Reliability and condition-based maintenance modeling for systems operating under performance-based contracting
Introduction
After-sales support and maintenance services have become a major source of revenue, profit and competitive advantage in manufacturing industries (Cohen, Agawal, & Agrawal, 2006). This is especially true for those firms selling capital-intensive products, such as aircraft engines, wind turbines, and oil pipelines, among others (Dennis & Kambil, 2003). Customers of these systems are more concerned with the performance outcome (e.g. reliability, availability), and less sensitive to the costs (Xiang, Zhu, Coit, & Feng, 2017). In recent years, a new service paradigm named as performance-based contracting (PBC) has emerged. This new form of support contract is often referred to as performance-based logistics (PBL) (Randall, Nowicki, & Hawkins, 2011) in the defense sector and Power by the Hour (PBH) (Neely, 2008) in the commercial sector. Different from the traditional fixed-price and cost-plus contracts, under PBC, service providers are paid for the realized system performance (Guajardo et al., 2012, Kim et al., 2007). A detailed review of various issues related to PBC can be referred to Selviaridis and Wynstra (2015).
Since PBC rewards the service provider based on the same performance outcome that the customer cares about, it motivates the problem of how to plan a viable maintenance policy to both improve the system reliability/availability at a lower cost and gain profitability under PBC. Different maintenance policies for complex industrial systems under support contracting approach have been received growing attention (Qiu et al., 2018, Sultana et al., 2013). Using modern sensor technology, condition-based maintenance (CBM) suggests “necessary” preventive maintenance actions based on the inspection of working conditions. See Alaswad and Xiang (2017) for an excellent review of mathematical models and optimization approaches of CBM. In practice, many dynamic systems are not only subject to degradation with operational age but also to sudden shocks which can lead to a complete system failure (Singpurwalla, 1995). The stochastic degradation-threshold-shock (DTS) models (Lehmann, 2009, Lemoine and Wenocur, 1985, Singpurwalla, 1995) are commonly applied to describe these two competing failure causes. Reliability analysis and CBM policies planning based on DTS models have been extensively investigated (Caballé et al., 2015, Che et al., 2018, Gao et al., 2019, Huynh et al., 2012, Liu et al., 2016, Peng et al., 2010, Rafiee et al., 2014, Rafiee et al., 2016, Shafiee et al., 2015, Shen et al., 2018, Tang et al., 2015, Zhao et al., 2018, Zhou et al., 2016, Zhu et al., 2015). In particular, increasing attention has been paid to the determination of optimal maintenance policies within the framework of PBC (Qiu et al., 2017, Wang et al., 2019, Xiang et al., 2017, Yang et al., 2019).
The DTS models applied in the above-mentioned literature assumed stationary degradation process for binary-state systems. Whereas a system may experience multiple states during its lifecycle; e.g., “normal”, “degraded”, “failed”, etc (Bian & Gebraeel, 2014). For such systems, their degradation processes are non-stationary and the degradation rates are not static but dependent on the system states. There has been research focusing on maintenance optimization for multi-state systems (Ma et al., 2020, Wei et al., 2019, Yang et al., 2017, Yang et al., 2017, Zhang et al., 2016), assuming constant degradation rate in each state for all units. However, due to heterogeneous working conditions or variations in the raw materials, degradation for systems within the same population may exhibit different degradation rates. Ignoring the effect of heterogeneity on the system deterioration may result in inaccurate reliability and cost estimation, and an inferior maintenance decision.
The present paper considers single-unit and non-repairable systems whose failures are due to the competing causes of degradation and shocks. The main objective is to develop system reliability and CBM models within the framework of PBC from the service provider’s perspective. There exists three possible states for a system, which are normal, degraded and failed. Transition from the normal state to the degraded state accelerates both the degradation and shock processes, resulting in increased degradation rate and higher shock failure intensity. Degradation is modeled as a two-stage inverse Gaussian (IG) process with random effects, allowing for unit-specific heterogeneity within the same population. Sudden shocks arrive according to a doubly stochastic Poisson process (DSPP). Based on the reliability model developed by incorporating the degradation-based and shock-based failure processes, we formulate a CBM model derived from the inspection-based preventive replacement policy. Along with the assessment of the long-run maintenance cost rate, the average availability of the system as the performance metric is evaluated. Adopting the profit-centric approach under PBC, the optimal maintenance policy is determined so as to maximize the expected profit rate to the service provider.
