| A predictor-corrector technique for solving inverse convection problems was developed, tested, and re fined. The methodology was tested against three inverse problems: inverse plume in a crossflow, inverse jet in a crossflow, and inverse plume in a cavity. The goal of the inverse plume in a crossflow was to solve for the strength and location (x, y) of the source. After refinement, the methodology was able to predict all three goals, utilizing three sample points to within 2.5%. Error analysis demonstrated that three sample points was unable to tolerate any simulation-experimental error. Therefore, when handling experimental data, an increase in the number of sample points is required, to a minimum of five. The error analysis also showed that the methodology, with five or more sample points, is remarkably stable in its prediction capability. The location prediction was minimally affected, less than 0.1%, by an artificial error of 10%. The goal of the inverse jet in a crossflow was to solve for the strength (velocity and temperature) and location (x, y) of the source. After testing, the methodology was not able to predict all four goals. The elevation location of the jet needed to be known to adequately solve the inverse problem. The methodology was able to predict the source velocity and temperature to within 10% and 3.3% respectively. The goal of the inverse plume in a cavity was to find the strength and location (x, y) of the source. Sensitivity analysis demonstrate it is very difficult, if not impossible, to resolve source location using this methodology. The method was able to predict the source strength within 5% using only one sample point. With future work, this approach could be extended to applied areas of interest, such as environmental flows, room fires, and thermal management systems. |
