Threshold-based method for elevating the system's constraint under theory of constraints

Abstract

The principal tenet of theory of constraints (TOC) is that there is at least one constraint in each system that limits the ability of achieving higher levels of performance relative to its goal. Maximum utilisation of the constraint leads to maximum output of the system. However, activation of a non-constraint resource at 100% of its capacity does not increase output. Therefore, some resources are not fully utilised. In this paper, the authors use the left capacity of a non-constraint resource (NC) to elevate the system's constraint. It is assumed that the capacity-constrained resource (CCR) is a continuous time Markov process having a two-dimensional state space. The work in the NC is interruptible, allowing a worker in the NC to switch to CCR. The switch from NC to CCR would occur when the queue of waiting parts in the CCR becomes 'too long' and vice versa, when there are few parts in the CCR. Returning to the NC from the CCR may require some 're-orientation time' on the part of the switched worker. The goal is to find the maximum output of CCR subject to the time-average number of workers in the NC must be greater than a pre-specified value.