A decision model for corrective maintenance management

Proper management of maintenance offers many companies significant potential for improving productivity and profitability. Traditional management thinking regards maintenance costs as accidental, rather than planned and controllable. Additionally, research in maintenance management has focused on preventive maintenance and has ignored corrective maintenance even though the latter is also considered to be a critical activity in industry. This study proposes a decision model that could assist in a comparative evaluation of alternative corrective maintenance policies. This decision model consists of a simulation model and economic analysis. The simulation model predicts inventory costs and delivery performance of a corrective maintenance policy in various production systems. Based on simulation results, an economic analysis, consisting of a net present value model and breakeven models, determines the economic value of alternative maintenance policies. A detailed example is offered to evaluate two particular correciive maintenance policies (machine redundancy and worker flexibility) although the decision model can be applied to other options. The results of the example demonstrate the decision model's capability to assist managers in selecting the best corrective maintenance policy.     

Plant-level maintenance decision support system for throughput improvement

In the current global scenario of extreme competition, factors such as productivity, availability, quality and cost of operations play a vital role in the success of a company. A critical component relating to all of the above is maintenance. The conventional maintenance decision support systems have primarily focused on maximising the gains of a single machine system. However, a real life application usually consists of multiple machines, and the operational level decisions are more complex. In this paper, an on line plant-level maintenance decision support system (PMDSS) is developed by combining the short term and long term decision making process to improve the overall system performance while continuously attempting to maximise immediate profits in the short term. The PMDSS works towards two basic aims: (1) unplanned downtime reduction by predicting the remaining useful life of the machines, and (2) efficient utilisation of the finite maintenance and production resources through identifying the throughput-critical machines. The benefits of this approach are presented by considering an industrial case study of an automotive assembly line. The results obtained using this PMDSS approach shows a big throughput improvement as compared to the conventional maintenance policies.     

Control Limit Policies for Warranty, Maintenance and Upgrade of Software Systems

Software systems continuously undergo change, mostly in response to customers' needs and expectations, which themselves change from one period to the next. Although the cost of these changes may be many times the original system cost, there is no framework for their analysis and for bringing customers' views into the decision processes. In this paper, we develop a Markov decision process model for warranty, maintenance, and upgrade of software systems, using the index of customer satisfaction as the state variable. We identify optimal and near-optimal policies within the space of control-limit policies     

Research into the use of statistical quality control in british manufacturing industry-part II

The work described in this paper is part of a large. Government-funded investigation into the usage of statistical methods of quality control (SQC) in British manufacturing industry. This paper reports the results of work in seventeen companies on the implementation of SQC. The results are summarized by the use of evaluation scales and attributes, and measures of success and barriers are identified. Some general conclusions are drawn which point to a methodology for the introduction of SQC.     

A portfolio approach to maintenance: a case study of a residential estate

The choice between repairing and replacing a defective piece of equipment is an economic decision that is faced by all maintenance managers, including housing estate managers. Such decisions need to be made within the limits and constraints set by maintenance expenditure budgets and by manpower availability. Our particular problem is concerned with the development of a maintenance policy for a residential estate of the Hong Kong Housing Authority. Our approach is to treat the numerous housing systems as a portfolio and to exploit flexibilities in performing or delaying the repair/replacement of these systems. The cost of a repair/replacement plan for the portfolio is formulated as an integer programme and genetic algorithms (GAs) are employed to generate optimal and sub‐optimal solution plans. The novel features of the approach are the model developed and use of GAs in this particular optimization context. The results and discussion of the case study will help practitioners to better understand the difficulties involved in collecting relevant cost data and in formulating repair/replacement plans for a group of buildings. Copyright © 2007 John Wiley & Sons, Ltd.     

Bayesian real options control chart

In this paper, we present the concept of a novel control chart, which uses economic considerations within the real options–inspired framework together with the principles of Bayesian statistics to produce a continuously updated estimate of the parameters of the actual process, and thus to decide whether to continue running the process or to recalibrate it instead. Bayesian estimate allows the decision maker to combine prior information about the process with the continuously incoming data in a natural flexible manner. In the real options framework, at any given moment, we compare the cost of recalibrating the process to the cost of postponing the (optimal) decision for later. The decision is thus based on cost‐benefit analysis rather than statistically significant deviations from the in‐control process. To have a clear focus on the conceptual representation of the novel methodology, we consider a continuously sampled binary process. We derive the algorithm for the control chart, which can also, in this discrete setting, be represented as a table, a matrix, or a tree. We also investigate the performance of the method in different settings with particular attention being paid to the role of Bayesian prior. Being flexible in prior beliefs leads to better results anywhere outside of the in‐control process. Together, Bayesian paradigm and dynamic decision‐making approach create a realistic representation of a real‐life decision‐making process.     

