[Paper Review] Open TURNS: An industrial software for uncertainty quantification in simulation

Michaël Baudin, Anne Dutfoy+2 more|arXiv (Cornell University)|Jan 21, 2015

Simulation Techniques and Applications28 references20 citations

TL;DR

OpenTURNS is an open-source, industrial software platform for uncertainty quantification (UQ) in complex simulation frameworks, offering a modular, transparent, and multi-accessible environment for probabilistic UQ. It integrates advanced methods for uncertainty propagation, sensitivity analysis, and metamodeling, and enables seamless coupling with external black-box models via generic wrappers, significantly reducing computational burden in high-fidelity simulations across engineering and scientific domains.

ABSTRACT

The needs to assess robust performances for complex systems and to answer tighter regulatory processes (security, safety, environmental control, and health impacts, etc.) have led to the emergence of a new industrial simulation challenge: to take uncertainties into account when dealing with complex numerical simulation frameworks. Therefore, a generic methodology has emerged from the joint effort of several industrial companies and academic institutions. EDF R&D, Airbus Group and Phimeca Engineering started a collaboration at the beginning of 2005, joined by IMACS in 2014, for the development of an Open Source software platform dedicated to uncertainty propagation by probabilistic methods, named OpenTURNS for Open source Treatment of Uncertainty, Risk 'N Statistics. OpenTURNS addresses the specific industrial challenges attached to uncertainties, which are transparency, genericity, modularity and multi-accessibility. This paper focuses on OpenTURNS and presents its main features: openTURNS is an open source software under the LGPL license, that presents itself as a C++ library and a Python TUI, and which works under Linux and Windows environment. All the methodological tools are described in the different sections of this paper: uncertainty quantification, uncertainty propagation, sensitivity analysis and metamodeling. A section also explains the generic wrappers way to link openTURNS to any external code. The paper illustrates as much as possible the methodological tools on an educational example that simulates the height of a river and compares it to the height of a dyke that protects industrial facilities. At last, it gives an overview of the main developments planned for the next few years.

Motivation & Objective

• Address the growing industrial need to quantify and manage uncertainties in complex simulation frameworks due to stricter regulatory demands in safety, security, and environmental impact.
• Overcome limitations of domain-specific, non-transparent, and non-generic approaches to uncertainty treatment in industrial applications.
• Develop a unified, open, and extensible software platform that supports transparency, modularity, multi-accessibility, and industrial-scale computing for uncertainty quantification.
• Enable seamless integration of external simulation codes (black-box models) into a probabilistic UQ workflow through generic wrappers.
• Support diverse user needs through multiple interfaces (Python, C++, GUI) and scalable computing capabilities for high-performance environments.

Proposed method

• Implement a C++ library and a Python TUI (Textual User Interface) to provide both low-level performance and high-level usability.
• Support a wide range of probabilistic methods including Monte Carlo simulation, polynomial chaos expansion, and stochastic collocation for uncertainty propagation.
• Integrate advanced sensitivity analysis techniques such as Sobol’ indices and derivative-based methods to quantify input influence on model output.
• Provide metamodeling capabilities using kriging and polynomial chaos to reduce computational cost in expensive simulations.
• Use generic wrappers (e.g., DistributedPythonFunction) to interface with external simulation codes via file-based data exchange, enabling distributed and high-performance computing.
• Enable integration with the Salome platform and tools like YACS and ParaVIS for workflow automation and in-situ visualization of uncertainty results.

Experimental results

Research questions

• RQ1 How can a generic, open-source software platform effectively support uncertainty quantification across diverse industrial and scientific simulation domains?
• RQ2 What architectural and algorithmic features are required to ensure transparency, modularity, and multi-accessibility in industrial UQ workflows?
• RQ3 How can high-performance computing be efficiently leveraged to reduce the computational burden of uncertainty propagation in expensive simulation models?
• RQ4 To what extent can metamodeling techniques like kriging and polynomial chaos reduce the number of required model evaluations in uncertainty analysis?
• RQ5 How can uncertainty analysis results be effectively visualized and interpreted, especially when dealing with large-scale, high-dimensional data from complex simulations?

Key findings

• OpenTURNS successfully supports a broad spectrum of UQ tasks, including uncertainty quantification, propagation, sensitivity analysis, and metamodeling, with consistent and reliable results across different numerical methods.
• The software enables efficient uncertainty propagation on high-performance computing clusters, with one study requiring 8000 CPU hours across 32 nodes to perform 6000 evaluations of a computationally expensive nuclear waste storage model.
• The integration of OpenTURNS with the Salome platform and tools like YACS and ParaVIS allows for automated workflow execution and advanced visualization, including functional boxplots for uncertainty data.
• The use of metamodels such as kriging significantly reduces the number of required model evaluations, making uncertainty analysis feasible for expensive simulations.
• The software’s open-source nature and LGPL licensing have enabled widespread adoption, with integration into major Linux distributions and industrial platforms like Salome.
• Ongoing developments focus on enhancing kriging with advanced optimization techniques and improving in-situ visualization for high-volume uncertainty data, supported by national research funding (ANR Chorus project).