Bayesian Identification of Thermodynamic Parameters from Shock Tube Data

Bayesian Identification of Thermodynamic Parameters from Shock Tube Data

Butler, Jacob (TU Delft Aerospace Engineering)

Dwight, R.P. (mentor)
Pini, M. (graduation committee)
Hickel, S. (graduation committee)

Delft University of Technology

Aerospace Engineering

2018-02-20

The project concerns uncertainty reduction of parameters of a thermodynamic equation of state for a dense gas, using Bayesian inference. The dense gas considered is D6 siloxane and the equation of state used is the polytropic van der Waals equation. The shock tube data comes from the flexible asymmetric shock tube (FAST) experiment. This is modeled using the quasi-one-dimensional Euler equations with a source term that depends on time. A surrogate model based on sparse grids and a sensitivity analysis using Sobol' indices are both applied. The Markov chain Monte Carlo technique is applied to sample from the posterior probability distribution on the chosen parameters of the computer model. The results indicated that some of the thermodynamic parameters were identified, but that their mean values showed a disagreement with the true values in the literature.

Bayesian Inference
Dense gas flow
Markov Chain Monte Carlo

http://resolver.tudelft.nl/uuid:6c0a3871-aaca-45d3-80dc-a4265f95088a

Student theses

master thesis

© 2018 Jacob Butler