Joint production and maintenance operations in smart custom-manufacturing systems

Machines in custom manufacturing environments with IoT (Internet-of-Things) capability are predicted to be pervading enterprises. However, there is a need to develop new algorithms that reap the benefits of such technologies. We consider a system where jobs with stochastic workloads arrive to a machine in an arbitrary fashion and upon arrival, their workload is revealed (enabled by IoT). The tool on the machine gets used up based on the speed at which the jobs are processed. Knowing that tool-replacement consumes a significant amount of time, we want to develop online algorithms that maximize the capacity of the machine by determining: (i) the speed at which each job is processed; and (ii) the epoch when the tool is replaced. We provide online approaches that leverage the ability to reveal workload in real-time and effectively balance future uncertainties. We derive asymptotic bounds for the online algorithm performance and show using numerical experimentation that a little revealed information could result in a tremendous improvement in performance. Our online algorithms also work under realistic conditions of non-stationary batch arrivals and correlated workloads. Our work opens up research directions for a variety of operational settings that may benefit from revealing stochastic quantities by mining information.