A deep learning approach for the dynamic dispatching of unreliable machines in re-entrant production systems

This research combines deep neural network (DNN) and Markov decision processes (MDP) for the dynamic dispatching of re-entrant production systems. In re-entrant production systems, jobs enter the same workstation multiple times and dynamic dispatching oftentimes aims to dynamically assign different priorities to various job groups to minimise weighted cycle time or maximise throughput. MDP is an effective tool for dynamic production control, but it suffers from two major challenges in dynamic control problems. First, the curse of dimensionality limits the computational performance of solving large MDP problems. Second, a different model should be built and solved after system configuration is changed. DNN is used to overcome both challenges by learning directly from optimal dispatching policies generated by MDP. Results suggest that a properly trained DNN model can instantly generate near-optimal dynamic control policies for large problems. The quality of the DNN solution is compared with the optimal dynamic control policies through the standard K-fold cross-validation test and discrete event simulation. On average, the performance of the DNN policy is within 2% of optimal in both tests. The proposed artificial intelligence algorithm illustrates the potential of machine learning methods in manufacturing applications.     

Joint production and maintenance planning with machine deterioration and random yield

Production, yield and maintenance are three key components for sustaining the competitiveness of a manufacturing firm. In this paper, we investigate a joint production and maintenance planning problem in a periodic review environment subject to stochastic demand and random yields. The manufacturing system deteriorates from period to period according to a discrete-time Markov chain. The objective is to find an integrated lot sizing and maintenance policy for the system such that the aggregate cost associated with production, holding, backlogging and maintenance is minimised. We formulate this integrated planning problem as a Markov decision process and analyse the structural properties of the optimal policies. We prove that the optimal production and the maintenance policies both exhibit a control limit structure and show that the solution to the finite-horizon problems converges to that of the infinite-horizon problem.     

Using reliability analysis to support decision making in phased mission systems

Because of the environments in which they will operate, future autonomous systems must be capable of reconfiguring quickly and safely following faults or environmental changes. Past research has shown how, by considering autonomous systems to perform phased missions, reliability analysis can support decision making by allowing comparison of the probability of success of different missions following reconfiguration. Binary decision diagrams (BDDs) offer fast, accurate reliability analysis that could contribute to real‐time decision making. However, phased mission analysis using existing BDD models is too slow to contribute to the instant decisions needed in time‐critical situations. This paper investigates 2 aspects of BDD models that affect analysis speed: variable ordering and quantification efficiency. Variable ordering affects BDD size, which directly affects analysis speed. Here, a new ordering scheme is proposed for use in the context of a decision‐making process. Variables are ordered before a mission, and reordering is unnecessary no matter how the mission configuration changes. Three BDD models are proposed to address the efficiency and accuracy of existing models. The advantages of the developed ordering scheme and BDD models are demonstrated in the context of their application within a reliability analysis methodology used to support decision making in an unmanned aerial vehicle.     

Online maintenance policy for a deteriorating system with random change of mode

Most of maintenance policies proposed in the literature for gradually deteriorating systems, consider a stationary deterioration process. This paper is an attempt to take into account stochastically deteriorating systems which are subject to a sudden change in their degradation process. A technical device subject to gradual degradation is considered. It is assumed that the level of degradation can be resumed by a single scalar variable. An online maintenance decision rule is proposed, which makes it possible to take into account in real time the online information available on the operating mode of the system as well as its actual deterioration level. We show the efficiency of considering online decision rules for maintenance with respect to traditional maintenance policies based on a static alarm threshold. Numerical simulations are given, to assess and optimize the performance of the maintained system from its asymptotic unavailability point of view. It is compared to the results obtained with classical control-limit maintenance policies.     

A Bayesian Approach to Controlling Percent Defective A Model and Application

The purpose of this paper is to develop a general model for controlling the percent defective of an ongoing production process. The model is developed in a Bayesian decision theory framework so that, using dynamic programming, optimal (least cost) control decisions can be found. An application of the model to a real world production process is described in detail. The problems of estimating the model parameters are discussed along with some approaches to overcoming the estimation problems. Finally, the optimal control policies for the real world process are presented and are shown to be straightforward and easily implemented.     

