After 10 years of research and development between CASTLE Lab at Princeton University and Norfolk Southern Railway, PLASMA was put into production at Norfolk Southern to perform strategic simulations and operational planning in 2008. PLASMA (Princeton Locomotive Allocation and Shop Management System) is an optimization engine based on recently developed strategy called high-dimensional approximate dynamic programming. PLASMA was integrated within NS into a large integrated planning system called LARS (Locomotive Assignment and Routing System) developed by NS. In this talk, Warren Powell is going to review the underlying modeling and algorithmic technology and describe how a single engine was used for strategic, tactical and operational planning of locomotives for NS. Then, Clark Cheng will describe what it took to get the modeling running at NS, and then keep it running over the 14+ years since the LARS fleet sizing application was deployed.