New approximate solution of CONWIP and a method for determining appropriate level of WIP

Abstract

CONWIP (Constant Work in Process), which is a simple pull production control method keeping WIP (Work in Process) of a whole production line constant, has been considered a better alternative to Kanban in some ways. However, CONWIP involves a trial-and-error process to determine the appropriate constant WIP level by commonly employing MVA (Mean Value Analysis) for its performance measure estimation. This trial-and-error approach with the recursive nature of MVA might prohibit industries from widely implementing CONWIP in practice. This paper proposes a new simpler approximate solution of CONWIP by focusing on the bottlenecks in the system. The precision of the proposed approximation method is evaluated by simulation. It is found that the proposed method tends to outperform MVA by far, particularly in the range of relatively small process variability. In addition, a practical method determining the appropriate level of WIP is proposed to maximize TH (throughput) based on the current system performance. Moreover, through the formulation incorporating the influence of process variability such as machine failures, practical implications are derived to accommodate CONWIP to situations such as unstable production ramp-ups or overseas productions where Kaizen could not be readily expected.

Keywords

• tps
• jit
• conwip
• mva
• queueing theory
• pull production
• kanban
• bottle-neck
• process variability