Researchers used numerical simulation to model unburned carbon in fly ash from a coal-fired boiler. They created a support vector machine model that accurately predicts unburned carbon content with only a 1.01% error. This method offers a new way to monitor combustion processes online, even though it requires more computational time.