Combustor Design Optimisation Using Co-Kriging of Steady and Unsteady Turbulent Combustion

2011 ◽  
Author(s):  
Moresh J. Wankhede ◽  
Neil W. Bressloff ◽  
Andy J. Keane
Keyword(s):  
Navier Stokes ◽  
Design Stage ◽  
Reacting Flow ◽  
High Fidelity ◽  
Design Optimisation ◽  
Design Problems ◽  
Function Calls ◽  
Unsteady Rans ◽  
Computational Budget ◽  
High Level

In the gas turbine industry, computational fluid dynamics (CFD) simulations are often used to predict and visualize the complex reacting flow dynamics, combustion environment and emissions performance of a combustor at the design stage. Given the complexity involved in obtaining accurate flow predictions and due to the expensive nature of simulations, conventional techniques for CFD based combustor design optimisation are often ruled out, primarily due to the limits on available computing resources and time. The design optimisation process normally requires a large number of analyses of the objective and constraint functions which necessitates a careful selection of fast, reliable and efficient computational methods for the CFD analysis and the optimization process. In this study, given a fixed computational budget, an assessment of a co-Kriging based optimisation strategy against a standard Kriging based optimisation strategy is presented for the design of a 2D combustor using steady and unsteady Reynolds-averaged Navier Stokes (RANS) formulation. Within the fixed computational budget, using a steady RANS formulation, the Kriging strategy successfully captures the underlying response, however with unsteady RANS the Kriging strategy fails to capture the underlying response due to the existence of a high level of noise. The co-Kriging strategy is then applied to two design problems, one using two levels of grid resolutions in a steady RANS formulation and the other using steady and unsteady RANS formulations on the same grid resolution. With the co-Kriging strategy, the multi-fidelity analysis is expected to find an optimum design in comparatively less time than that required using the high-fidelity model alone since less high-fidelity function calls should be required. However, using the applied computational setup for co-Kriging, the Kriging strategy beats the co-Kriging strategy under the steady RANS formulation whereas under the unsteady RANS formulation, the high level of noise stalls the co-Kriging optimisation process.

Combustor Design Optimization Using Co-Kriging of Steady and Unsteady Turbulent Combustion

2011 ◽  
Vol 133 (12) ◽  
Author(s):  
Moresh J. Wankhede ◽  
Neil W. Bressloff ◽  
Andy J. Keane
Keyword(s):  
Design Optimization ◽  
Optimization Process ◽  
Design Stage ◽  
Reacting Flow ◽  
High Fidelity ◽  
Optimization Strategy ◽  
Function Calls ◽  
Unsteady Rans ◽  
Computational Budget ◽  
High Level

In the gas turbine industry, computational fluid dynamics (CFD) simulations are often used to predict and visualize the complex reacting flow dynamics, combustion environment and emissions performance of a combustor at the design stage. Given the complexity involved in obtaining accurate flow predictions and due to the expensive nature of simulations, conventional techniques for CFD based combustor design optimization are often ruled out, primarily due to the limits on available computing resources and time. The design optimization process normally requires a large number of analyses of the objective and constraint functions which necessitates a careful selection of fast, reliable and efficient computational methods for the CFD analysis and the optimization process. In this study, given a fixed computational budget, an assessment of a co-Kriging based optimization strategy against a standard Kriging based optimization strategy is presented for the design of a 2D combustor using steady and unsteady Reynolds-averaged Navier Stokes (RANS) formulation. Within the fixed computational budget, using a steady RANS formulation, the Kriging strategy successfully captures the underlying response; however with unsteady RANS the Kriging strategy fails to capture the underlying response due to the existence of a high level of noise. The co-Kriging strategy is then applied to two design problems, one using two levels of grid resolutions in a steady RANS formulation and the other using steady and unsteady RANS formulations on the same grid resolution. With the co-Kriging strategy, the multifidelity analysis is expected to find an optimum design in comparatively less time than that required using the high-fidelity model alone since less high-fidelity function calls should be required. However, using the applied computational setup for co-Kriging, the Kriging strategy beats the co-Kriging strategy under the steady RANS formulation whereas under the unsteady RANS formulation, the high level of noise stalls the co-Kriging optimization process.

High-Fidelity Numerical Analysis of Per-Rev-Type Inlet Distortion Transfer in Multistage Fans—Part II: Entire Component Simulation and Investigation

2010 ◽  
Vol 132 (4) ◽  
Author(s):  
Jixian Yao ◽  
Steven E. Gorrell ◽  
Aspi R. Wadia
Keyword(s):  
Total Pressure ◽  
Navier Stokes ◽  
High Fidelity ◽  
Inlet Distortion ◽  
Unsteady Rans ◽  
Total Temperature ◽  
Computational Fluid Dynamics Cfd ◽  
Last Stage ◽  
Remarkable Agreement ◽  
Engine Development

Part I of this paper validated the ability of the unsteady Reynolds-Averaged Navier-Stokes (RANS) solver PTURBO to accurately simulate distortion transfer and generation through selected blade rows of two multistage fans. In this part, unsteady RANS calculations were successfully applied to predict the 1/rev inlet total pressure distortion transfer in the entirety of two differently designed multistage fans. This paper demonstrates that high-fidelity computational fluid dynamics (CFD) can be used early in the design process for verification purposes before hardware is built and can be used to reduce the number of distortion tests, hence reducing engine development cost. The unsteady RANS code PTURBO demonstrated remarkable agreement with the data, accurately capturing both the magnitude and the profile of total pressure and total temperature measurements. Detailed analysis of the flow physics identified from the CFD results has led to a thorough understanding of the total temperature distortion generation and transfer mechanism, especially for the spatial phase difference of total pressure and total temperature profiles. The analysis illustrates that the static parameters are more revealing than their stagnation counterpart and that pressure and temperature rise are more revealing while the pressure and temperature ratio could be misleading. The last stage is effectively throttled by the inlet distortion even though the overall engine throttle remains unchanged. The total temperature distortion generally grows as flow passes through the fan stages.

Multi-Objective Design Problem of Tire Wear and Visualization of Its Pareto Solutions2

2006 ◽  
Vol 34 (3) ◽  
pp. 170-194 ◽  
Author(s):  
M. Koishi ◽  
Z. Shida
Keyword(s):  
Inverse Problems ◽  
Conceptual Design ◽  
Dimensional Space ◽  
Design Stage ◽  
Objective Functions ◽  
Pareto Solutions ◽  
Design Problems ◽  
Multi Objective ◽  
Design Engineers ◽  
Design Variables

Abstract Since tires carry out many functions and many of them have tradeoffs, it is important to find the combination of design variables that satisfy well-balanced performance in conceptual design stage. To find a good design of tires is to solve the multi-objective design problems, i.e., inverse problems. However, due to the lack of suitable solution techniques, such problems are converted into a single-objective optimization problem before being solved. Therefore, it is difficult to find the Pareto solutions of multi-objective design problems of tires. Recently, multi-objective evolutionary algorithms have become popular in many fields to find the Pareto solutions. In this paper, we propose a design procedure to solve multi-objective design problems as the comprehensive solver of inverse problems. At first, a multi-objective genetic algorithm (MOGA) is employed to find the Pareto solutions of tire performance, which are in multi-dimensional space of objective functions. Response surface method is also used to evaluate objective functions in the optimization process and can reduce CPU time dramatically. In addition, a self-organizing map (SOM) proposed by Kohonen is used to map Pareto solutions from high-dimensional objective space onto two-dimensional space. Using SOM, design engineers see easily the Pareto solutions of tire performance and can find suitable design plans. The SOM can be considered as an inverse function that defines the relation between Pareto solutions and design variables. To demonstrate the procedure, tire tread design is conducted. The objective of design is to improve uneven wear and wear life for both the front tire and the rear tire of a passenger car. Wear performance is evaluated by finite element analysis (FEA). Response surface is obtained by the design of experiments and FEA. Using both MOGA and SOM, we obtain a map of Pareto solutions. We can find suitable design plans that satisfy well-balanced performance on the map called “multi-performance map.” It helps tire design engineers to make their decision in conceptual design stage.

