AIAA Guide for Verification and Validation of Computational Fluid Dynamics Simulation

Guide AIAA G-077-1998 Guide for the Verification and Validation of Computational Fluid Dynamics Simulations AIAA standards are copyrighted by the American Institute of Aeronautics and Astronautics (AIAA), 1801 Alexander Bell Drive, Reston, VA 20191-4344 USA. All rights reserved. AIAA grants you a license as follows: The right to download an electronic file of this AIAA standard for temporary storage on one computer for purposes of viewing, and/or printing one copy of the AIAA standard for individual use. Neither the electronic file nor the hard copy print may be reproduced in any way. In addition, the electronic file may not be distributed elsewhere over computer networks or otherwise. The hard copy print may only be distributed to other employees for their internal use within your organization. AIAA G-077-1998 Guide Guide for the Verification and Validation of Computational Fluid Dynamics Simulations Sponsor American Institute of Aeronautics and Astronautics Approved January 14, 1998 Abstract This document presents guidelines for assessing the credibility of modeling and simulation in computational fluid dynamics. The two main principles that are necessary for assessing credibility are verification and validation. Verification is the process of determining if a computational simulation accurately represents the conceptual model, but no claim is made of the relationship of the simulation to the real world. Validation is the process of determining if a computational simulation represents the real world. This document defines a number of key terms, discusses fundamental concepts, and specifies general procedures for conducting verification and validation of computational fluid dynamics simulations. The document’s goal is to provide a foundation for the major issues and concepts in verification and validation. However, this document does not recommend standards in these areas because a number of important issues are not yet resolved. It is hoped that the guidelines will aid in the research, development, and use of computational fluid dynamics simulations by establishing common terminology and methodology for verification and validation. The terminology and methodology should also be useful in other engineering and science disciplines. Administrator 打字机 Administrator 打字机 Administrator 打字机 Administrator 打字机 Administrator 打字机 CFD仿真验证及确认 指南 验证指南下载验证指南下载验证指南下载星度指南下载审查指南PDF Administrator 打字机 Administrator 打字机 发起者 Administrator 打字机 美国航空航天协会 Administrator 打字机 批准时间1998年2月14日 Administrator 打字机 摘要 Administrator 注释 本文档提供评估计算流体动力学建模及仿真可信度的指南。目前存在两种主要的评估可信度的方法：verification及validation。Verification主要判定计算仿真表征概念模型的精度，而不会关注模型与真实世界的关系。Validation为评判计算仿真表征真实世界能力的过程。本文档定义了一些关键术语，讨论基本概念，并且指定一些计算流体动力学仿真中Verification及Validation的一般过程。本文档的目标是提供v&v中的主要问题及概念基础。然而，由于一些重要问题未得到解决，因此本文档不建议用于这些方面的标准版。希望本只能将会有助于建立计算流体动力学仿真V&V中的通用术语及方法。这些术语及方法应当对于其他 工程 路基工程安全技术交底工程项目施工成本控制工程量增项单年度零星工程技术标正投影法基本原理 及科学领域同样有用。 AIAA G-077-1998 Library of Congress Cataloging-in-Publication Data AIAA guide for the verification and validation of computational fluid dynamics simulations / sponsor, American Institute of Aeronautics and Astronautics. p. cm. Includes bibliographical references. ISBN 1-56347-285-6 1 . Fluid dynamics--Computer simulation--Evaluation. 2. Fluid dynamics--Mathematical models--Evaluation. I. American Institute of Aeronautics and Astronautics. QA911.A54 1998 620.1 '064 '0113--dc21 9 8 - 8 1 3 8 CIP Published by American Institute of Aeronautics and Astronautics 1801 Alexander Bell Drive, Suite 500, Reston, VA 20191-4344 Copyright © 1998 American Institute of Aeronautics and Astronautics All rights reserved. No part of this publication may be reproduced in any form, in an electronic retrieval system or otherwise, without prior written permission of the publisher. Printed in the United States of America. ii AIAA G-077-1998 Table of Contents Foreword...................................................................................................................... v Executive Summary................................................................................................... vii 1 . Introduction................................................................................................................. 1 1.1 Background..................................................................................................................... 1 1.2 Scope............................................................................................................................. 1 1.3 Outline............................................................................................................................ 2 2 . Concepts and Terminology....................................................................................... 2 2.1 Modeling and Simulation...................................................................................................2 2.2 Verification and Validation................................................................................................. 3 2.3 Uncertainty and Error........................................................................................................ 4 2.4 Prediction and Levels of Credibility.................................................................................... 5 3 . Verification Assessment........................................................................................... 7 3.1 Grid and Time-Step Convergence..................................................................................... 7 3.2 Iterative Convergence and Consistency Tests................................................................... 8 3.3 Highly Accurate Solutions................................................................................................ 9 4 . Validation Assessment.............................................................................................10 4.1 Validation Phases...........................................................................................................11 4.2 Calibration......................................................................................................................13 4.3 Requirements for Experimental Data................................................................................14 5 . Summary and Conclusions...................................................................................... 14 6 . References................................................................................................................ 