Development of Verification and Validation Procedures for Computer

Development of Verification and Validation Procedures for Computer Simulation use in Roadside Safety Applications NCHRP 22 -24 DEFINITIONS AND PROCEDURES Worcester Polytechnic Institute Battelle Memorial Institute Politecnico di Milano

FEA IN ROADSIDE HARDWARE DESIGN • Specialty codes prior to 1990 (i. e. , NARD, Guard, Barrier. VII, etc. ) • Shift to DYNA/LSDYNA in mid-1990’s • Today – Almost exclusively LSDYNA – Used in nearly all new product developments – Requests for approval now coming based partly or entirely on LSDYNA results. – Decision-makers need a way to judge good from bad results. Who do you trust? How do you make an acceptance decision based on simulations?

MOTIVATION • We have formal standards on how to perform and evaluate full-scale crash tests. • Decision makers want a formal standard on how to perform and evaluate FEA simulations used in the approval process. • NCHRP 22 -24 was initiated to develop these procedures. • Develop a procedure and format for validation and verification report for simulations that can be used like a crash test report.

OBJECTIVE … to develop guidelines for verification and validation of detailed finite element analysis for crash simulations of roadside safety features. The focus of these guidelines will be on establishing accuracy, credibility, and confidence in the results of crash test simulations intended (1) to support policy decisions and (2) to be used for approval of design modifications to roadside safety devices that were originally approved with full-scale crash testing.

Existing V&V Procedures Several organizations have developed V&V procedures in recent years. • • • NASA • Lockhead These are general purpose, broadbased guides that out-line general Do. D procedures and provide AIAA terminology definitions. They are not step-by-step guides. Sandia Los Almos LLNL ASME Guide draws on all the above guides in developing its recommendations. ASME FHWA/NARD These are domain-specific guides with more step-by-step FAA • EU (rail) procedures and defined metrics. FRA • Chrysler

ASME V&V 10 -2006 History 1999 An ad hoc verification & validation specialty committee was formed under the auspices of the United States Association for Computational Mechanics (USACM). 2001 ASME approved the committee’s charter: To develop standards for assessing the correctness and credibility of modeling and simulation in computational solid mechanics. 2006 2007 Committee was assigned the title and designation of the ASME Committee for Verification & Validation in Computational Solid Mechanics (PTC 60). ASME published the “Guide for verification and validation in computational solid mechanics. ” ASME V&V 10 -2006. Developing a series of “best practices guides”

ASME V&V 10 -2006 Committee Members The committee derives its authority from the diversity of its membership and consensus of opinion. M. C. Anderson, Los Alamos National Laboratory J. A. Cafeo, General Motors Corporation R. L. Crane, The American Society of Mechanical Engineers S. W. Doebling, Los Alamos National Laboratory J. H. Fortna, ANSYS M. E. Giltrud, Defense Threat Deduction Agency J. K. Gran, SRI International T. K. Hasselman, Acta Inc. H. M. Kim, Boeing R. W. Logan, Lawrence Livermore National Laboratory H. U. Mair, Institute for Defense Analyses A. K. Noor, Old Dominion University W. L. Oberkampf, Sandia National Laboratories J. T. Oden, University of Texas D. K. Pace, Consultant T. Paez, Sandia National Laboratories A. B. Pifko, Consultant L. Proctor, MSC Software J. N. Reddy, Texas A & M University P. J. Roache, Consultant L. E. Schwer, Schwer Engineering P. E. Senseny, Consultant M. S. Shephard, Rensselaer Polytechnic Institute D. A. Simons, Northrop Grumman B. H. Thacker, Southwest Research Institute T. G. Trucano, Sandia National Laboratories R. J. Yang, Ford Motor Company Y. Zhao, St. Jude Medical 10

ASME V&V 10 -2006 • The Guide does provide a: • Framework and process for V&V activities. • Standard definitions for V&V terms. • The Guide does not provide: • A step-by-step procedure for V&V. • Specific recommendation for metrics.

ASME V&V 10 -2006 Some Definitions • Validation -- 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. – Model results are compared to physical experiments. • Verification -- The process of determining that a computational model accurately represents the underlying mathematical model and its solution. – Model results are compared to known mathematical solution. • Calibration -- The process of adjusting physical modeling parameters in the computational model to improve agreement with experimental data. – Physical experiments used to estimate model parameters.

ASME V&V 10 -2006 Validation 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.

ASME V&V 10 -2006 Verification The process of determining that a computational model accurately represents the underlying mathematical model and its solution.

ASME V&V 10 -2006 Calibration The process of adjusting physical modeling parameters in the computational model to improve agreement with experimental data.

ASME V&V 10 -2006 Model Development Before we begin to develop a model, a reality of interest is identified (i. e. , what is the physical system to be modeled). “We need to know the wing tip deflection of the ABC experimental aircraft under a distributed load of X Newtons/meter, ” in this case the reality of interest is the aircraft wing. The airplane is manufactured by Karel Klenor - KLN, Choceň, the firm is one of the biggest producers of composites in the Czech Republic www. kln. cz

ASME V&V 10 -2006 Model Development Conceptual Model – “the collection of assumptions and descriptions of physical processes representing the solid mechanics behavior of the reality of interest from which the mathematical model and validation experiments can be constructed. ”

ASME V&V 10 -2006 Model Development Mathematical Model – “The mathematical equations, boundary values, initial conditions, and modeling data needed to describe the conceptual model. ”

ASME V&V 10 -2006 Model Development Computational Model – “The numerical implementation of the mathematical model, usually in the form of numerical discretization, solution algorithm, and convergence criteria. ” Commercial Software

ASME V&V 10 -2006 V&V Process

ASME V&V 10 -2006 V&V Process Verification

ASME V&V 10 -2006 V&V Process Verification Validation

ASME V&V 10 -2006 Validation Process The goal of the validation process is to assess the predictive capability of the model by comparing the predictive results of the model with validation experiments. Three key elements of Validation: 1. Precision Testing 2. Uncertainty Quantification 3. Comparative Metrics 25

ASME V&V 10 -2006 Comments on the V&V Process • • • The V&V process diagram is valid not only for whole models but for components, assemblies, parts, etc. While most roadside safety work uses LSDYNA, this process and definitions are applicable to any numerical simulation software (e. g. , MADYMO, BVII, HVOSM, HVE, etc. ). We can not usually do code verification – we do not generally have access to the code. – • We can do calculation verification – this is another word for benchmarking. – • Counter example: Yvonne Murray’s soil and timber models for LSDYNA. Example: do different versions of LSDYNA produce the same result? Do different computational platforms produce the same result? Notice the comparison is quantitative. – Qualitative validation is not really validation because it is subjective.

Recommendation • The project team recommends that we adopt the ASME V&V 10 -2006 Guide as a basis for the basic V&V process and definition of terms because … • The 22 -24 panel definitions are more or less consistent with the ASME definitions and … • The ASME definitions have been adopted by a broad cross-section of the computational solid mechanics community and. . . • The ASME process includes all the essential pieces needed in computation roadside safety. • Discussion?

Hierarchical Modelling • The ultimate goal is to use a validated model to extrapolate results to an untested situation. • We need to have confidence in the model before we can use it to predict untested situations.

Hierarchical Modelling Barrier assembly Vehicle assembly Whole model level Top Rail assembly Middle Rail assembly Post part Assembly Level Rubrail part Guardrail part Posts parts Part Level Spacer part Blockout part Main -rail part Stiffner parts

Hierarchical Modelling

Hierarchical Modelling

Hierarchical Modelling