International Journal of Reliability, Quality and Safety Engineering

This study discusses methods for designing two types of ARQ policy; (1) the usual ARQ policy and (2) the extended ARQ policy, which is proposed in this study. In the protocol of the usual ARQ policy, the upper bound for the number of retransmissions of the data is empirically determined. A statistical model and a model based on Kullback–Leibler information (K–L model) are proposed for determining such a upper bound of retransmissions. Both the models are also applied to the extended ARQ policy. Numerical examples are presented to illustrate the K–L model which makes the design of ARQ policies easier than the statistical model.

The data transmission fails due to errors which have been generated by disconnections, noises and distortions in a communication line. To protect such failures, the following three schemes of error-control procedures are mainly used : (i) automatic-repeat-request (ARQ), (ii) forward-error-correction (FEC) and (iii) hybrid ARQ which combines FEC with ARQ. This paper proposes two stochastic models of hybrid ARQ policies with retransmission number K and response time T, and derives the mean times to success of data transmission. Comparing the mean times with each other, we discuss analytically which policy is better. These results are also evaluated in numerical examples.

When it is not economical to carry out maintenance in-house, out-sourcing to an external agent is an alternate viable option. In this paper, we study a simple maintenance service contract involving a single agent (providing the maintenance service) and a single customer (owner of the equipment and recipient of the maintenance service). We develop a simple model to obtain the optimal strategies for both the customer and the agent. We give a complete characterization of the strategies along with some sensitivity analysis and discuss some extensions.

A two-stage method is presented for analysing reliability data with suspended items. The method proposed takes account of the ages of the suspended items when calculating plotting positions for the failure events. This sensitivity to suspension timings provides a more precise method than other plotting-based methods currently in use.

This is a proposal for the implementation of "Science SQC" (a program based on the new scientific SQC method in Manufacturing) in order to elevate the quality of the business process. We construct the "Availability and Reliability Information Administration System — ARIM-BL" as a practical example of "Inline-Online SQC." As for the key technology of "Science SQC," it is indispensable to implement "Inline-Online SQC" by utilizing both methods consistently. It makes the dark points in business process clear and manages to communicate between man and information. This study describes the total integrated process information system, which links together between Production Engineering division and Manufacturing Engineering division. It is representative of intelligent "Inline-Online SQC." We have introduced the demonstrative and effective system into FBL (Flexible Body Line), which was the core of TPS (Toyota Production System).

The Modulated Power Law process (MPLP) has been recently proposed as a suitable model to describe the failure pattern of repairable mechanical units subject to imperfect or hazardous maintenance. In this paper, an informative Bayes procedure is proposed to analyze failure data arising from a MPLP sample, which allows prior information on the failure/repair process to be incorporated into the inferential procedure. Inference both of the MPLP parameters and of some functions thereof (such as the unconditional mean number of failures and the unconditional failure intensity), as well as prediction on failure times in a future sample, is developed. Finally, a numerical example is given to illustrate the proposed procedure.

For some products that degrade over time, failure is defined in terms of a specified critical value of performance characteristic. The life of such a product depends on the specified critical value. The more severe the critical value, the shorter the life would be, and vice versa. This paper presents a method of accelerating life by using severe critical values. The method, based on degradation measurements, provides estimates of the relationship between life and critical value and reliability. An example is given to illustrate the application of the proposed method. Comparison between the method and traditional time-to-failure analysis shows that the proposed method is more efficient and cost-effective.