This thesis develops a computational model of a household biomass cookstove for use by the 2.7 billion people currently cooking with biomass in developing countries. This traditional practice results in a number of detrimental effects to health, ecosystems, and global climate, including indoor air pollution, which is responsible for 4 million premature deaths per year and represents the second leading cause of death for women globally. Despite several decades of engineering improved biomass cookstoves, to date there has been relatively little research regarding the computational modeling of such widely used devices. Development of a flexible, comprehensive, computationally inexpensive, and coupled model with detailed experimental validation will allow the design of cookstoves to benefit from the same engineering tools used in design for the developed world.

Through investigation of techniques employed in the literature, a flexible steady-state model is developed for a single pot, natural draft, shielded fire stove burning traditional wood fuel to predict the fluid flow and heat transfer characteristics of the system, which is separated into the (1) bed, (2) flame, and (3) heat transfer zones. The model incorporates 15 design parameters, including 10 geometrical, 2 material, and 3 operating variables spanning the region of interest for household biomass cookstoves. The model is validated from a unified experimental data set developed from three studies in the literature that report thermal performance characteristics in terms of design characteristics. The data set includes 63 data points incorporating variation of all 15 parameters and is shown to be consistent and supportive of qualitative thumb rules regarding the effects of stove geometry, operating variables, and material on overall thermal performance. Several adjustable coefficients and convective heat transfer correlations are fitted to the data using particle swarm optimization. The model utilizes contracting mapping to predict air flow and temperature profile, resulting in 94% of the data points predicted within 5% of measured thermal efficiency and a L2 norm error of 3%. The model can be used to optimize heat transfer efficiency given local constraints for design and allows for conceptual design and sensitivity analysis without the need for extensive experimentation. In addition, the temperature and velocity profiles, location and magnitude of losses, and heat transfer contributions through various modes and regions of the pot are detailed to lead to greater understanding of the cookstove system.