Simulation of partially observed Markov decision process and dynamic quality improvement

Computer simulation is used to compare three quality policies. The first policy is ‘do-nothing’. The second is an appraisal policy. The third policy includes prevention that leads to process quality improvement. The simulation model is based on a Partially Observed Markov Decision Process (POMDP). The unobserved states of the process depend on the failure rate, λ. The observed process output is the number of conforming and nonconforming products. The process performance is measured by quality costs per unit. The simulation language used is SLAM II. The power of using computer simulation to model the dynamics of process quality improvement is discussed.