Monika Kuznia

Date of Award

January 2021

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Chemical Engineering

First Advisor

Gautham Krishnamoorthy

Abstract

Oxy-combustion is one of the most competitive technologies for retrofitting existing pulverized coal-fired power plants to facilitate carbon capture and sequestration processes to reduce fossil fuel-derived carbon dioxide emissions into the atmosphere. However, one of the obstacles hindering the widespread commercialization of this technology is the need to recirculate large volumes of flue gas to the boiler. Second generation atmospheric pressure oxy-combustion technologies have been developed to reduce the volume of recirculated flue gas by using high oxygen concentrations in the inlet oxidizer streams. However, recent experiments have shown that an increased ash deposition propensity is associated with these oxygen-enriched environments. This increase has been primarily attributed to aerodynamic effects, namely the higher ash concentrations associated with the reduction in flue gas volumetric flow rates and ash particle size distribution variations possibly due to a more intense combustion at the higher temperatures in the oxygen-enriched environments. Since the Stokes number and impaction efficiencies both decrease as velocity decreases for a fixed particle size, ash deposition rates under oxy-combustion conditions should be lower than those under air combustion conditions. The primary hypothesis of this thesis is that the ash particle size distribution variations is the aerodynamic effect that most influences numerical predictions of ash impaction and outside ash deposition rates.In order to test this hypothesis, 25 highly resolved numerical simulations of well-characterized pulverized coal combustion tests that were performed at The University of Utah under air combustion, first generation oxy-combustion, and second generation atmospheric pressure oxy-combustion conditions were completed. Two different coal types from the combustion tests were examined in these numerical simulations: a non-swelling, sub-bituminous Powder River Basin coal and a swelling, bituminous Utah Sufco coal. The measured outside ash deposition rates from the Utah Sufco coal combustion tests were approximately 5x larger than the outside ash deposition rates from the Powder River Basin coal combustion tests. Additionally, the outside ash deposition rates of the second generation atmospheric pressure oxy-combustion tests were approximately 2x and 3x larger than the outside ash deposition rates of the air combustion tests for the Powder River Basin coal and the Utah Sufco coal, respectively. The outside ash deposition rates were measured at a location within the experimental apparatus where the flow was predominantly laminar and complete combustion had been achieved. In all 25 numerical simulations, predictions of temperature, velocity, and flue gas volumetric flow rate agreed well with the respective experimental measurements and estimates. The first 13 numerical simulations were completed for a sensitivity study of Powder River Basin coal combustion to investigate the effect of changing 3 different parameters on predicted ash impaction rates: 1) the number of bins, or resolution, specified for the inlet coal particle size distribution model, 2) the model for the coal density-diameter variations (shrinking core versus shrinking sphere), and 3) the inlet coal particle size distribution model (to account for the significant variations in the measured sieve mass fractions of the larger sized particles). The following combustion conditions were investigated: AIR, 27 vol% oxygen with 73 vol% carbon dioxide (OXY27), and 50 vol% oxygen with 50 vol% carbon dioxide (OXY50). The ability of these numerical simulations to accurately predict the outside ash deposition rates from the Powder River Basin coal combustion tests was also evaluated. The predicted ash impaction rate showed an obvious sensitivity to all three numerical simulation parameters, which supported the hypothesis of this thesis and reaffirmed the recent findings that numerical ash impaction rate predictions are critically dependent upon numerical ash particle size distribution predictions. 120 bins were deemed necessary for accurately resolving the inlet coal particle size distribution, which is significantly larger than the 40 to 80 bins that have been reported in the ash deposition literature. The measured trends in the outside ash deposition rates from the Powder River Basin coal combustion tests could not be accurately predicted (qualitatively and quantitatively) by these numerical simulations despite using established best Reynolds-averaged Navier-Stokes simulation practices. This was attributed to the overestimation of the impaction and deposition rates on the leeward side of the probe likely due to the unavailability of an accurate fly ash particle size distribution. The accuracy of the predictions may be improved with the availability of fly ash particle size distribution data in the 10-400 μm range and incorporating/modeling this distribution in the numerical simulations. The remaining 12 numerical simulations were completed to investigate the effect of changing 3 different parameters on predictions of outside ash deposition rates for the Utah Sufco coal: 1) the swelling coefficient of the combusting particle, 2) the spread parameter of the inlet coal particle size distribution model, and 3) the mean diameter of the inlet coal particle size distribution model. The following combustion conditions were investigated: AIR, 27 vol% oxygen with 73 vol% carbon dioxide (OXY27), and 70 vol% oxygen with 30 vol% carbon dioxide (OXY70). Again, the predicted ash deposition rates were noticeably affected by changing any of these three parameters. 120 bins were deemed necessary for accurately resolving the inlet coal particle size distribution. Utilizing accurate inlet coal particle size distribution measurements in the numerical simulations could not predict the outside ash deposition rates from the Utah Sufco coal combustion tests, which further supported the hypothesis of this thesis and reaffirmed the recent findings that numerical ash deposition predictions are critically dependent upon numerical ash particle size distribution predictions. However, the measured ash deposition trends could be replicated successfully when the swelling coefficient and spread parameter were adjusted such that the measured and simulated ash deposit particle size distributions matched. For the range of velocities investigated in this research (0.2-1 m/s), measurements of the fly ash particle size distribution in the 10-400 μm range were identified as a critical variable influencing the deposition rate predictions in the numerical simulation of these experimental combustion tests.

Download

Share

COinS