RELIABILITY ASSESSMENT OF A MULTI-STATE SYSTEM USING DISCRETE TIME MARKOV CHAIN

Reliability assessment of a working system is perfunned to identuy the most likely
fuLiures and time-to-fuilure so that appropriate actions can be planned to diminish the
effucts of the fuilures. Traditional model of reliability assessment assumes every
working system would P.ave two states wP.ich are the st.ate du.ring perfectly workiP.g
and the complete fuilure state. However, systems that exhibit multi-state behavior
rr.ay have 1iPite number of fuilure rate V>il.~h are called multi-state system (MSS). The
MSS system will degrade from a perfect working system to certain minor fuilure
states befure it complete~' .taiL Hence, new approach is needed to predict tire
reliability of such system. This report contains the selected MSS analysis as the new
approach to assess the reliabiL;cy of a MSS system and the findings about the selected
method which is Discrete-Time Markov Chain (DTMC) analysis. The data was taken
from UTP Gas District CooliP.g (GDC) production report fOcusing on the perfunnance
of a gas turbine in terms of kW. The perfunnance data was clustered into some
perfunnance states and the state transition probabilities were estimated. From tlte
estimation, reliability function and distribution parameter were obtained to be used to
calculate tlte Mean Time Between Failure (M1BF) of the system. At the end of the
project, the reliability of the MSS that predicted using DTMC analysis was compared
to the reliability predicted using traditioml method wpjch was the exponential
distribution method. The analysis shows that DTMC analysis has better prediction
th.an the exponential distdbutjon method. Moreover, by exploiting the state transition
probabilities estimation process, the change of operation demand as well as
Preventive Maintenance planning could be included in the analysis. Briefly, MSS
analysis gave better reliability prediction of the MSS and the behavior of the system
could be analyzed.