15 iii AIAA G-077-1998 List of Figures Figure 1 Phases of Modeling and Simulation...................................................................... 3 Figure 2 Verification Process............................................................................................. 7 Figure 3 Validation Process..............................................................................................10 Figure 4 Validation Phases...............................................................................................11 Figure 5 Characteristics of Validation Phases.....................................................................11 Figure 6 Use of Completed Validation Cases for New Applications...................................... 12 iv AIAA G-077-1998 Foreword The American Institute of Aeronautics and Astronautics (AIAA) Standards Program sponsored development of this document, Guide for the Verification and Validation of Computational Fluid Dynamics Simulations. This document originated within the AIAA Computational Fluid Dynamics Committee on Standards, which is composed of AIAA members and others who are not affiliated with AIAA. Committee members come from industry, government, and academia, and serve voluntarily without compensation. This document represents a consensus of the Committee’s opinions on the terminology and methodology for verification and validation of computational fluid dynamics (CFD) simulations. This document is primarily a synthesis of opinions from the published literature on verification and validation in modeling and simulation. Perspectives from a wide variety of sources were assembled in order to develop the most useful, self-consistent, and logical framework. Even though there is a variety of opinion on verification and validation in the literature, there is inc g agreement on the fundamental aspects. It is hoped that this document will promote consensus e major issues among the CFD community at large. The goal of this document is to support researchers, develop common terminology and methodology for verification and validati and methodology should also be useful in other engineering and s The AIAA Standards Procedures provides that all approved standards are advisory only. The use of these publications by a entirely voluntary. There is no agreement to adhere to any AIAA st to conform to or be guided by any standards report. This guide is or accreditation of codes. In formulating, revising, and approv Committees on Standards will not consider patents that may ap users of the publications are responsible for protecting themsel patents, or copyrights, or both. This document is subject to change based on developments i received from readers. Comments are welcome from any intere affiliation with AIAA. Comments should be directed to: American Institute of Aeronautics and Astronautics Standards Department 1801 Alexander Bell Drive, Suite 500 Reston, VA 22091 or, by electronic mail to: standards@aiaa.org The first draft of this guide was prepared by Unmeel B. Mehta Oberkampf, Munir M. Sindir, and A. Terrence Conlisk. A num improvements of the document were made by members of the Committee on Standards and by several interested individuals appreciate and value all input provided. The following committee members voted on this document: John L. Porter, Committee Chair (Sverdrup Technology) Ramesh Agarwal (Wichita State University) reasin on th ers, and users of CFD by establishing on of CFD simulations. The terminology cience disciplines. guides, recommended practices, and nyone engaged in industry or trade is andards publication and no commitment not intended to be used for certification ing standards publications, the AIAA ply to the subject matter. Prospective ves against liability for infringement of n the state of the art and on comments sted party, regardless of membership . This draft was prepared by William L. ber of comments and suggestions for AIAA Computational Fluid Dynamics who were not on the Committee. We v AIAA G-077-1998 Ram S. Azad (University of Manitoba) Donald Bain (CFD Research Corporation) John A. Benek (Microcraft Corporation) Bobby L. Berrier (NASA Langley Research Center) A. Terrence Conlisk (Ohio State University) Raymond R. Cosner (The Boeing Company) Robert A. Delaney (Allison Engine Company) Klaus Hoffmann (Wichita State University) Michael S. Holden (Calspan Corporation) Louis G. Hunter (Lockheed Martin Corporation) Yuji Ikeda (Kobe University) R. E. Luxton (University of Adelaide) Joseph G. Marvin (NASA Ames Research Center) Unmeel B. Mehta (NASA Ames Research Center) Robert E. Melnik (Northrop Grumman Corporation) Michele Napolitano (Politechnico Di Bari) William L. Oberkampf (Sandia National Laboratories) Gerald A. Paynter (The Boeing Company) Louis A. Povinelli (NASA Lewis Research Center) Cary Presser (National Institute of Standards and Technology) Balu Sekar (U.S. Air Force Wright Laboratory) Munir M. Sindir (The Boeing Company) Ashok K. Singhal (CFD Research Corporation) Ambady Suresh (NYMA, Inc.) John C. Tannehill (Iowa State University) The CFD Committee on Standards approved the document on January 14, 1998. The AIAA Standards Executive Council accepted it for publication on May 6, 1998. vi AIAA G-077-1998 Executive Summary Computer simulations of fluid flow processes are now used to design, investigate, and operate engineered systems and to determine the performance of these systems under various conditions. Computational fluid dynamics (CFD) simulations are also used to improve understanding of fluid physics and chemistry, such as turbulence and combustion, and to aid in weather prediction and oceanography. Although CFD simulations are widely conducted in industry, government, and academia, there is presently little agreement on procedures for assessing their credibility. These guidelines are predicated upon the notion that there is no fixed level of credibility or accuracy that is applicable to all CFD simulations. The accuracy level required of simulations depends on the purposes for which the simulations are to be used. The two main principles that are necessary for establishing credibility are verification and validation (V&V). As defined here, verification is the process of determining that a model implementation accurately represents the developer’s conceptual description of the model and the solution to the model. Validation is defined as the process of determining the degree to which a model is an accurate representation of the real world from the perspective of the intended uses of the model. These definitions point out that V&V are ongoing activities that do not have a clearly defined completion point. Completion or