Chance constrained models for air pollution monitoring and control.
详细信息
文摘
This dissertation develops new methodology for dealing with skewed distributions of matrix coefficients in chance constrained programming applications to air pollution monitoring and control. The investigations were directed to examination of transfer of benzene from point sources in the Houston Ship Channel area to selected monitoring sites. It has been observed that distribution of the benzene readings at the receptor sites exhibited nonnormal behavior mainly due to different annual meteorological conditions that are often encountered. Standard management science techniques applied to the problem of air pollution control fail to capture this kind of behavior which, in turn, often leads to less accurate interpretation of costs involved with emissions reductions as well as risks to human health and the environment. A series of experiments use the simulated data based on Gaussian diffusion model which describes the transfer of pollution from sources to receptors. Statistical estimations of parameters and implementation of the newly developed chance constrained programming models are then presented. This is followed by interpretations of solutions and comparisons across the models using postoptimality analysis. Joint chance constraints formulations are also considered since they provide more flexibility with respect to choices of risk levels at particular monitoring stations. Finally, the technique of data envelopment analysis enabled us to rank emission sources according to efficiency of their pollution prevention activities. An extensive review of past studies that are to be found in the literature on mathematical programming applied to air pollution monitoring and control is also provided.