As computing technologies are maturing, Finite element Analysis (FEA) has become a major computer-based analytical tool for structural analysis. Now it is feasible to analyze a simple beam to large scale complex structures, linear and non-linear regimes, fluids and solids, and variety of materials such as foams, composites and metals using FEA. In fact, it is also feasible to analyze all these in a single FE model. Due to such robustness, FEA has become popular in all industries.
If we consider aircraft certification, the cost factors for achieving certification are very large. This is also evident by the lack of competition in the civil aviation industry. The certification process requires a significant number of tests, starting from coupon level to component level and in some cases even full-scale tests [1]. With the emerging markets of EVTOL and electric propulsion, and the desire to bring certification costs down to increase competition, there is a push for adoption of FEA for supporting certification by the means of virtual testing.
The idea of virtual testing is to replace some or all structural tests by validated FEA. In the authors opinion, this is something that can be achieved and will become reality in the future but currently we are far off from that goal. The robustness that FEA provides is somewhat the issue that is also hindering it from replacing physical tests [1]. That is because it has been particularly difficult to build confidence on FEA results within the design and certification communities.
Not only FEA but any structural analysis performed by engineers should always be scrutinized – for good reason. The initial conditions and inputs used in the analysis and the interpretation of results can all have big implications on the product design. The results of the analysis influence the weight of the product, material selection, test outcomes, manufacturing and more. All these issues have a direct impact on the cost and timeline of the product development. Further implications can also arise if the analysis is used to support certification of the product.
Let’s look at some methods for product development using FEA and some processes one can follow to build confidence in their FEA analysis. For complex models a combination of these would be required.
Methods for Product Development using FEA
- Building block approach
This approach is useful for large scale models or component level models. The basic idea of the building block approach is to model complex analytical models that are supported by increasingly difficult intermediate analyses which have been validated by tests. The cost increases as you step through these intermediate analyses and tests, leading up to your most complex model and most expensive test [2][3].
As you work through different building blocks, you obtain the knowledge and experience of the performance of your product at different levels. This enables you to build databases supported by physics-based analyses. You can then create standards, databases and processes that can be used for product development and certification. The knowledge can also be more easily transferred from an experienced engineer to a beginner as the information is broken down into building blocks.
- The SAFESA approach
The SAFESA approach is to formalize the process of structural qualification. The objective is to minimize errors – especially in the idealization process. This approach provides a systematic framework for carrying out FEA within a product qualification. The framework is designed to help to minimize, identify and treat errors. Further details can be found here 🡪 NAFEMS – SAFESA Technical Manual
- Databases, Libraries and Standards
Another method for qualifying products using FEA is to rely on extensive databases of similar FEA performed in the past. This can be especially useful if the FEA in the database has been validated against experiments. In addition, following some standard work procedures in creating your FEA would provide consistency in results. The issue with this approach is that most companies do not have such extensive databases and it is costly to generate them. However, if a plan is put in place early on, these databases can be generated for future application of FEA for product qualification.
The focus should be on creating libraries that are well organized and validated so that FEA engineers can directly use the data from these libraries to build their simulations. For example, having access to a library of material cards that can be used by all FEA engineers.
Processes for building confidence in your FEA
- Basic Checklist
At a minimum, the analyst must follow a checklist that covers the major steps involved in their FEA. A basic checklist is available to download under our FEA Tools section.
- MSB-NET Checklist
This checklist includes a report form defining recommendations for FEA in the field of Orthopedic and Trauma biomechanics. The motivation behind this work is to address problems and controversies that arise in the verification and validation process as well as in the set-up and evaluation of FEA. The checklist is intended to identify serious errors of the FEA and to improve the credibility of the FEA in biomechanical investigations. Although their work is focused on biomechanics, the author believes that their extensive checklist can be applied to any FEA analysis. Further details can be found here 🡪 Reporting checklist for verification and validation of finite element analysis in orthopedic and trauma biomechanics – ScienceDirect
- ASME Verification and Validation (V&V) Guide [4]
The American Society of Mechanical Engineers (ASME) has developed guides and standards that provide a roadmap for forming and documenting a V&V plan. This plan generally includes four sections:- Introduction – The role of V&V plan is described, and concepts of V&V are introduced.
