Method of normalization of dependability indicators of railway transport facilities

Aim. The results of evaluation of a technical system’s (facility’s) factual state allow making a decision on a further life (operation continuation, maintenance assignment, decommissioning and a facility’s replacement etc.). Under the conditions of resource limits, it is vital to identify most “problematic” facilities that require primary investments. The aim of the research is to develop a method of normalization of dependability indicators whose application is intended to improve targeted investment allocation for maintenance of facilities, which allows fulfilling the requirement of uninterruptible transportation under the conditions of resource scarcity. Methods. The research uses methods of system analysis, probability theory, mathematical statistics, and correlation analysis. It proposes approximation of a time series of factual values related to a dependability indicator by a three-parameter gamma distribution based on a scarcity function q(x). Findings. The research has considered the criteria of choice of railway transport facilities requiring the enhancement of dependability for the cases of unavailability and availability of a normalized dependability indicator. It has been shown that if introducing normalization of indicators one should take into account non-similar maintenance conditions for facilities in different enterprise units, which are determined by differences in climatic factors, technical capabilities for maintenance and repair, staffing levels, grades of tear and wear of facilities, requirements for their productivity. The research has analyzed the conditions of association of a service supplier’s and user’s requirements for normalization of a dependability indicator value. It has been demonstrated that it is reasonable to establish a single threshold normalized value xη of a dependability indicator, in which case a normalized value xη for the attribute x shall comply with the requirements of a service user as well as a service supplier. In the case of a single threshold value, the risk Qη = P{x > xη} of noncompliance of an indicator with the specified requirements is in fact split between a service user and a service supplier according to their agreement. Conclusions. The paper proposes a method of normalization of a dependability indicator based on statistical data assuming that in general this indicator may be evaluated for a certain period of observance as acceptable for a service user. For to choose and justify the normalized value of a dependability indicator, the authors have studied the relations between a service supplier and a service user, have analyzed statistics using the method of estimation of empirical sufficiency of a raw data series as well as approximation of an ordered initial series by a three-parameter gamma distribution. The paper provides an example of normalizing a value of a facility failure rate indicator as per the criterion of a specified risk of its violation based on the quantiles of an obtained function of sufficiency. It has been shown that the proposed approach allows establishing a correlation between a normalized value and a risk of its violation via a function of sufficiency, which can be obtained on the basis of existing statistical data on a facility’s dependability for the past periods. This correlation makes it possible to guarantee the ensuring of compliance of factual and normalized indicator values with a specified risk level for a facility working in normal mode