Computational Modelling of Rotating Cavity Flows
Lead Research Organisation:
University of Oxford
Department Name: Engineering Science
Abstract
This project falls within the EPSRC fluid dynamics and aerodynamics research area.
Rotating cavities are a feature of many parts of a gas turbine internal air system: they can be used to model compressor inter-disk and turbine rotor-stator cavities, and turbine rim seals. The internal air system is the part of the gas turbine that conveys cold compressor bleed gas to the turbine blades and vanes so that they stay at a temperature where their integrity is not compromised. In modern gas turbines the internal air system mass flow makes up a significant minority of the total gas turbine mass flow rate. Heat transfer inside these rotating cavities plays an important role in gas turbine internal air systems. In the compressor inter-disk cavities (the focus of this project) the heat transfer within these cavities plays a large role in the thermal growth of the disks: this in turn determines the blade tip clearance.
As higher bypass ratios, geared turbofans, and optimised short haul aircraft are developed by engine and airframe manufacturers to increase engine efficiency and reduce emissions, it is likely that the core size for a turbofan engine will decrease. Further into the future, technologies such as embedded and distributed engines and open rotor architectures are also likely to require smaller core sizes in comparison to modern engines. As the core size of an engine shrinks, the tip clearance makes up a larger proportion of the main gas path, and the efficiency of the engine is more sensitive to the thermal expansion of the compressor disks - this means that heat transfer in rotating cavities is a topic that will become more and more relevant and important in the future.
The flow in rotating cavities is three dimensional and inherently unsteady, with a distinctive feature being the formation of a large-scale inertial wave flow structure (in terms of the centrifugal and Coriolis forces, the physics of this flow structure is similar to the polar jet stream in the earth's atmosphere.) Investigating the flow and heat transfer in these cavities is difficult experimentally, which has led to significant interest in using computational fluid dynamics to simulate them. However, the necessary turbulence modelling fidelity is currently unclear, with the lowest modelling fidelities (and cheapest unsteady approach), Unsteady Reynolds Averaged Navier-Stokes (URANS), sometimes giving acceptable results, despite it being a commonly stated opinion in the literature that it is unsuitable. The highest modelling fidelity - Large Eddy Simulation (LES) - is often avoided by researchers as it is very computationally costly to resolve the near-wall flow. To mitigate this, Detached Eddy Simulation (DES) is sometimes used, where the near-wall flow is modelled with URANS, and the bulk flow is modelled with LES. However, there is currently very little work examining the validity of DES-type turbulence approaches in rotating cavity flows. This project addresses this by systematically investigating the suitability of the above approaches, and how they interact with the small- and large-scale flow structures present in rotating cavities.
As rotating cavities are a buoyancy-driven flow, it seems likely that simulating both the unsteady fluid and the solid domain at the same time - Conjugate Heat Transfer (CHT) - is necessary. Unsteady CHT is difficult to do in a consistent manner: the solid and fluid time scales differ by a factor of 10,000, and the solid near-wall mesh needs to be very small to properly resolve the thermal penetration depth of high frequency thermal unsteadiness.
This project will also develop novel multi-scale methodologies to enable unsteady CHT to be conducted in a consistent manner. The time-average state, low-frequency unsteadiness, and high frequency unsteadiness in the solid domain will be considered separately, and each be coupled to the fluid domain.
Rotating cavities are a feature of many parts of a gas turbine internal air system: they can be used to model compressor inter-disk and turbine rotor-stator cavities, and turbine rim seals. The internal air system is the part of the gas turbine that conveys cold compressor bleed gas to the turbine blades and vanes so that they stay at a temperature where their integrity is not compromised. In modern gas turbines the internal air system mass flow makes up a significant minority of the total gas turbine mass flow rate. Heat transfer inside these rotating cavities plays an important role in gas turbine internal air systems. In the compressor inter-disk cavities (the focus of this project) the heat transfer within these cavities plays a large role in the thermal growth of the disks: this in turn determines the blade tip clearance.
