Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.768897
Title: Uncertainty and sensitivity analysis of computational simulations of industrial jet flows
Author: Granados Ortiz, Francisco Javier
ISNI:       0000 0004 7655 8334
Awarding Body: University of Greenwich
Current Institution: University of Greenwich
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The main purpose of this thesis has been to improve the deterministic computational simulations of two industrial problems with a stochastic approach. For this purpose, uncertainty quantification and global sensitivity analysis have been carried out in two industrial jet flows with different objectives. First, an impinging swirling air jet for heat transfer purposes is generated by an axisymmetric rotating pipe. Despite that many mechanisms to generate swirling jets for heat transfer can be found in literature, there is no application with rotating pipes. To develop such study, two Computational Fluid Dynamic (CFD) simulations are carried out in FLUENT (Simulation 1 for the rotating pipe to generate the swirling flow and Simulation 2 for its impingement on a heated flat plate) in order to compare this heat transfer mechanism with others available in literature. Once the simulations are validated, experimental uncertainties are taken into account in the simulation by means of the Stochastic Collocation Method, as to provide measures of uncertainty is typical in experimental studies but also relevant in CFD. Moreover, as this Uncertainty Quantification (UQ) method requires to run Simulation 1 several times, mathematical models are also investigated for the outflow turbulent and velocity profiles to reduce the complexity and time invested in the analysis. The simulated results show that this swirling mechanism can significantly beat others in literature, specially by increasing the angular velocity in small nozzle-to-plate distances, increasing the heat transfer at the stagnation point. This is the only mechanism that provides such increase in short nozzle-to-plate distances. The investigated models were also implemented in User Defined Functions (UDF) introducing negligible errors in the simulations. Second, a framework is presented for the propagation of uncertainties in the computational research of under-expanded jets. For this aim, experimental and turbulence uncertainty is input to the CFD simulations of the jet flow and its impact is accounted and apportioned (UQ and sensitivity analysis) by means of generalised Polynomial Chaos and Kriging surrogates. The results show that some interesting parts in the computational domain suffer significant variations due to the input uncertainty. In addition to this, the last part of the framework is the stability analysis by the Parabolised Stability Equations (PSE). Stochastic Collocation is used to demonstrate that the impact of experimental and turbulence uncertainty can be linked to PSE in this framework, despite the fact that the PSE do not include them in their formulation.
Supervisor: Lai, Choi-Hong ; Pericleous, Kyriacos A. Sponsor: AeroTraNet 2 FP7-People PITN-GA-2012-317142
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.768897  DOI: Not available
Keywords: QA Mathematics
Share: