Sum learnings:
The anathema to operations is variability and uncertainty. It is impossible to perfectly crystal ball demand, therefore we must tailor our supply to expected demand. First, attack the closed system within the current period; understand throughput (using TOC) to determine whether process is capacity or demand constrained. If the former, cycle time becomes the game. Waiting time within cycle time (typically modeled using Kingman's) is the largest contributor. Reduce arrival and processing variability while keeping utilization in check. Predictive maintenance increases resource availability therefore increasing throughput. Cycle time can be malleable with a fixed throughput, shortest processing time decreases inventory in system (little's). Obviously analysis/implementation is context dependent. SPT in hospitals would be unethical.
Supply chain: conflicting incentives leads to variability -> bullwhip effect. Hence the argument for vertical integration (quality control an incentive too)
Cool models: Newsvendor w standard normal, six sigma