The Assessment of Simulation Credibility in Virtual Engineering Workflows: The Challenge of Uncertainty Quantification

This presentation was made at the 2019 NAFEMS World Congress in Quebec Canada

Resource Abstract

Manufacturing processes play a significant role in creating variability in the mechanical properties of industrial products, thus increasing the potential for suboptimal performance and early failure. Computer Aided Engineering (CAE) tools can help to mitigate the risk of major deviations from the target performance, breaking down the complexity of the physical and chemical phenomena involved in the different production stages into a set of mathematical models which provide quantitative insights to understand the leading design factors and devise optimization strategies.



A variety of CAE tools is typically used to model manufacturing processes and performance verification tests and a substantial amount of work is required to link the different tools and secure that the output from the simulation of each process is correctly transferred to the input of the model representing the process which comes next in the manufacturing chain. A joint effort to secure the interoperability of CAE tools in several branches of manufacturing industry has led to initiate the project VMAP, whose goal is the development of an efficient and vendor-neutral protocol for the exchange of material data across various software platforms.



In addition to enable seamless data exchange, any interface built to facilitate the integration of CAE tools should include a structured and transparent system to manage the information related to the credibility of the simulations. The assessment of credibility of engineering models is a key requirement even in the simplest CAE ecosystems, i.e. where a single software tool generates models to mitigate the risk and increase the efficiency of product development. The critical challenge of evaluating the predictive capability of engineering simulations and describe it in some standardized form is greatly amplified in the context of virtual manufacturing workflows comprising a range of models with different characteristics in terms of physics represented, typical length and time scales and numerical implementation.



All models require parameters and simplifying assumptions (regarding, for example, boundary conditions and material properties) and ignoring the level of uncertainty associated to these factors, being them determined by experimental measurements or numerical simulations, might severely compromise the predictive capability of the model. Therefore, a desirable feature of a standard interface for material data exchange would be a sensible way to document and propagate uncertainty data about material properties throughout the entire simulation chain, i.e. from manufacturing to performance verification.



Here, we present an effort to clarify the requirements that a format for material data exchange as envisaged in the VMAP project is expected to fulfil in order to enable consistent uncertainty quantification throughout the whole chain of integrated simulation steps. First, the role played by model credibility assessment and the potential of some general methods for sensitivity analysis and uncertainty propagation is illustrated through a simple, but representative example on modelling-based design of reliable bungee jumping cords. Existing standardization practices and methods already developed in other areas of modelling (and particularly within Life Cycle Assessment) for minimizing the risks connected to lack of uncertainty assessment and poor data quality are then briefly reviewed and their potential to address specific issues in engineering simulation workflows will be examined.

Document Details

ReferenceNWC_19_416
AuthorSantandrea. F
LanguageEnglish
TypePresentation
Date 18th June 2019
OrganisationRISE Research Institutes of Sweden
RegionWorld

Download


Back to Search Results