Browsing by Subject "QMOM"
Now showing 1 - 3 of 3
- Results Per Page
- Sort Options
Item A study of time-dependent general dynamic equation models for aerosol particle coagulation(2018-05-04) Omana, Michael Alexis; Ezekoye, Ofodike A.Aerosol dynamics are complex processes that tend to be difficult to model. As a result, various methods have been developed. This makes the selection of a model an ambiguous process. This study focuses on three models: QMOM, DQMOM, and the constant sectional method. They serve as baseline models for the families of moment and sectional methods. This work aimed to provide some basis from which comparisons could be made among methods. A focus was placed on providing a more thorough comparison among families of solutions rather than simply the runtime and results. To do so, each method was formulated in a manner that would provide comparable run-times and solution accuracies. Coagulation was selected as the phenomenon of interest due to the mathematical difficulties inherent to the process. During the model formulation, assumptions and requirements of each method were identified. A focus was also placed on any algorithms that each method may rely on. Then, the methods were implemented for several benchmark cases. This bottom-up approach allowed for a more involved comparison among the methods. All methods were first optimized for a simple scenario with a known solution. During this process, controllable parameters were varied and resulting impacts on the models were monitored. QMOM proved to be the least sensitive to parameter variation while sectional method performance heavily depended on proper parameter selections. Parameter selections that caused model failure were also identified and were typically able to be attributed to either being a fault in the model or a result of machine arithmetic. Further benchmarking of the optimized models showed DQMOM outperforming QMOM for equivalent quadrature point selections. The sectional method proved capable of providing decent moment values while providing insight as to what the distribution looks like. If such details are desired, then it is the obvious choice. Else, moment methods are preferable when bulk properties suffice. QMOM is good for quick calculations and has a decent application range. DQMOM, however, is the more robust and accurate of the two, with a vastly larger application range. Its main downside is more user intuition being required for proper implementation.Item Experimental and computational characterization of strong vent flow enclosure fires(2011-08) Weinschenk, Craig George; Ezekoye, Ofodike A.; da Silva, Alexandre K.; Engelhardt, Michael D.; Howell, John R.; Raman, Venkatramanan; Nicks, RobertFirefighters often arrive at structures in which the state of fire progression can be described as ventilation-controlled or under-ventilated. This means that inside the enclosure the pyrolyzed fuel has consumed most, if not all of the available oxygen, resulting in incomplete combustion. Under-ventilated (fuel rich) combustion is particularly dangerous to occupants because of the high yield of toxins such as carbon monoxide and to firefighters because once firefighters enter the structure and introduce oxidizer, the environment can rapidly change into a very dangerous, fast burning condition. The fuel load in many compartment fires would support a several megawatt fire if the fire were not ventilation controlled. In the process of making entrance to the fire compartment, firefighters will likely provide additional ventilation paths for the fire and may initiate firefighting tactics like positive pressure ventilation to push the hot flammable combustion products out of the attack pathway. Forced ventilation creates a strongly mixed flow within the fire compartment. Ventilation creates a complex fluid mechanics and combustion environment that is generally not analyzed on the scale of compartment fires. To better understand the complex coupling of these phenomena, compartment scale non-reacting and reacting experiments were conducted. The experiments, which were conducted at The University of Texas at Austin’s fire research facility, were designed to gain insight into the effects of ventilation on compartment thermal characteristics. Computational models (low and high order) were used to augment the non-reacting and reacting experimental results. Though computationally expensive, computational fluid dynamics models provided significant detail into the coupling of buoyantly driven fire products with externally applied wind or fan flow. A partially stirred reactor model was used to describe strongly driven fire compartment combustion processes because previously there was not an appropriate low dimensional computational tool applicable to this type of problem. This dissertation will focus on the experimental and computational characterization of strong vent flows on single room enclosure fires.Item A toolkit for characterizing uncertainties in hypersonic flow-induced ablation(2010-12) Anzalone, Reed Anthony; Ezekoye, Ofodike A.; Upadhyay, Rochan R.A one-dimensional, quasi-steady ablation model with finite rate surface chemistry and frozen equilibrium pyrolysis gases is developed and discussed. This material response model is then coupled to a film-transfer boundary layer model to enable the computation of heat and mass transfer to and from the ablating surface. A shock model is outlined, as well, and all three components are then coupled together to form a stand-alone ablation code. The coupled models in the code are validated with respect to arcjet experiments, and comparisons are drawn between the ablation code and the unsteady ablation code Chaleur, as well as other computations for a graphite ablator in an arcjet. The coupled code is found to compare very well to both the experimental results and the other calculations. It is also found to have unique computational capabilities due to the use of finite-rate surface chemistry. Finally, uncertainty propagation using the quadrature method of moments (QMOM) is discussed. The method is applied to a number of simplified sample problems, for both univariate and multivariate scenarios. QMOM is then used to compute the uncertainty in an application of the coupled ablation code using a graphite ablator. The results of this study are discussed, and conclusions about the utility of the method as well as the properties of the ablation code are drawn.