scholarly journals Developing a BIM-Based MUVR Treadmill System for Architectural Design Review and Collaboration

2021 ◽  
Vol 11 (15) ◽  
pp. 6881
Author(s):  
Calvin Chung Wai Keung ◽  
Jung In Kim ◽  
Qiao Min Ong
Keyword(s):  
Architectural Design ◽  
Human Locomotion ◽  
Virtual Space ◽  
Design Stage ◽  
Design Review ◽  
Team Members ◽  
Physical Constraints ◽  
Simulated Environment ◽  
Set Up ◽  
High Level

Virtual reality (VR) is quickly becoming the medium of choice for various architecture, engineering, and construction applications, such as design visualization, construction planning, and safety training. In particular, this technology offers an immersive experience to enhance the way architects review their design with team members. Traditionally, VR has used a desktop PC or workstation setup inside a room, yielding the risk of two users bump into each other while using multiuser VR (MUVR) applications. MUVR offers shared experiences that disrupt the conventional single-user VR setup, where multiple users can communicate and interact in the same virtual space, providing more realistic scenarios for architects in the design stage. However, this shared virtual environment introduces challenges regarding limited human locomotion and interactions, due to physical constraints of normal room spaces. This study thus presented a system framework that integrates MUVR applications into omnidirectional treadmills. The treadmills allow users an immersive walking experience in the simulated environment, without space constraints or hurt potentialities. A prototype was set up and tested in several scenarios by practitioners and students. The validated MUVR treadmill system aims to promote high-level immersion in architectural design review and collaboration.

scholarly journals TRACING THE EMERGENCE OF DESIGN PROBLEMS AND THEIR IMPACTS ON THE COMPLEXITY OF ENGINEERING SOLUTIONS

2021 ◽  
Vol 1 ◽  
pp. 3229-3238
Author(s):  
Torben Beernaert ◽  
Pascal Etman ◽  
Maarten De Bock ◽  
Ivo Classen ◽  
Marco De Baar
Keyword(s):  
Large Scale ◽  
Fusion Reactor ◽  
Design Stage ◽  
Engineering Model ◽  
Design Problems ◽  
Nuclear Fusion Reactor ◽  
Early Design Stage ◽  
Iterative Design ◽  
Problems And Solutions ◽  
Emergent Design

AbstractThe design of ITER, a large-scale nuclear fusion reactor, is intertwined with profound research and development efforts. Tough problems call for novel solutions, but the low maturity of those solutions can lead to unexpected problems. If designers keep solving such emergent problems in iterative design cycles, the complexity of the resulting design is bound to increase. Instead, we want to show designers the sources of emergent design problems, so they may be dealt with more effectively. We propose to model the interplay between multiple problems and solutions in a problem network. Each problem and solution is then connected to a dynamically changing engineering model, a graph of physical components. By analysing the problem network and the engineering model, we can (1) derive which problem has emerged from which solution and (2) compute the contribution of each design effort to the complexity of the evolving engineering model. The method is demonstrated for a sequence of problems and solutions that characterized the early design stage of an optical subsystem of ITER.

scholarly journals Numerical simulation and experimental validation of cavitating flow in a multi-hole diesel fuel injector

2021 ◽  
pp. 146808742199863
Author(s):  
Aishvarya Kumar ◽  
Ali Ghobadian ◽  
Jamshid Nouri
Keyword(s):  
Computational Cost ◽  
Cavitating Flow ◽  
Field Analysis ◽  
Navier Stokes ◽  
Fuel Injector ◽  
Fluid Behavior ◽  
Flow Field Analysis ◽  
Charged Couple Device ◽  
Biodiesel Fuels ◽  
High Level

