Surrogate model based sequential sampling estimation of conformance probability for computationally expensive systems: application to fire safety science

Demeyer, Séverine; Fischer, Nicolas; Marquis, Damien

Journal de la société française de statistique, Special Issue : Computer Experiments, Uncertainty and Sensitivity Analysis, Tome 158 (2017) no. 1, pp. 111-138.

Le recours à la simulation numérique est devenue courant lorsque les expériences réelles sont impossibles ou réalisables qu’en très petit nombre. La modélisation statistique d’expériences numériques à partir de modèles de krigeage offre un cadre de décision probabiliste pour évaluer la probabilité de défaillance d’un système. La combinaison de simulations rapides de basse fidélité avec des simulations coûteuses de haute fidélité s’est avérée une méthode efficace pour diminuer le coût en simulations lors de la prévision de sorties d’un système. Par ailleurs, l’échantillonnage séquentiel est couramment utilisé pour estimer une probabilité de défaillance d’un système modélisé par krigeage. Dans cette étude, une méthodologie est exposée d’utilisation d’un plan séquentiel dans un cadre multi-fidélité pour prédire la probabilité de défaillance d’un système numérique coûteux et son incertitude. La méthodologie est appliquée à un cas d’étude en ingénierie de la sécurité incendie pour évaluer la probabilité de non-conformité d’un système d’évacuation de fumée à partir d’outils numériques complexes de simulation incendie.

The use of complex simulation systems has become common practice when physical experiments are not feasible or when too few are feasible. The statistical modelling of numerical experiments with kriging models yields a probabilistic decision framework to assess the probability of failure of the system. Combining fast low-fidelity simulations with costly high-fidelity simulations has proved an efficient method to decrease the burden of costly simulations when predicting the output of a system. In addition, sequential design is commonly used to estimate the probability of failure of a system modelled by kriging. In this work, a methodology is derived to benefit from sequential design in a multi-fidelity framework to predict the probability of failure of a computationally expensive system and its uncertainty. The methodology is applied to a fire safety engineering case study to assess the probability of non-conformity of a smoke control system from complex numerical fire tools.

[1] Alpert, R. L. Ceiling jet flows, SFPE handbook of fire protection engineering (Nenno, P.J. Di, ed.), National fire protection association, Quincy, MA, 2008, p. 2.21-2.36

[2] Au, S.K; Wang, Z.-H; Lo, S.-M Compartment fire risk analysis by advanced Monte Carlo simulations, Engineering Structures (2007), pp. 2381-2390

[3] Bichon, B. J.; Eldred, M. S.; Swiler, L. P.; Mahadevan, S.; McFarland, J. M. Efficient Global Reliability Analysis for Nonlinear Implicit Performance Functions, AIAA Journal, Volume 46 (2008) no. 10, pp. 2459-2468

[4] Bect, J.; Ginsbourger, D.; L, L.; Picheny, V.; Vazquez, E. Sequential design of computer experiments for the estimation of a probability of failure, Statistics and Computing, Volume 22 (2012), pp. 773-793 | Zbl

[5] Chevalier, Clément; Bect, Julien; Ginsbourger, David; Vazquez, Emmanuel; Picheny, Victor; Richet, Yann Fast Parallel Kriging-Based Stepwise Uncertainty Reduction With Application to the Identification of an Excursion Set, Technometrics, Volume 56 (2014) no. 4, pp. 455-465 | DOI

[6] Courrier, N.; Boucard, P. A.; Soulier, B. The use of partially converged simulations in building surrogate models, Advances in Engineering Software, Volume 67 (2014), pp. 186-197

[7] Chen, R.B.; Hung, Y.C.; nad S.W. Yen, W.Wang Contour estimation via two fidelity computer simulators under limited resources, Computational Statistics, Volume 28 (2013), pp. 1813-1834 | Zbl

[8] Carlotti, P; Lamalle, D Sensitivity to boundary conditions for simulation of fire plume in enclosure (2013)

[9] Chevalier, C.; Picheny, V.; Ginsbourger, D. The KrigInv package : An efficient and user-friendly R implementation of Kriging-based inversion algorithms, Computational Statistics and Data Analysis, Volume 71 (2014), pp. 1021-1034 | Zbl

[10] Drysdale, D. An introduction to fire dynamics, John Wiley, 2011

[11] Demeyer, S.; Wright, L.; Pendrill, L.; Bär, M. Intermediate report on case study on fire engineering, A report of the EMRP joint research project NEW04 Novel mathematical and statistical approaches to uncertainty evaluation (2015)

[12] Echard, B.; Gayton, N.; Lemaire, M. AK-MCS: An active learning reliability method combining Kriging and Monte Carlo Simulation, Structural Safety, Volume 33 (2011) no. 2, pp. 145 -154 | DOI

[13] French regulation Instruction technique, N${}^{\circ }$246 relative au désenfumage dans les établissements recevant du public, Annexe pour le calcul du taux alpha, Arrêté du 22 mars 2004, Documentation BATISS (French Regulation), 2004

[14] Fernández-Godino, G.; Park, C.; Kim, N.H.; Haftka, R.T. Review of multi-fidelity models, ArXiv e-prints (2016) | arXiv

[15] Geraci, G.; Eldred, M.; Iaccarino, G. A multifidelity control variate approach for the multilevel Monte Carlo technique (2015), pp. 169-181 (Center for Turbulence Research Annual Research Briefs)

[16] Giles, Michael B. Multilevel Monte Carlo Methods, Springer Berlin Heidelberg, Berlin, Heidelberg (2013), pp. 83-103 | DOI | Zbl