The main contributions of the present study are:
- •
Extending DTS models through investigating the impact of system state transition on the evolution of degradation and shock processes;
- •
Establishing a two-stage IG degradation model accounting for unit-specific heterogeneity, and analyzing system reliability based on competing failure processes modeling;
- •
Developing a CBM model and evaluating system performance outcome (average availability) based on inspection-based preventive replacement policy;
- •
Providing service providers for choosing the optimal maintenance strategy adopting the profit-centric approach under PBC.
The remainder of this article is organized as follows. Section 2 lists the assumptions and notations used in the reliability and maintenance modeling studies. In Section 3, we develop the system reliability model based on two competing failure processes due to degradation and sudden shocks. Section 4 proposes the CBM optimization model based on the periodic inspection-replacement policy under PBC. Section 5 presents a numerical example along with sensitivity analysis illustrating the reliability model and maintenance policy. Finally, Section 6 summarizes the main conclusions and provides suggestions for future research.
Section snippets
System description
We consider a single-unit and non-repairable system which goes through natural degradation along with sudden shocks. The system is regarded as failed when the overall degradation is beyond a pre-determined threshold or when a sudden shock arrives although the degradation has not reached the threshold level, whichever occurs first.
System reliability analysis
In this section, reliability analysis is performed considering the two competing failure processes due to degradation and sudden shocks.
Maintenance modeling and optimization under PBC
A maintenance policy which minimizes the cost but ignores the system performance outcome may not be effective. In this section, we formulate the CBM model derived from inspection-based preventive replacement policy. That is, based on what the inspection reveals, preventive replacement is performed if the degradation is beyond a threshold or corrective replacement is performed if the system has failed. To determine the optimal maintenance policy, we adopt a profit-centric approach commonly
Numerical example and sensitivity analysis
In this section, we provide a numerical example and sensitivity analysis to demonstrate the applicability of the proposed reliability and CBM models. The values of several model parameters used in the analysis are adopted from existing literature (Chen et al., 2015, Tanner and Dugger, 2003, Xiang et al., 2017, Yang et al., 2017) with slight modifications. Others are assumptions based on typical and plausible values.Solution procedure based on improved PSO algorithm Step 1: Input model
Conclusions
In this paper, we propose new reliability and CBM models for single-unit and non-repairable systems operating under PBC. The two competing failure causes are internal degradation and sudden shocks. The lifecycle evolution of a system involves three states that are normal, degraded and failed. Considering state transition, a reliability model is developed by incorporating degradation modeled as two-stage IG process with random effects characterizing heterogeneity across systems, and sudden
Acknowledgment
This research is supported by the National Natural Science Foundation of China (Grant No. 71801171).
References (53)
- et al.
A review on condition-based maintenance optimization models for stochastically deteriorating system
Reliability Engineering & System Safety
(2017) - et al.
Analysis of the reliability and the maintenance cost for finite life cycle systems subject to degradation and shocks
Applied Mathematical Modelling
(2017) - et al.
A condition-based maintenance of a dependent degradation-threshold-shock model in a system with multiple degradation processes
Reliability Engineering & System Safety
(2015) - et al.
Reliability modeling for dependent competing failure processes with mutually dependent degradation process and shock process
Reliability Engineering & System Safety
(2018) - et al.
Condition-based maintenance using the inverse Gaussian degradation model
European Journal of Operational Research
(2015) - et al.
Reliability modeling for degradation-shock dependence systems with multiple species of shocks
Reliability Engineering & System Safety
(2019) - et al.
A condition-based maintenance policy for stochastically deteriorating systems
Reliability Engineering & System Safety
(2002) - et al.
Modeling age-based maintenance strategies with minimal repairs for systems subject to competing failure modes due to degradation and shocks
European Journal of Operational Research
(2012) Joint modeling of degradation and failure time data
Journal of Statistical Planning and Inference
(2009)- et al.
An imperfect maintenance policy for mission-oriented systems subject to degradation and external shocks
Computers & Industrial Engineering
(2016)