As a graduate student in the Environmental Assessment and Optimization group at Stanford, Kemp became interested in developing a new method for finding natural gas leaks. Before mercaptan has been added, natural gas is an invisible odorless gas. Leaks often go unnoticed with significant impact on both climate and profit. There was interest in several classes of detector at the time: continuous monitoring of methane concentration using distributed detectors, manual inspection of natural gas infrastructure with a flame ionization detector, periodic aerial surveys of methane concentration using drones, and periodic surveys using infrared cameras. Before investing effort in a particular technology, Kemp needed a way to compare the disparate approaches. To that end, he developed a model that would become the Fugitive Emissions Abatement Simulation Toolkit (FEAST). The model garnered enough interest that Kemp ultimately abandoned efforts to develop any particular method in order to focus on developing the model for others to use.
Kemp published his masters thesis describing FEAST in 2015, and moved back to Alaska. However, interest in FEAST continued to grow and Kemp accepted several contracts to improve and expand the model–first to support analysis of a new methane detection method for an aerospace startup company, then to transfer the model from the Matlab programming language to Python to enable a collaboration between Stanford and Sandia National Lab, and most recently to adapt FEAST to enable regulators to use it to determine equivalence between new and existing leak detection and repair programs. Documentation and results from FEAST were published by Environmental Science and Technology in 2016: https://pubs.acs.org/doi/10.1021/acs.est.5b06068.