Date of Graduation


Document Type



Abandoned mine lands are areas affected by past coal mining activities that often pose significant water quality problems from acid mine drainage (AMD). Weather, a primary determinant of AMD, is inherently stochastic. The economic costs of treating AMD problems are an issue of importance to policymakers and citizens. This study develops a spatial model to devise cost effective plans for treating AMD in a watershed when pollution loads are stochastic and the level of certainty of treating all of the AMD is allowed to vary. Mixed integer and stochastic chance constrained programming methods are applied to a model of AMD treatment cost under conditions of uncertainty in a spatial context. The AMD loadings are stochastic variables in the model and treatment options are based on specified probability levels of certainty represented in a chance constrained programming model. The results give the selection of segments to treat, the technology to be used, and the treatment level required by probability level of treating the full AMD problem within the watershed. Paint Creek Watershed serves as a case study for this dissertation. Findings are based on cost estimates for AMD treatment technologies obtained from past studies. The water quality data, posed as net acidity loadings by segment on a daily basis, were based on output from the Total Acid Mine Drainage Loading model developed by Stiles. The economic model considers two chemical treatment technologies to treat the AMD problem, ammonia and hydrated lime, and considers the implications of four probability target levels of treating all of the net acidity loading; 50%, 80%, 90% and 95%. Ammonia is the least cost AMD treatment alternative for the Paint Creek watershed in all cases. The annualized costs for treating the maximum anticipated acid load increase by 90%, 195% and 245% when acid loads treated increase from the average to the 80%, 90% and 95% probability levels, respectively. The increased costs arise from treating more segments to meet the higher targets, increased cost from higher loads in all segments, and larger plant size requirements to meet the treatment goals in some segments.