TOPAAS: An Alternative Approach to Software Reliability Quantification

In the realm of safety related systems, a growing number of functions are realized by software, ranging from ‘firmware’ to autonomous decision‐taking software. To support (political) real‐world decision makers, quantitative risk assessment methodology quantifies the reliability of systems. The optimal choice of safety measures with respect to the available budget, for example, the UK (as low as reasonably practicable approach), requires quantification. If a system contains software, some accepted methods on quantification of software reliability exist, but none of them is generally applicable, as we will show. We propose a model bringing software into the quantitative risk assessment domain by introducing failure of software modules (with their probabilities) as basic events in a fault tree. The method is known as ‘TOPAAS’ (Task‐Oriented Probability of Abnormalities Analysis for Software). TOPAAS is a factor model allowing the quantification of the basic ‘software’ events in fault tree analyses. In this paper, we argue that this is the best approach currently available to industry. Task‐Oriented Probability of Abnormalities Analysis for Software is a practical model by design and is currently put to field testing in risk assessments of programmable electronic safety‐related systems in tunnels and control systems of movable storm surge barriers in the Netherlands. The TOPAAS model is constructed to incorporate detailed fields of knowledge and to provide focus toward reliability quantification in the form of a probability measure of mission failure. Our development also provides context for further in‐depth research. Copyright © 2013 John Wiley & Sons, Ltd.     

Optimal control of an unreliable machine using fuzzy-logic control: from design to implementation

In this paper, the development, implementation and validation of fuzzy logic to control an unreliable machine in manufacturing systems are presented. The fuzzy-logic controller developed in this context is based on the optimal control policy using hedging-point methodology. The fuzzy-logic controller has two-subsets of fuzzy-logic controls. The first provides a decision whether the system should produce part at make-to-stock or at make-to-order mode. The decision provided by the first fuzzy-logic control is then used by the second fuzzy-logic control to specify at what production rate the part should be produced. Simulation and implementation have been performed by controlling an unreliable machine using the developed fuzzy-logic controller. Then, the simulation results are compared with the simulation results given by the optimal control policy (hedging-point methodology). The results show that the performance of the developed fuzzy logic outperforms and is more precise than the hedging-point method under certain conditions.     

基于支持向量机改进算法的船舶类型识别研究

利用船舶目标辐射噪声DEMON谱特征,采用改进的支持向量机算法,实现了对船舶目标的分类识别研究。针对支持向量机算法对噪声比较敏感和最优分类面求解时约束较多不利于支持向量机最优分类面寻优的问题,在保持支持向量稀疏性和应用径向基核函数的条件下,对支持向量机算法在松弛变量和决策函数两方面进行了改进,提出了基于径向基核函数的齐次决策二阶损失函数支持向量机改进算法,并应用于利用船舶目标辐射噪声DEMON谱进行船舶目标类型分类识别实验。理论分析、数据仿真与实验结果表明,该改进算法实现了在二次规划中的较少约束条件下最优分类面求解,具有模型参数寻优空间广阔、总体分类性能优的特点,其分类性能优于原支持向量机算法,是一种适合于船舶辐射噪声DENOM分类识别的有效的支持向量机改进算法。     

Preventive maintenance and replacement policies for deteriorating production systems subject to imperfect repairs

This paper presents a special case of integration of the preventive maintenance into the repair/replacement policy of a failure-prone system. The machine of the considered system exhibits increasing failure intensity and increasing repair times. To reduce the failure rate and subsequent repair times following a failure, there is an incentive to perform preventive maintenance on the machine before failure. When a failure occurs, the machine can be repaired or replaced by a new one. Thus the machine's mode at any time can be classified as either operating, in repair, in replacement or in preventive maintenance. The decision variables of the system are the repair/replacement switching age or number of failures at the time of the machine's failure and the preventive maintenance rate. The problem of determining the repair/replacement and preventive maintenance policies is formulated as a semi-Markov decision process and numerical methods are given in order to compute optimal policies which minimise the average cost incurred by preventive maintenance, repair and replacement over an infinite planning horizon. As expected, the decisions to repair or to replace the machine upon a failure are modified by performing preventive maintenance. A numerical example is given and a sensitivity analysis is performed to illustrate the proposed approach and to show the impact of various parameters on the control policies thus obtained.     

