A Numerical Model for Ash Deposition Based on Actual Operating Conditions of a 700 MW Coal-Fired Power Plant: Validation Feedback Loop via Structural Similarity Indexes (SSIMs)

Abstract

The combustion of coals will result in significant ash-related issues, which will ultimately lead to the efficiency loss of coal-fired utility boilers. While there have been numerous attempts to predict ash deposition dynamics using numerical approaches, the majority of these models were constructed using experimental data from pilot-scale furnaces and without integration with combustion models. Therefore, the current study collects meaningful power plant data from ash sampling activities at one of Malaysia's 700 MW sub-critical coal-fired power plants, enabling the ash deposition behavior in a real coal-fired utility boiler to be adequately captured and converted into a reliable ash deposition numerical model. The validation feedback loop of the ash deposition model was run using in-situ measurement data (ash sampling picture) and the actual power plant operating conditions during the ash sampling activities. The image processing algorithm was used to determine the degree of similarity between the actual ash sampling image and the predicted ash deposition image from the numerical model. Prior to the validation feedback loop, the overall numerical model (solver, combustion, turbulence, radiation) was successfully validated with the FEGT from the actual power plant, revealing a difference of less than 5 %. The current study found that the baseline ash deposition model (created from experimental data) underestimates the quantity of ash deposition gathered. The validation feedback loop of the baseline ash deposition model successfully established a new set of impaction efficiency constants, which increased the similarity of the images between the actual and predicted ash depositions. The current study's drawback, however, is mostly in the validation basis, which is largely qualitative in nature. Although the Structural Similarity Index (SSIM) value is useful for comparing the similarity of images between actual and predicted ash depositions, a more quantitative measurement that can provide extra meaningful data points and higher accuracy on the deposited ash is preferable. However, based on this modified version of the ash deposition model, the agreement is found to be satisfactory in terms of gaining a rudimentary insight of the ash deposition behavior in a coal-fired boiler.