sufficiency is usually determined by practical issues such as budgetary constraints and intended uses of the model. All encompassing proofs of correctness, such as those developed in mathematical analysis, do not exist in complex modeling and computational simulation. The definitions of V&V also stress the evaluation of accuracy. In verification activities, accuracy is generally measured with respect to benchmark solutions of simplified model problems. In validation activities, accuracy is measured with respect to experimental data, i.e., reality. Uncertainty and error can be considered as the broad categories that are normally associated with loss in accuracy in modeling and simulation. Uncertainty is defined as a potential deficiency in any phase or activity of the modeling process that is due to lack of knowledge. Lack of knowledge is commonly caused by incomplete knowledge of a physical characteristic or parameter, as in the inadequate characterization of the distribution of surface roughness on a turbine blade. Lack of knowledge can also be caused by the complexity of a physical process, e.g., turbulent combustion. Error is defined as a recognizable deficiency in any phase or activity of modeling and simulation that is not due to lack of knowledge. Error can be categorized as either acknowledged or unacknowledged. Examples of acknowledged errors are round-off error in a digital computer and physical approximations made to simplify the modeling of a physical process. Unacknowledged errors include blunders and mistakes, such as programming errors. In the context of V&V, the meaning of the word “prediction” is restricted from its general usage to consider the history of validation activities with the CFD model. Prediction is defined as the use of a CFD model to foretell the state of a physical system under conditions for which the CFD model has not been validated. This definition of prediction is a subset of the general meaning of prediction because it eliminates past comparisons with experimental data. If this restriction is not made, then one is only demonstrating previous agreement with experimental data in the validation database. The processes or activities of V&V should be viewed as historical statements, i.e., reproducible evidence that a model has achieved a given level of accuracy in the solution of specified problems. Viewed in this light, it becomes clear that the V&V processes do not directly make claims about the accuracy of predictions. The fundamental strategy of verification is the identification and quantification of error in the computational solution. In CFD simulations, there are four predominant sources of error, namely insufficient spatial discretization convergence, insufficient temporal discretization convergence, lack of iterative convergence, and computer programming. The most important activity in verification testing is systematically refining the grid size and time step. The objective of this activity is to estimate the discretization error of the numerical solution. As the grid size and time step approach zero, the discretization error should asymptotically approach zero. When the asymptotic region has been vii AIAA G-077-1998 demonstrated, Richardson’s extrapolation can be used to estimate zero-grid spacing and time step. In most cases, CFD equations are highly nonlinear, and the vast majority of methods of solving these equations requires iteration. These iterations normally occur in two situations: 1) globally for boundary value problems (i.e, over the entire domain); and 2) within each time step for initial-boundary value problems. In verification testing, the sensitivity of the solution to the magnitude of the convergence criteria should be varied, and a value should be established that is consistent with the objectives of the simulation. In verification activities, comparing a computational solution to a highly accurate solution is the most accurate and reliable way to quantitatively measure the error in the computational solution. However, highly accurate solutions are known only for a relatively small number of simplified problems. These highly accurate solutions can be classified into three types: analytical solutions, benchmark numerical solutions to ordinary differential equations (ODEs), and benchmark numerical solutions to partial differential equations (PDEs). As one moves from analytical solutions to ODE solutions to PDE solutions, the accuracy of the benchmark solutions clearly becomes more of an issue. The fundamental strategy of validation is the identification and quantification of error and uncertainty in the conceptual and computational models. The recommended validation method is to employ a building- block approach. This approach divides the complex engineering system of interest into three progressively simpler phases: subsystem cases, benchmark cases, and unit problems. The strategy in this approach is the assessment of how accurately the computational results compare with experimental data (with quantified uncertainty estimates) at multiple levels of complexity. Each phase of the process represents a different level of flow physics coupling and geometrical complexity. The complete system consists of the actual hardware or system for which a validated CFD tool is needed. Thus all the geometric and flow physics effects occur simultaneously; commonly, the complete system includes multidisciplinary physical phenomena. Subsystem cases represent the first decomposition of the actual hardware into simplified or partial flow paths. Each of these cases commonly exhibits restricted geometric or flow features compared to the complete system. Benchmark cases represent another level of successive decomposition of the complete system. For these cases, separate hardware is fabricated to represent key features of each subsystem. The benchmark cases are geometrically simpler than those at the subsystem level, as only two separate features of the flow physics and two flow features are commonly coupled in the benchmark cases. Unit problems represent the total decomposition of the complete system. High- precision, special-purpose hardware is fabricated and inspected. Unit problems are characterized by very simple geometries, one flow-physics feature, and one dominant flow feature. Each of these phases is also characterized by different quantities of experimental information available for the initial conditions and boundary conditions that are used to solve the PDEs at each phase. In addition, the estimate of experimental measurement uncertainty varies considerably from one phase to another. viii AIAA G-077-1998 1.