This study assesses the predictive capability of the ZGB (Zwart-Gerber-Belamri) cavitation model with the RANS (Reynolds Averaged Navier-Stokes), the realizable k-epsilon turbulence model, and compressibility of gas/liquid models for cavitation simulation in a multi-hole fuel injector at different cavitation numbers (CN) for diesel and biodiesel fuels. The prediction results were assessed quantitatively by comparison of predicted velocity profiles with those of measured LDV (Laser Doppler Velocimetry) data. Subsequently, predictions were assessed qualitatively by visual comparison of the predicted void fraction with experimental CCD (Charged Couple Device) recorded images. Both comparisons showed that the model could predict fluid behavior in such a condition with a high level of confidence. Additionally, flow field analysis of numerical results showed the formation of vortices in the injector sac volume. The analysis showed two main types of vortex structures formed. The first kind appeared connecting two adjacent holes and is known as “hole-to-hole” connecting vortices. The second type structure appeared as double “counter-rotating” vortices emerging from the needle wall and entering the injector hole facing it. The use of RANS proved to save significant computational cost and time in predicting the cavitating flow with good accuracy.

Transfer Reinforcement Learning for Autonomous Driving

2021 ◽  
Vol 31 (3) ◽  
pp. 1-26
Author(s):  
Aravind Balakrishnan ◽  
Jaeyoung Lee ◽  
Ashish Gaurav ◽  
Krzysztof Czarnecki ◽  
Sean Sedwards
Keyword(s):  
Decision Making ◽  
Reinforcement Learning ◽  
Transfer Problem ◽  
Autonomous Driving ◽  
High Fidelity ◽  
Rule Based ◽  
High Level ◽  
Real Vehicle

Reinforcement learning (RL) is an attractive way to implement high-level decision-making policies for autonomous driving, but learning directly from a real vehicle or a high-fidelity simulator is variously infeasible. We therefore consider the problem of transfer reinforcement learning and study how a policy learned in a simple environment using WiseMove can be transferred to our high-fidelity simulator, W ise M ove . WiseMove is a framework to study safety and other aspects of RL for autonomous driving. W ise M ove accurately reproduces the dynamics and software stack of our real vehicle. We find that the accurately modelled perception errors in W ise M ove contribute the most to the transfer problem. These errors, when even naively modelled in WiseMove , provide an RL policy that performs better in W ise M ove than a hand-crafted rule-based policy. Applying domain randomization to the environment in WiseMove yields an even better policy. The final RL policy reduces the failures due to perception errors from 10% to 2.75%. We also observe that the RL policy has significantly less reliance on velocity compared to the rule-based policy, having learned that its measurement is unreliable.

Automatic Three-Dimensional Optimisation of a Modern Tandem Compressor Vane

2014 ◽  
Author(s):  
R. C. Schlaps ◽  
S. Shahpar ◽  
V. Gümmer
Keyword(s):  
Pressure Ratio ◽  
Three Dimensional ◽  
Trust Region ◽  
Automatic Generation ◽  
Navier Stokes ◽  
Design Parameters ◽  
High Fidelity ◽  
Stall Margin ◽  
Design Choice ◽  
Multipoint Approximation

In order to increase the performance of a modern gas turbine, compressors are required to provide higher pressure ratio and avoid incurring higher losses. The tandem aerofoil has the potential to achieve a higher blade loading in combination with lower losses compared to single vanes. The main reason for this is due to the fact that a new boundary layer is generated on the second blade surface and the turning can be achieved with smaller separation occurring. The lift split between the two vanes with respect to the overall turning is an important design choice. In this paper an automated three-dimensional optimisation of a highly loaded compressor stator is presented. For optimisation a novel methodology based on the Multipoint Approximation Method (MAM) is used. MAM makes use of an automatic design of experiments, response surface modelling and a trust region to represent the design space. The CFD solutions are obtained with the high-fidelity 3D Navier-Stokes solver HYDRA. In order to increase the stage performance the 3D shape of the tandem vane is modified changing both the front and rear aerofoils. Moreover the relative location of the two aerofoils is controlled modifying the axial and tangential relative positions. It is shown that the novel optimisation methodology is able to cope with a large number of design parameters and produce designs which performs better than its single vane counterpart in terms of efficiency and numerical stall margin. One of the key challenges in producing an automatic optimisation process has been the automatic generation of high-fidelity computational meshes. The multi block-structured, high-fidelity meshing tool PADRAM is enhanced to cope with the tandem blade topologies. The wakes of each aerofoil is properly resolved and the interaction and the mixing of the front aerofoil wake and the second tandem vane are adequately resolved.