Predicting availability functions in time-dependent complex systems with SAEDES simulation algorithms

In this paper, we propose the use of discrete-event simulation (DES) as an efficient methodology to obtain estimates of both survival and availability functions in time-dependent real systems—such as telecommunication networks or distributed computer systems. We discuss the use of DES in reliability and availability studies, not only as an alternative to the use of analytical and probabilistic methods, but also as a complementary way to: (i) achieve a better understanding of the system internal behavior and (ii) find out the relevance of each component under reliability/availability considerations. Specifically, this paper describes a general methodology and two DES algorithms, called SAEDES, which can be used to analyze a wide range of time-dependent complex systems, including those presenting multiple states, dependencies among failure/repair times or non-perfect maintenance policies. These algorithms can provide valuable information, specially during the design stages, where different scenarios can be compared in order to select a system design offering adequate reliability and availability levels. Two case studies are discussed, using a C/C++ implementation of the SAEDES algorithms, to show some potential applications of our approach.     

Optimal joint maintenance and operation policies to maximise overall systems effectiveness

In this paper, the problem of determining the optimal maintenance and operation policies for a multi-state, multi-stage machine maintenance problem is considered. This problem has been formulated in the literature as a Partially Observed Markov Decision Process (POMDP). A new formulation that explicitly ties maintenance, operation, and quality within the POMDP framework is provided. The new formulation maximises Overall Systems Effectiveness for an n-state system with multiple speed and maintenance actions. The model provides, for each time epoch, a set of optimal maintenance and production-rate actions. The decision-maker (controller) can select the optimal policy depending on the system state occupancy vector (belief state). A realistic numerical model is presented to demonstrate the model utility.     

Optimal decision of an economic production quantity model for imperfect manufacturing under hybrid maintenance policy with shortages and partial backlogging

Considering the characteristics of the stochastic shift of the machine state and the uncertainty of the product quality of production, in this paper, we develop an optimisation decision of economic production quantity model for an imperfect manufacturing system under hybrid maintenance policy with shortages and partial backlogging. We assume that the production process is imperfect stemming from the machine reliability and the probability of out-of-control, a hybrid maintenance policy combined of emergency maintenance and preventive maintenance is executed during each production run. Three decision models based on the scenarios of machine breakdown and repair time are developed. The optimal production quantity and maintenance inspection number during each production run are solved with minimising the expected average cost of the system. Numerical examples are used to demonstrate the effectiveness and feasibility of the model. Sensitivity analysis is conducted to analyse the impacts of key parameters on the optimal decision. Some implications related to the effective and economical execution of maintenance policy for practitioners are derived.     

Graphical Analysis for Operation Management: A Graphical Method to Support Operation Decision Making

This paper proposes a graphical method to easy decision‐making in industrial plants operations. The proposed tool ‘Graphical Analysis for Operation Management (GAOM) method’ allows to visualizing and analyzing production‐related parameters, integrating assets/systems maintenance aspects. This integration is based on the Total Productive Maintenance model, using its quantitative management techniques for optimal decision‐making in day‐to‐day operations. On the one hand, GAOM monitors possible production target deviations, and on the other, the tool illustrates different aspects to gain control on the production process, such as availability, repair time, cumulative production, or overall equipment effectiveness. Through appropriate information filtering, individual analysis by class of intervention (corrective maintenance, preventive maintenance, or operational intervention) and production level can be developed. Graphical Analysis for Operation Management (GAOM) integrates maintenance information (number of intervention, type of intervention, required/not required stoppage) with production information (cumulative production, cumulative defective products, and cumulative production target) during a certain timeframe (cumulative calendar time, duration of intervention). Then the tool computes basic performance indicators supporting operational decision‐making. GAOM provides interesting graphical outputs using scatter diagrams integrating indicators on the same graph. GAOM is inspired in the Graphical Analysis for Maintenance Management method, published by the authors (LB, AC, and PV) in 2012. Copyright © 2015 John Wiley & Sons, Ltd.     

Selection of dispatching rules on multiple dispatching decision points in real-time scheduling of a semiconductor wafer fabrication system

Semiconductor wafer fabrication involves one of the most complex manufacturing processes ever used. To control such complex systems, it is a challenge to determine appropriate dispatching strategies under various system conditions. Dispatching strategies are classified into two categories: a vehicle-initiated dispatching policy and a machine-initiated dispatching policy. Both policies are important to improve the system performance, especially for the real time control of the system. However, there has been little research focusing on combining them under various situations for the semiconductor manufacturing system. In addition, it is shown that no single dispatching strategy consistently dominates others in all situations. Therefore, the goal of this study is to develop a scheduler for selection of dispatching rules for dispatching decision variables in order to obtain the desired performance measures given by a user for each production interval. For the proposed methodology, simulation and competitive neural network approaches are used. The results of the study indicate that applying our methodology to obtaining a dispatching strategy is an effective method considering the complexity of semiconductor wafer fabrication systems.