As higher bypass ratios, geared turbofans, and optimised short haul aircraft are developed by engine and airframe manufacturers to increase engine efficiency and reduce emissions, it is likely that the core size for a turbofan engine will decrease. Further into the future, technologies such as embedded and distributed engines and open rotor architectures are also likely to require smaller core sizes in comparison to modern engines. As the core size of an engine shrinks, the tip clearance makes up a larger proportion of the main gas path, and the efficiency of the engine is more sensitive to the thermal expansion of the compressor disks - this means that heat transfer in rotating cavities is a topic that will become more and more relevant and important in the future.
The flow in rotating cavities is three dimensional and inherently unsteady, with a distinctive feature being the formation of a large-scale inertial wave flow structure (in terms of the centrifugal and Coriolis forces, the physics of this flow structure is similar to the polar jet stream in the earth's atmosphere.) Investigating the flow and heat transfer in these cavities is difficult experimentally, which has led to significant interest in using computational fluid dynamics to simulate them. However, the necessary turbulence modelling fidelity is currently unclear, with the lowest modelling fidelities (and cheapest unsteady approach), Unsteady Reynolds Averaged Navier-Stokes (URANS), sometimes giving acceptable results, despite it being a commonly stated opinion in the literature that it is unsuitable. The highest modelling fidelity - Large Eddy Simulation (LES) - is often avoided by researchers as it is very computationally costly to resolve the near-wall flow. To mitigate this, Detached Eddy Simulation (DES) is sometimes used, where the near-wall flow is modelled with URANS, and the bulk flow is modelled with LES. However, there is currently very little work examining the validity of DES-type turbulence approaches in rotating cavity flows. This project addresses this by systematically investigating the suitability of the above approaches, and how they interact with the small- and large-scale flow structures present in rotating cavities.
As rotating cavities are a buoyancy-driven flow, it seems likely that simulating both the unsteady fluid and the solid domain at the same time - Conjugate Heat Transfer (CHT) - is necessary. Unsteady CHT is difficult to do in a consistent manner: the solid and fluid time scales differ by a factor of 10,000, and the solid near-wall mesh needs to be very small to properly resolve the thermal penetration depth of high frequency thermal unsteadiness.
This project will also develop novel multi-scale methodologies to enable unsteady CHT to be conducted in a consistent manner. The time-average state, low-frequency unsteadiness, and high frequency unsteadiness in the solid domain will be considered separately, and each be coupled to the fluid domain.
Organisations
People |
ORCID iD |
| L He (Primary Supervisor) | |
| Tom Hickling (Student) |
Publications
Hickling T
(2023)
A multiscale framework for unsteady conjugate heat transfer with turbulence resolving methods - With application to rotating cavities
in International Journal of Heat and Fluid Flow
Hickling T
(2021)
Some Observations on the Computational Sensitivity of Rotating Cavity Flows
in Journal of Engineering for Gas Turbines and Power
Hickling T
(2023)
LES-CHT for a Rotating Cavity With Axial Throughflow
in Journal of Turbomachinery
| Description | Understanding flow and heat transfer in compressor disk rotating cavities is important for designing more efficient gas turbine compressors. We simulated rotating cavities using three different turbulence modelling approaches: unsteady Reynolds-averaged Navier-Stokes (URANS), large eddy simulation (LES), and detached eddy simulation (DES). URANS is the cheapest, and attempts to model all turbulence while LES aims to simulate the most important (large) turbulence. DES combines URANS and LES in different regions of the flow to reduce the size and cost of the simulation. There is no consensus in the community which of these is necessary to simulate rotating cavities. We showed that, while URANS sometimes can give good results, it does this for the wrong reasons. We also found a key problem with the use of DES in rotating cavities: URANS regions mix with LES regions, resulting in turbulence being accounted for twice, corrupting the solution. We also found that how we specified the thermal behaviour of the cavity walls had an effect on the aerodynamics of the flow inside the cavity. In th future, we intend to develop methods that recreate the thermal behaviour of the cavity walls more accurately to remove this source of uncertainty. |
| Exploitation Route | The most significant way that the outcome of this funding (so far) may be used by other researchers (in academia and industry) is as a guide to how to model turbulence when they are simulating rotating cavities. In particular, it provides a justification for using more expensive approaches like LES, and warns researchers away from using DES unless they have a way of verifying that it is not adversely affecting the solution. |
| Sectors | Aerospace, Defence and Marine |