- Model Development – description of the mathematical, conceptual and computational models.
- Verification – code and calculation verification. It must be verified that the mathematical model and results produced using these mathematical models are error free. Verification must precede validation
- Validation – determine the degree to which model is an accurate representation of the real world for the problem. This requires generation of experimental data to compare to analytical model. The level of accuracy and uncertainty is quantified in the validation process.
Further details about the ASME V&V can be found here 🡪 Verification, Validation and Uncertainty Quantification (VVUQ) – ASME
AIAA G-077-1998
This is a guide for verification and validation of computational fluid dynamics simulations. It provides a foundation for fundamental issues related to verification and validation and establishes common terminology that can be used across a variety of engineering disciplines. The guide does not specify any standards citing important issues yet to be resolved. Further details can be found here 🡪 Guide: Guide for the Verification and Validation of Computational Fluid Dynamics Simulations (AIAA G-077-1998(2002)) | AIAA StandardsNASA-STD-7009
Established as an outcome of the Columbia Accident Investigation Board (CAIB), the NASA-STD-7009 ensures that models and simulations (MS) are developed, applied and interpreted appropriately for making decisions that may impact crew or mission safety. This document is focused on probabilistic and deterministic biological MS and provides applicable information, tools and techniques. Further details can be found here 🡪 NASA-STD-7009 Guidance Document for Human Health and Performance Models and Simulations – NASA Technical Reports Server (NTRS)Sandia Report SAND2002-0341
This report details the process used in planning, executing and assessing experimental validation of Accelerated Strategic Computing Initiative (ASCI) codes. The report focuses on model validation. Further details can be found here 🡪 General Concepts for Experimental Validation of ASCI Code Applications (Technical Report) | OSTI.GOVUncertainty Quantification
Uncertainty in modeling and simulation results should be reported. The sources of uncertainty should be identified and reported. Some references that discuss in detail about uncertainty estimation in modeling and simulation are 🡪 [PDF] Estimation of Total Uncertainty in Modeling and Simulation | Semantic Scholar and Guide: Guide for the Verification and Validation of Computational Fluid Dynamics Simulations (AIAA G-077-1998(2002)) | AIAA Standards
It is evident that there is significant effort across various industries to incorporate methods that build confidence in FEA. As the application of FEA increases, the methods and processes will also become more refined. If your company is looking to implement FEA for research, product development or certification, we highly recommend you think about a framework to include this methods and processes early on. Some of these methods require significant investment of time, money and resources and thus having a clear plan is important.
Having said that, Algo Engineering is working hard to develop validated FEA models (against existing test data) and tools that can help you in implementing these methods. Please feel free to contact our team for your FEA needs.
References:
[1] Symons, C. and Morris, A. (1997), “Analytic certification of airframes”, Aircraft Engineering and Aerospace Technology, Vol. 69 No. 3, pp. 235-240.
[2] Burger, U., Rochat, L., Breton, C., & Markmiller, J.F. (2015), “A methodology to assess damage tolerance of composite structures by fea simulation technique.”
[3] Zinzuwadia, C., Olivares, G., Ly, H., & Gomez, L. (2018), “Crashworthiness by Analysis: Verifying FEA Modeling Capabilities by Accident Reconstruction”, Earth and Space 2018.
[4] Schwer, L. (2007), “An overview of the PTC 60/V&V 10: Guide for verification and validation in computational solid mechanics: Transmitted by L. E. Schwer”, Chair PTC 60V&V 10. Eng. Comput. (Lond.). 23. 245-252. 10.1007/s00366-007-0072-z.