Centrifugal Compressor Aero-Mechanical Design: A Machine Learning Approach

2021 ◽  
Author(s):  
Dario Barsi ◽  
Andrea Perrone ◽  
Luca Ratto ◽  
Gianluca Ricci ◽  
Marco Sanguineti
Keyword(s):  
Machine Learning ◽  
Surrogate Model ◽  
Mechanical Design ◽  
Three Dimensional ◽  
Operating Conditions ◽  
Model Parameters ◽  
Design And Optimization ◽  
Machine Learning Approach ◽  
Computational Fluid Dynamics Cfd ◽  
Geometry Parameterization

Abstract The present paper presents an enhanced method for multi-disciplinary design and optimization of centrifugal compressors based on Machine Learning (ML) algorithms. The typical approach involves the preliminary design, the geometry parameterization, the generation of aero-mechanical databases and a surrogate-model based optimization. This procedure is able to provide excellent results, but it is time consuming and has to be repeated for each new design. The aim of the proposed procedure is to actively exploit the simulations performed in the past for subsequent designs thanks to the predictive capabilities of the ML surrogate model. A commercial 3D (three dimensional) computational fluid dynamics (CFD) solver for the aerodynamic computations and a commercial finite element code for the mechanical integrity calculations, coupled with scripting modules, have been adopted. Two different compressors, with different geometry and operating conditions, have been designed and two aero-mechanical databases have been developed. Then, these two databases have been joined and have been used for the training and validation of the surrogate model. To assess the performance of this approach, two new compressors have been designed, case 1 with operating conditions between those of the databases used for training and validation and case 2 with operating conditions far above. The use of an optimizer coupled to the prediction of the surrogate model has enabled to define the “best set” of model parameters, in compliance with aero-mechanical objectives and constraints. The accuracy of the ML algorithm forecast has been evaluated through CFD and FEM simulations carried out iteratively on the optimal samples, with new simulations added to the database for further training of the surrogate model. The results have been presented with reference to cases 1 and 2 and highlight all the benefits of the proposed approach.

On the Coupling of Inverse Design and Optimization Techniques for Turbomachinery Blade Design

2006 ◽  
Author(s):  
Duccio Bonaiuti ◽  
Mehrdad Zangeneh
Keyword(s):  
Mechanical Design ◽  
Design Method ◽  
Three Dimensional ◽  
Axial Compressor ◽  
Optimization Techniques ◽  
Operating Conditions ◽  
Inverse Design ◽  
Design Parameters ◽  
Optimization Strategy ◽  
Design And Optimization

Optimization strategies have been used in recent years for the aerodynamic and mechanical design of turbomachine components. One crucial aspect in the use of such methodologies is the choice of the geometrical parameterization, which determines the complexity of the objective function to be optimized. In the present paper, an optimization strategy for the aerodynamic design of turbomachines is presented, where the blade parameterization is based on the use of a three-dimensional inverse design method. The blade geometry is described by means of aerodynamic parameters, like the blade loading, which are closely related to the aerodynamic performance to be optimized, thus leading to a simple shape of the optimization function. On the basis of this consideration, it is possible to use simple approximation functions for describing the correlations between the input design parameters and the performance ones. The Response Surface Methodology coupled with the Design of Experiments (DOE) technique was used for this purpose. CFD analyses were run to evaluate the configurations required by the DOE to generate the database. Optimization algorithms were then applied to the approximated functions in order to determine the optimal configuration or the set of optimal ones (Pareto front). The method was applied for the aerodynamic redesign of two different turbomachine components: a centrifugal compressor stage and a single-stage axial compressor. In both cases, both design and off-design operating conditions were analyzed and optimized.

Conditioning Model Ensembles to Various Observed Data (Field and Regional Level) by Applying Machine-Learning-Augmented Workflows to a Mature Field with 70 Years of Production History

2021 ◽  
pp. 1-18
Author(s):  
Gisela Vanegas ◽  
John Nejedlik ◽  
Pascale Neff ◽  
Torsten Clemens
Keyword(s):  
Machine Learning ◽  
Oil Recovery ◽  
Numerical Models ◽  
Operating Conditions ◽  
Model Parameters ◽  
Large Set ◽  
Model Parameter ◽  
Production History ◽  
Hydrocarbon Fields ◽  
Parameter Distributions