[17] Gano, S.E.; Renaud, J.E.; Martin, J.D.; Simpson, T.W. Update strategies for kriging models used in variable fidelity optimization, Structural and Multidisciplinary Optimization, Volume 32 (2006), pp. 287-298

[18] Huang, D.; Allen, T.T.; Notz, W.I.; Miller, R.A. Sequential kriging optimization using multiple-fidelity evaluations, Structural and Multidisciplinary Optimization, Volume 32 (2006), pp. 369-382

[19] ISO 7730 NF EN ISO 7730 - Analytical determination and interpretation of thermal confort using calculation of the PMV and PPD indices and local thermal comfort criteria, Volume ICS: 13.18024, 2006

[20] ISO 13571 NF EN ISO 13571 - Guidelines for the estimation of time to compromised tenability in fires, Volume ICS: 13.220.01, 2012

[21] Jones, W.W; Peacock, R.D; Forney, G.P; P.A, Reneke CFAST - Consolidated Model of Fire Growth and Smoke Transport (Version 6), Technical Reference Guide, NIST SP - 1026 (2009), 125 pages

[22] Kandasamy, K.; Dasarathy, G.; Oliva, J. B.; Schneider, J.; Poczos, B. Multi-fidelity Gaussian Process Bandit Optimisation, ArXiv e-prints (2016) | arXiv | Zbl

[23] Kong, D; Johansson, N; van Hess, P; Lu, S; Lo, S A Monte Carlo analysis of the effect of heat release rate uncertainty on available safe egress time, Journal of Fire Protection Engineering (2012), pp. 5-29

[24] Kennedy, M. C.; O’Hagan, A. Predicting the Output from a Complex Computer Code when Fast Approximations are Available, Biometrika, Volume 87 (2000), pp. 1-13 | Zbl

[25] Le Gratiet, L. MuFiCokriging: Multi-Fidelity Cokriging models (2012) http://CRAN.R-project.org/package=MuFiCokriging (R package version 1.2)

[26] Le Gratiet, L. Bayesian analysis of hierarchical multifidelity codes, SIAM/ASA Journal on Uncertainty Quantification, Volume 1 (2013), pp. 244-269 | Zbl

[27] Le Gratiet, L.; Cannamela, C. Cokriging-Based Sequential Design Strategies Using Fast Cross-Validation Techniques for Multi-Fidelity Computer Codes, Technometrics, Volume 57 (2015) no. 3, pp. 418-427 | DOI

[28] Myers, R. H.; Montgomery, D. C. Response Surface Methodology: Process and Product Optimization Using Designed Experiments, John Wiley, 2015 | Zbl

[29] McGrattan, K.; McDermott, R; Weinschenk, C; Overholt, K Fire dynamics simulator - technical reference guide, Volume 1 Mathematical model, NIST SP - 1018 (2014), 175 pages

[30] Overholt, K. J.; Ezekoye, O. A. Characterizing heat release rates using an inverse fire modeling technique, Fire Technology, Volume 48 (2012), pp. 893-909

[31] Oakley, J.; O’Hagan, A Bayesian inference for the uncertainty distribution of computer model outputs, Biometrika, Volume 89 (2002), pp. 769-784

[32] Picheny, V.; Ginsbourger, D.; Roustant, O.; Haftka, R.T.; Kim, N.-H. Adaptive Designs of Experiments for Accurate Approximation of a Target Region, Journal of Mechanical Design, Volume 132 (2010)

[33] Perdikaris, P.; Venturi, D.; Royset, J. O.; Karniadakis, G. E. Multi-fidelity modelling via recursive co-kriging and Gaussian-Markov random fields, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science,, Volume 471 (2015)

[34] Peng, C.-Y.; Wu, J. On the choice of nugget in kriging modeling for deterministic computer experiments, Journal of Computational and Graphical Statistics, Volume 23 (2014), pp. 151-168

[35] Peherstorfer, B.; Willcox, K.; Gunzburger, M. Survey of multifidelity methods in uncertainty propagation, inference, and optimization (2016) (ACDL Technical Report TR16-1)

[36] Ranjan, Pritam; Bingham, Derek; Michailidis, George Sequential experiment design for contour estimation from complex computer codes, Technometrics, Volume 50 (2008) no. 4, pp. 527-541 | DOI

[37] Roustant, Olivier; Ginsbourger, David; Deville, Yves DiceKriging, DiceOptim: Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization, Journal of Statistical Software, Volume 51 (2012) no. 1, pp. 1-55 http://www.jstatsoft.org/v51/i01/

[38] Rubinstein, R. Y.; Kroese, R. Y. Simulation and the Monte Carlo Method, Wiley, 2008 | Zbl

[39] Rasmussen, C. E; Williams, C K. I. Gaussian Processes for Machine Learning, MIT Press, 2006 | Zbl

[40] Stroh, R.; Bect, J.; Demeyer, S.; Fischer, N.; Vazquez, E. Assessing Fire Safety using Complex Numerical Models with a Bayesian Multi-fidelity Approach, Fire Safety Journal (2017) (submitted)

[41] Santner, T. J.; Notz, W. The design and analysis of computer experiments, Springer Verlag, Springer Series in Statistics, 2003 | Zbl

[42] Sacks, J.; Welch, W. J.; Mitchell, T. J.; Wynn, H. P. Design and analysis of computer experiments, Statistical Science, Volume 4 (1989), pp. 409-423 | Zbl

[43] Xiong, S.; Qian, P.Z.G.; Wu, J. Sequential design and analysis of high-accuracy and low-accuracy computer codes, Technometrics, Volume 55 (2013), pp. 37-46