Skip to main content

Empowered decision-making in simulation-based engineering: Advanced Model Reduction for real-time, inverse and optimization in industrial problems

Periodic Reporting for period 2 - AdMoRe (Empowered decision-making in simulation-based engineering: Advanced Model Reduction for real-time, inverse and optimization in industrial problems)

Reporting period: 2017-10-01 to 2019-09-30

Computational mechanics tools are well integrated in the technological industrial practice. Nonetheless, simulation-based techniques are not extensively used in real-time for decision-making, due to their overall cost (preprocess, solve and postprocess). Real-time (fast-queries) is critical for control of manufacturing processes, non-destructive testing and fast decision-making at production phases. This is also the case for multiple-queries: optimization and parameter identification with uncertainty quantification. Addressing fast and multiple queries is crucial in applications of major industrial interest, independently of the manufactured product (engines, cars, planes, helicopters, medical devices). The actual bottleneck lies in the computational effort to be furnished in solving each of the queries with the desired accuracy.

AdMoRe focused on frontier research on reduced order models – in particular, the proper generalised decomposition - to develop innovative algorithms to solve fast and multiple queries in industrial problems. To maximise the impact of the project, AdMoRe industrial partners identified four strategic sectors for the European industry: composites and forming technology; fast simulations for geometrical design; inverse problems and identification; integration of reduced order models into commercial codes.

The following five actions were performed:
1. Development of innovative solutions targeting specific needs of the industry by means of fundamental and application-oriented research. The proactive participation of the industrial partners in the definition of the research goals and their continuous interaction with the academic partners were crucial for addressing current and future needs of the European industry.
2. Strengthening of the position of European industry in simulation-based engineering via the training of a new generation of research engineers, their integration in the R&D departments of the industrial partners and the consolidation of long-time strategic collaborations with the academic partners.
3. Production of original research by the eight ESRs and the senior researchers from academia and industry.
4. Realization of code demonstrators to illustrate the capabilities of novel methodologies to provide added value to existing software, fostering faster transfer of knowledge from academia to industry.
5. Empowering of the eight trainees on translational research and entrepreneurial skills by immersing them in a dynamic industrial environment and by exploiting business incubators within academic institutions.
The major outcomes of AdMoRe are the successful training of a new generation of engineers on industrially-oriented research topics, the continuous interaction and mutual transfer of knowledge between academia and industry and the consequent reduction of the time required to incorporate new research solutions in a daily industrial production environment.
Reduced order models are the core of AdMoRe being the topic common to all ESRs projects. The eight individual research projects applied model reduction techniques to various fields of interest and led to fundamental and application-oriented contributions, as testified by the varied high-quality scientific production of AdMoRe.

Concerning thermo-mechanics problems, a simulation-based strategy for monitoring and controlling in real-time the production processes of thermoplastic composites in the aerospace industry was developed. For flow problems, a non-intrusive reduced order model, exploiting a computational fluid dynamics solver validated by the industry, was proposed to treat parametric problems of interest for automotive applications. Concerning inverse problems, fast analysis tools for MRI scanners were developed to assist engineers during different stages of the manufacturing process. Finally, the novel algorithmic solutions were integrated in existing software to maximise the impact of the strategies developed within AdMoRe.

Besides the scientific achievements above, dissemination and outreach activities were crucial for the success of AdMoRe. On the one hand, they involved trainings on communications skills and oral presentations in international conferences and seminars, in academic and industrial environments. The main goal was to develop the confidence of ESRs when presenting their research in a variety of scenarios and in front of different audiences, to prepare them to work in interdisciplinary teams on multidisciplinary projects. On the other hand, public engagement activities gave the ESRs the opportunity to gently explain their findings to a non-expert audience, contributing to the development of soft skills crucial for the future success of the research engineers trained by AdMoRe. Presentation of results to an extended audience, beyond academia and industry, was practiced in written and oral form via vulgarisation articles and blog posts on the AdMoRe website, short talks during Ambassadors’ Day and OpenDay activities and videoclips released on the internet.
Many computational models are not currently used to their full potential by the European industry and a competitive edge is being lost with respect to other well-established and new-emerging economies. AdMoRe contributed to maintain the worldwide leadership of the four European industrial partners involved, by reducing the time required to transfer knowledge from academia to industrial innovation, maximizing the profitability of European research.

The achievements of AdMoRe will have a major impact in different fields of European industry in the upcoming years:
1. The production of large thermoplastic composite parts is a challenging task in the aerospace industry. Within AdMoRe, a simulation-based strategy for monitoring and controlling in real-time the thermo-mechanical production processes was developed to assist engineers and experimentalists in such operations.
2. Design and optimization of vehicles require simulating complex flows for a large number of configurations. AdMoRe achieved solving parametric flow problems efficiently via a non-intrusive model reduction technique which exploits a computational fluid dynamics solver validated by the industry. These results pave the way for the industrialization of model reduction strategies and their integration in a daily production environment.
3. MRI scanners use pulsed field gradients to form an image but they are often affected by artifacts. The research of AdMoRe led to the development of an efficient reduced order model to predict magnetic field variations induced by user defined parameters, providing a set of tools for the analysis and benchmarking of MRI scanners at different stages of the manufacturing process.
4. The incorporation of scientific and technical developments into commercial codes was a key aspect of the research developed within AdMoRe. Code demonstrators were produced to illustrate the capabilities of novel methodologies to provide added value to existing software, fostering faster transfer of knowledge from academic research to industrial innovation.
Code integration: commercial codes and reduced order models
Fast simulations for geometrical design
Composites & forming technology
Inverse problems & identification
Inverse problems & identification