Summary Forecasting production from hydrocarbon fields is challenging because of the large number of uncertain model parameters and the multitude of observed data that are measured. The large number of model parameters leads to uncertainty in the production forecast from hydrocarbon fields. Changing operating conditions [e.g., implementation of improved oil recovery or enhanced oil recovery (EOR)] results in model parameters becoming sensitive in the forecast that were not sensitive during the production history. Hence, simulation approaches need to be able to address uncertainty in model parameters as well as conditioning numerical models to a multitude of different observed data. Sampling from distributions of various geological and dynamic parameters allows for the generation of an ensemble of numerical models that could be falsified using principal-component analysis (PCA) for different observed data. If the numerical models are not falsified, machine-learning (ML) approaches can be used to generate a large set of parameter combinations that can be conditioned to the different observed data. The data conditioning is followed by a final step ensuring that parameter interactions are covered. The methodology was applied to a sandstone oil reservoir with more than 70 years of production history containing dozens of wells. The resulting ensemble of numerical models is conditioned to all observed data. Furthermore, the resulting posterior-model parameter distributions are only modified from the prior-model parameter distributions if the observed data are informative for the model parameters. Hence, changes in operating conditions can be forecast under uncertainty, which is essential if nonsensitive parameters in the history are sensitive in the forecast.

Investigation of the Flow Phenomenon Inside Gas Ejectors With Moist Gas Entrainment

2016 ◽  
Vol 9 (1) ◽  
Author(s):  
Yuping Wang ◽  
Mark Pellerin ◽  
Pravansu Mohanty ◽  
Subrata Sengupta
Keyword(s):  
Water Vapor ◽  
Phase Change ◽  
Gas Flow ◽  
Phase Distribution ◽  
Light Attenuation ◽  
Three Dimensional ◽  
Operating Conditions ◽  
Second Phase ◽  
Symmetric Model ◽  
Design And Optimization

This paper focuses on the gas flow study of an ejector used in applications where moist gases are being entrained. Two parts of work are presented. In the first part, characteristics of gas flow inside an ejector, as well as the ejector's performance under various operating and geometric configurations, were studied with a three-dimensional computational model. Measurements were also performed for validation of the model. In the second part, focus was given to the potential condensation or desublimation phenomena that may occur inside an ejector when water vapor is included in the entrained stream. Experiments using light-attenuation method were performed to verify the presence of a second phase; then, the onset of phase change and the phase distribution were obtained numerically. A two-dimensional axis-symmetric model was developed based on the model used in the first part. User-defined functions were used to implement the phase-change criteria and particle prediction. A series of simulations were performed with various amounts of water vapor added into the entrained flow. It was found that both frost particles and water condensate could form inside the mixing tube depending on the operating conditions and water vapor concentrations. When the concentration exceeds 3% by mass, water vapor could condense throughout the mixing tube. Some preliminary results of the second phase particles formed, e.g., critical sizes and distributions, were also obtained to assist with the design and optimization of gas ejectors used in similar applications.

Fixing a Boiler with CFD

1998 ◽  
Vol 120 (04) ◽  
pp. 59-61
Author(s):  
Kevin Parker
Keyword(s):  
Three Dimensional ◽  
Paper Mill ◽  
Operating Conditions ◽  
Clear Understanding ◽  
Post Processing ◽  
Processing Program ◽  
Hidden Line ◽  
Computational Fluid Dynamics Cfd ◽  
Animated Sequence ◽  
Light Shading

This article focuses on carryover at a paper mill that had been solved using computational fluid dynamics (CFD) to visualize flow within the boiler. Technicians had tried adjusting airflow and firing arrangements without success. They turned the problem over to analysts who simulated the airflow within the boiler using CFD. An animated sequence of streamlines showing airflow provided engineers with a clear understanding of exactly what was happening inside the boiler, making it relatively easy to adjust operating conditions and solve the problem. McDermott analysts use FIELDVIEW, a commercial post-processing program from Intelligent Light in Lyndhurst, NJ. With the software, the analyst can create three-dimensional perspective views with hidden-line removal and light shading. She or He can trace the path of a marker traveling along with the fluid through a series of animated views. The analysts made a second FIELDVIEW movie of the airflow conditions with the new arrangement, showing the elimination of the center core. They played the two movies simultaneously on two monitors set side-by-side to demonstrate for the customer’s engineers how the recommended changes would solve the problem.

Plasto-Elastohydrodynamic Lubrication in Point Contacts for Surfaces With Three-Dimensional Sinusoidal Waviness and Real Machined Roughness

2014 ◽  
Vol 136 (3) ◽  
Author(s):  
Tao He ◽  
Ning Ren ◽  
Dong Zhu ◽  
Jiaxu Wang
Keyword(s):  
Plastic Deformation ◽  
Rough Surface ◽  
Mechanical Design ◽  
Three Dimensional ◽  
Elastohydrodynamic Lubrication ◽  
Operating Conditions ◽  
Von Mises Stress ◽  
Asperity Contacts ◽  
Lubrication Characteristics ◽  
Von Mises