Fully Parametric High-Fidelity CFD Model for the Design Optimisation of the Cyclic Stagger Pattern of a Set of Fan Outlet Guide Vanes

2009 ◽  
Author(s):  
Andrea Milli ◽  
Olivier Bron
Keyword(s):  
Complex Geometry ◽  
Single Point ◽  
Sensitivity Study ◽  
Operating Conditions ◽  
High Fidelity ◽  
Cfd Model ◽  
Design Optimisation ◽  
Guide Vanes ◽  
High Loss ◽  
Design Variables

The present paper deals with the redesign of cyclic variation of a set of fan outlet guide vanes by means of high-fidelity full-annulus CFD. The necessity for the aerodynamic redesign originated from a change to the original project requirement, when the customer requested an increase in specific thrust above the original engine specification. The main objectives of this paper are: 1) make use of 3D CFD simulations to accurately model the flow field and identify high-loss regions; 2) elaborate an effective optimisation strategy using engineering judgement in order to define realistic objectives, constraints and design variables; 3) emphasise the importance of parametric geometry modelling and meshing for automatic design optimisation of complex turbomachinery configurations; 4) illustrate that the combination of advanced optimisation algorithms and aerodynamic expertise can lead to successful optimisations of complex turbomachinery components within practical time and costs constrains. The current design optimisation exercise was carried out using an in-house set of software tools to mesh, resolve, analyse and optimise turbomachinery components by means of Reynolds-averaged Navier-Stokes simulations. The original configuration was analysed using the 3D CFD model and thereafter assessed against experimental data and flow visualisations. The main objective of this phase was to acquire a deep insight of the aerodynamics and the loss mechanisms. This was important to appropriately limit the design scope and to drive the optimisation in the desirable direction with a limited number of design variables. A mesh sensitivity study was performed in order to minimise computational costs. Partially converged CFD solutions with restart and response surface models were used to speed up the optimisation loop. Finally, the single-point optimised circumferential stagger pattern was manually adjusted to increase the robustness of the design at other flight operating conditions. Overall, the optimisation resulted in a major loss reduction and increased operating range. Most important, it provided the project with an alternative and improved design within the time schedule requested and demonstrated that CFD tools can be used effectively not only for the analysis but also to provide new design solutions as a matter of routine even for very complex geometry configurations.

Proving the Labeled Interval Calculus for Inferences on Catalogs

1990 ◽  
Author(s):  
Walid Habib ◽  
Allen C. Ward
Keyword(s):  
Mechanical Design ◽  
Formal System ◽  
Constraint Propagation ◽  
Operating Conditions ◽  
Manufacturing Processes ◽  
Entire System ◽  
High Level Language ◽  
Design Problems ◽  
High Level ◽  
Interval Constraint

Abstract The “labeled interval calculus” is a formal system that performs quantitative inferences about sets of artifacts under sets of operating conditions. It refines and extends the idea of interval constraint propagation, and has been used as the basis of a program called a “mechanical design compiler,” which provides the user with a “high level language” in which design problems for systems to be built of cataloged components can be quickly and easily formulated. The compiler then selects optimal combinations of catalog numbers. Previous work has tested the calculus empirically, but only parts of the calculus have been proven mathematically. This paper presents a new version of the calculus and shows how to extend the earlier proofs to prove the entire system. It formalizes the effects of toleranced manufacturing processes through the concept of a “selectable subset” of the artifacts under consideration. It demonstrates the utility of distinguishing between statements which are true for all artifacts under consideration, and statements which are merely true for some artifact in each selectable subset.

Sign in / Sign up

Export Citation Format

Share Document