked difference in sensitivity to damage at low- and high-speed operating points.
E. J. Gunn, T. Brandvik and M. J. Wilson
ASME Turbo Expo 2021
DOI
Abstract: The current trend in civil engine fans towards lower pressure ratio and larger diameter is accompanied by a need to shorten the engine intake length to reduce weight and drag. This paper uses full-annulus, unsteady CFD simulations of two coupled fan-intake configurations to explain the impact of flow field coupling and intake length on fan and intake performance. On-design and off-design operating points are considered at cruise and high angle of attack, respectively.
The fan efficiency at cruise is shown to be determined by a trade-off between two effects. Cruise efficiency is reduced by 0.11% with a short intake due to increased potential flow field distortion, which alters the incidence and diffusion of the rotor. This is partially offset by a reduction in casing boundary layer thickness due to lower intake wetted area.
At high angle of attack conditions, a short intake leads to increased potential flow field distortion and an earlier onset of intake flow separation due to a higher adverse pressure gradient approaching the fan. Both effects combine to reduce the fan thrust at such conditions, although the fan is shown to remain stable at attack angles up to 35°. The reduction in performance is shown to be dominated by flow separations in the rotor, which increase in size and severity for a given attack angle as the intake length is decreased. The fan is also shown to have a stronger influence on the form of the intake flow field in a short intake, suggesting that it is necessary to model the fan in the intake design process for a successful design.
T. Cao, T. Kanzaka, L. Xu and T. Brandvik
ASME Turbo Expo 2019
DOI
Abstract: In this paper, an unsteady tip leakage flow phenomenon is identified and investigated in a centrifugal compressor with a vaneless diffuser at near-stall conditions. This phenomenon is associated with the inception of a rotating instability in the compressor. The study is based on numerical simulations which are supported by experimental measurements. The study confirms that the unstable flow is primarily driven by the Kelvin-Helmholtz instability inherent in the shear layer between the main-stream flow and the tip leakage flow. The shear layer instability induces large scale vortex shedding which propagates circumferentially, leading to pressure perturbations with short wavelength and high amplitude which rotates at about half of the blade speed. The shed vortex is also found to interact with the main blade leading edge, resulting in the reduction of the blade loading identified in the experiment. The paper also reveals that the downstream volute imposes a long wavelength circumferential non-uniform back pressure at the impeller exit, which is found to either excite or supress the vortex shedding within the blade passage depending on the relative circumferential position of the impeller and the volutes.
N. Sohoni, C. Hall, T. Brandvik and A. B. Parry
ASME Turbo Expo 2015
DOI
Abstract: In this paper we present experimental measurements from high frequency unsteady pressure transducers on the blade surfaces of a counter-rotating open rotor and results obtained with full-annulus URANS simulations. A quantitative metric is introduced which can be used to evaluate the quality of the transmission of the front rotor flow features through the interpolating interface between the rotors. Comparisons are made for the open rotor operating at two angles of attack: 0° and 12°. The former setting is a datum case for "uninstalled" operation; the latter setting represents "installed" operation at an extreme angle of attack (AoA). Operation at 12° AoA causes the front rotor wake to vary around the annulus, and leads to shaft-order side-bands for interaction tones.
C. Hall, A. Zachariadis, T. Brandvi and N. Sohoni
The Aeronautical Journal, 2014
DOI
Abstract: A key challenge in open rotor design is getting the optimum aerodynamics at both the cruise and take-off conditions. This is particularly difficult because the operation and the requirements of an open rotor are very different at cruise compared to takeoff. This paper uses CFD results to explore the impact of various design changes on the cruise and take-off flow-fields. The paper then considers how a given open rotor design is best operated at take-off to minimise noise whilst maintaining high thrust. The main findings are that various design modifications can be applied to control the flow features that lead to lost efficiency at cruise and increased noise emission at take-off.
T. Brandvik, C. Hall and A. B. Parry
ASME Turbo Expo 2011
DOI
Abstract: Due to their potential for significant fuel consumption savings, Counter-Rotating Open Rotors (CRORs) are currently being considered as an alternative to high-bypass turbofans. When CRORs are mounted on an aircraft, several ‘installation effects’ arise which are not present when the engine is operated in isolation. This paper investigates how flow features arising from one such effect — the angle-of-attack of the engine centre-line relative to the oncoming flow — can influence the design of CROR engines. Three-dimensional full-annulus unsteady CFD simulations are used to predict the time-varying flow field experienced by each rotor and emphasis is put on the interaction of the front-rotor wake and tip vortex with the rear-rotor.
T. Brandvik and G. Pullan
CIT-2010
DOI
Abstract: We present a new software framework for the implementation of applications that use stencil computations on block-structured grids to solve partial differential equations. A key feature of the framework is the extensive use of automatic source code generation which is used to achieve high performance on a range of leading multi-core processors. Results are presented for a simple model stencil running on Intel and AMD CPUs as well as the NVIDIA GT200 GPU. The generality of the frame- work is demonstrated through the implementation of a complete application consisting of many different stencil computations, taken from the field of computational fluid dynamics.
T. Brandvik and G. Pullan
ASME Turbo Expo 2009
DOI
Abstract: A new three-dimensional Navier–Stokes solver for flows in turbomachines has been developed. The new solver is based on the latest version of the Denton codes but has been implemented to run on graphics processing units (GPUs) instead of the traditional central processing unit. The change in processor enables an order-of-magnitude reduction in run-time due to the higher performance of the GPU. The scaling results for a 16 node GPU cluster are also presented, showing almost linear scaling for typical turbomachinery cases. For validation purposes, a test case consisting of a three-stage turbine with complete hub and casing leakage paths is described. Good agreement is obtained with previously published experimental results. The simulation runs in less than 10 minutes on a cluster with four GPUs.
T. Brandvik and G. Pullan
AIAA 2008
DOI
Abstract: The porting of two- and three-dimensional Euler solvers from a conventional CPU implementation to the novel target platform of the Graphics Processing Unit (GPU) is described. The motivation for such an effort is the impressive performance that GPUs offer: typically 10 times more floating point operations per second than a modern CPU, with over 100 processing cores and all at a very modest financial cost. Both codes were found to generate the same results on the GPU as the FORTRAN versions did on the CPU. The 2D solver ran up to 29 times quicker on the GPU than on the CPU; the 3D solver 16 times faster.
T. Brandvik and G. Pullan
IMechE Journal, 2007
DOI
Abstract: The implementation of a two-dimensional Euler solver on graphics hardware is described. The graphics processing unit is highly parallelized and uses a programming model that is well suited to flow computation. Results for a transonic turbine cascade test-case are presented. For large grids (10^6 nodes) a 40 times speed-up compared with a Fortran implementation on a contemporary CPU is observed.