Efficiency and durability are among the top concerns in mechanical design to minimize environmental impact and conserve natural resources while fulfilling performance requirements. Today mechanical systems are more compact, lightweight, and transmit more power than ever before, which imposes great challenges to designers. Under the circumstances, some simplified analyses may no longer be satisfactory, and in-depth studies on mixed lubrication characteristics, taking into account the effects of 3D surface roughness and possible plastic deformation, are certainly needed. In this paper, the recently developed plasto-elastohydrodynamic lubrication (PEHL) model is employed, and numerous cases with both sinusoidal waviness and real machined roughness are analyzed. It is observed that plastic deformation may occur due to localized high pressure peaks caused by the rough surface asperity contacts, even though the external load is still considerably below the critical load determined at the onset of plastic deformation in the corresponding smooth surface contact. It is also found, based on a series of cases analyzed, that the roughness height, wavelength, material hardening property, and operating conditions may all have significant influences on the PEHL performance, subsurface von Mises stress field, residual stresses, and plastic strains. Generally, the presence of plastic deformation may significantly reduce some of the pressure spikes and peak values of subsurface stresses and make the load support more evenly distributed among all the rough surface asperities in contact.

CFD Analysis of Batch-Type Reheating Furnace

2007 ◽  
Author(s):  
Bin Wu ◽  
Andrew M. Arnold ◽  
Eugene Arnold ◽  
George Downey ◽  
Chenn Q. Zhou
Keyword(s):  
Three Dimensional ◽  
Furnace Operation ◽  
Operating Conditions ◽  
Temperature Fields ◽  
Cfd Analysis ◽  
Reheating Furnace ◽  
Batch Type ◽  
Billet Quality ◽  
Actual Operating ◽  
Computational Fluid Dynamics Cfd

In the steelmaking industry, reheating furnaces are used to heat the billets or blooms to the rolling temperature; the uniformity of the temperature in the furnace determines billet quality. In order to obtain a better understanding of the furnace operation, which influences the temperature distribution; Computational Fluid Dynamics (CFD) analysis is conducted to examine the transient and three dimensional temperature fields in a reheating furnace using the commercial software Fluent®. A number of actual operating conditions, based on the ArcelorMittal Steelton No.3 reheating furnace, are computed. The numerical results are used to optimize the operating parameters and thus help to improve the steel quality.

Prediction of Raceways in a Blast Furnace

2006 ◽  
Author(s):  
Naresh K. Selvarasu ◽  
D. Huang ◽  
Zumao Chen ◽  
Mingyan Gu ◽  
Yongfu Zhao ◽  
...  
Keyword(s):  
Particle Size ◽  
Blast Furnace ◽  
Three Dimensional ◽  
Operating Conditions ◽  
Gas Oil ◽  
Cfd Model ◽  
Coke Particle ◽  
Fuel Gas ◽  
Computational Fluid Dynamics Cfd ◽  
Insight Into

In a blast furnace, preheated air and fuel (gas, oil or pulverized coal) are often injected into the lower part of the furnace through tuyeres, forming a raceway in which the injected fuel and some of the coke descending from the top of the furnace are combusted and gasified. The shape and size of the raceway greatly affect the combustion of, the coke and the injected fuel in the blast furnace. In this paper, a three-dimensional (3-D) computational fluid dynamics (CFD) model is developed to investigate the raceway evolution. The furnace geometry and operating conditions are based on the Mittal Steel IH7 blast furnace. The effects of Tuyere-velocity, coke particle size and burden properties are computed. It is found that the raceway depth increases with an increase in the tuyere velocity and a decrease in the coke particle size in the active coke zone. The CFD results are validated using experimental correlations and actual observations. The computational results provide useful insight into the raceway formation and the factors that influence its size and shape.

Application of an Improved Streamline Curvature Approach to a Modern, Two-Stage Transonic Fan: Comparison With Data and CFD

2002 ◽  
Author(s):  
Keith M. Boyer ◽  
Walter F. O’Brien
Keyword(s):  
Three Dimensional ◽  
Operating Conditions ◽  
Navier Stokes ◽  
Model Parameters ◽  
Two Stage ◽  
Streamline Curvature ◽  
Wide Range ◽  
Improved Model ◽  
Transonic Fan ◽  
Streamline Curvature Method

A streamline curvature method with improvements to key loss models is applied to a two-stage, low aspect ratio, transonic fan with design tip relative Mach number of approximately 1.65. Central to the improvements is the incorporation of a physics-based shock model. The attempt here is to capture the effects of key flow phenomena relative to the off-design performance of the fan. A quantitative analysis regarding solution sensitivities to model parameters that influence the key phenomena over a wide range of operating conditions is presented. Predictions are compared to performance determined from overall and interstage measurements, as well as from a three-dimensional, steady, Reynolds-averaged Navier-Stokes method applied across the first rotor. Overall and spanwise comparisons demonstrate that the improved model gives reasonable performance trending and generally accurate results. The method can be used to provide boundary conditions to higher-order solvers, or implemented within novel approaches using the streamline curvature method to explore complex engine-inlet integration issues, such as time-variant distortion.

Sign in / Sign up

Export Citation Format

Share Document