Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Mining Engineering

Committee Chair

Aaron Noble

Committee Co-Chair

Keith Heasley

Committee Member

Mark Sinedlar


The chemical treatment of mine-influenced waters is a longstanding environmental challenge for many coal operators, particularly in Central Appalachia. Mining conditions in this region present several unique obstacles to meeting NPDES effluent limits. Outlets that discharge effluent are often located in remote areas with challenging terrain where conditions do not facilitate the implementation of large-scale commercial treatment systems. Furthermore, maintenance of these systems is often laborious, expensive, and time consuming. Many large mining complexes discharge water from numerous outlets, while using environmental technicians to assess the water quality and treatment process multiple times per day. Unfortunately, this treatment method when combined with the lower limits associated with increased regulatory scrutiny can lead to the discharge of non-compliant water off of the mine permit. As an alternative solution, this thesis describes the ongoing research and development of automated protocols for the treatment and monitoring of mine water discharges. In particular, the current work highlights machine learning algorithms as a potential solution for pH control.;In this research, a bench-scale treatment system was constructed. This system simulates a series of ponds such as those found in use by Central Appalachian coal companies to treat acid mine drainage. The bench-scale system was first characterized to determine the volumetric flow rates and resident time distributions at varying flow rates and reactor configurations. Next, data collection was conducted using the bench scale system to generate training data by introducing multilevel random perturbations to the alkaline and acidic water flow rates. A fuzzy controller was then implemented in this system to administer alkaline material with the goal of automating the chemical treatment process. Finally, the performance of machine learning algorithms in predicting future water quality was evaluated to identify the critical input variables required to build these algorithms. Results indicate the machine learning controllers are viable alternatives to the manual control used by many Appalachian coal producers.