توسعه یک مدل تعمیر و نگهداری پیشگیرانه بهینه مبتنی بر ارزیابی قابلیت اطمینان برای تجهیزات تهویه مطبوع در ساختمان های اداری

The purpose of this study is to examine the failure trend of HVAC systems in high-rise office buildings using the reliability assessment, and propose a method of predicting an optimal inspection period for condition-based preventive maintenance (CBM) using the Monte Carlo method, in point of view of randomness and independency between failure and inspection. This study describes the probability process method of measuring the effect of condition-based preventive maintenance on HVAC system's reliability and optimization of condition-based preventive maintenance. A simulation model is presented to analyze condition-based preventive maintenance through a fixed period maintenance inspection by maintenance personnel conducted to detect failure occurring. Based on this simulation model, the effects of condition-based preventive maintenance on units of HVAC system's reliability are quantitatively obtained, and the mean time between failures of units under CBM action is suggested. In addition, the basic characteristics of the condition-based preventive maintenance are analyzed by sensitivity analysis. As a result of this study, the method to predict an optimal inspection period is also suggested in order to increase the reliability of units, and effect on expected profit of optimal preventive maintenance inspection period is computed.

1.1. Backgrounds and aim As office buildings are getting bigger, the information technology (IT) and office automation (OA) equipment are developing and the facility systems are complex, the demand for building facilities’ reliability is rising. Especially, the reliability of HVAC systems is seriously considered as an important aspect of occupants’ productivity and comfort. To improve the reliability of building facilities, the most important issues requiring immediate attention are to grasp and remove the factors causing problems in all steps of the life cycle such as building planning, design, construction, maintenance and to evaluate quantitatively the reliability model of failure history data [1]. There are two ways to improve the reliability of an air-conditioning facility. One way is to design HVAC systems with special consideration for redundancy and the other way is to implement preventive maintenance. The former is the design mode which is necessary in the case where air-conditioning function requires a high-degree of reliability such as telecommunication rooms and computer rooms [2] and the latter is the maintenance mode which should be applied to most HVAC systems in general buildings. The preventive maintenance can be divided into time-based preventive maintenance and condition-based maintenance. The time-based preventive maintenance is mainly applied to the non-repairable items which have a life distribution and its research and theory are established as a maintenance policy by Barrow [3]. The condition-based preventive maintenance, also called prediction maintenance, is applied to the items where failure happens accidentally [4]. Because the HVAC system's reliability sometimes breaks down within a life period of durability through Weibull analysis,1 most of the main air-conditioning equipment is compatible for condition-based preventive maintenance. Condition-based prevention maintenance with the object of detecting symptoms of a failure is the mode to decrease an emergency stop and maintenance through the execution of preventive measures that is on the basis of consecutive operating condition's supervision by the real time-observation system and regular inspection by maintenance personnel. Such a mode makes an equipment's MTBF augment [5]. Therefore, it is necessary to present the optimal inspection period as a preventive maintenance policy to improve the reliability of facilities utilizing mean time between failures (MTBF) based on reliability statistical information. In this research, failure trend is analyzed through investigation of the failure history data for the units and parts of the main HVAC systems in high-rise office buildings located in Seoul, Korea. Furthermore, the models of detailed inspections and simplified inspections applied in practical maintenance are established and grasp the independence between failure occurrence's randomness and inspection. Then the method to seek the optimal inspection period is shown by the Monte Carlo simulation, where the probability process containing the probability distribution of the condition-based preventive maintenance models in the HVAC system. In addition, an optimal inspection period model of CBM based on reliability assessment is made, and effects on reliability improvement of this model, and on expected profit of this reliability improvement's result are analyzed. 1.2. Former works and methods Matsuura [6] found that for the air-conditioning equipment in Japanese office buildings the failures which appear between second and fifth years occur randomly. Takakusagi [4] insisted that the main units of HVAC system in a Japanese computer center building indicate the reliability property of random failure in the period of life and that can be expressed by Weibull's distribution [4]. In addition, although there is a mathematical model by Kawasaki [5] and Takakusagi [4], which optimizes both the reliability improvement effect of the condition-based preventive maintenance and the inspection policy by considering the probability theory inquiry, and detection of fault and the symptom of failure, it is difficult for this model to expand to the model that operates by separating detailed inspections and simplified inspections applied in practical maintenance. To solve these limitations of existing preventive maintenance research, a method to obtain the optimal inspection period of preventive maintenance for the practical maintenance through the Monte Carlo simulation that processes the probable factors not by theories, but by random numbers was introduced. The summary of this research method is as follows: (1) The failure history data of the HVAC system was analyzed, and the failure trend of the units and the parts of the HVAC systems were evaluated by applying the reliability assessment method, and the MTBF was proposed. (2) The random digits’ generation and verification methods needed for Monte Carlo simulation were investigated. (3) After setting up the condition-based preventive maintenance model, it was applied for the independence relationship between failure occurrence and inspection for the randomness existing in random numbers, then the degree of improvement by condition-based preventive maintenance was simulated through a case study.

In this research, the failure history data for the units and parts of the HVAC system were inspected and the failure trend was analyzed by applying a reliability assessment method. In addition, the models of detailed inspections and simplified inspections applied practically in practical maintenance were established and the method to seek the optimal inspection period is shown by Monte Carlo simulation, the probability process containing the probability distribution of the condition-based preventive maintenance models in the HVAC system by noticing the independence between failure occurrence's randomness and inspection. Moreover, the sensitivity analysis of the maintenance factor was executed by a prepared simulation program. As a result of the above processes, the factors that improve the reliability of HVAC system through the condition-based preventive maintenance were considered and the method to seek an optimal inspection period for improving the units’ reliability was proposed. (1) As a result of the failure trend analysis of the units of studied HVAC system, which consists of the burner, fan, safety valve, and water tube, it is revealed that not only the units have a tendency to have random failure, but most of the parts have the same tendency as well. Therefore, the fact that the condition-based maintenance policy executing preventive maintenance, which preliminarily predicts the failure, is proper for HVAC systems was revealed. In addition, the failure characteristic was grasped by estimating the MTBF for the units and parts and the parameter of the failure time distribution. (2) The model which can easily predict the condition-based preventive maintenance policy for the HVAC system was established by the Monte Carlo simulation. Then, using this model, the relationship between the HVAC system's MTBF and the fault occurrence period and detection was grasped. Moreover, the method to seek the optimal inspection period of condition-based preventive maintenance as a preventive maintenance policy to improve reliability was proposed by applying MTBF based on the probability information of reliability for the HVAC system. (3) The effect of the preventive maintenance factors such as the inspection frequency, the original MTBF, the fault occurrence period, the diagnosis time, to affect the reliability improvement of the air-conditioning equipment was investigated by a sensitivity analysis, and the effect of the original MTBF and the inspection frequency to affect the expected profit was also grasped. On the expected profit side, the more the inspection frequency is, the more expected profit is increased, but the ratio is decreased from a specific point. As a result, by presenting optimal inspection frequency, it is found that the effect of expected profit of preventive maintenance by CBM can be increased. (4) It is required to optimize inspection period in order to inspect HVAC system economically. To attain this purpose, optimal inspection period is suggested by using a model of relation between inspection number per unit time and reliability, and model of relation between expected profit by CBM and inspection type. (5) In comparing expected profit of periodical inspection maintenance with expected profit of optimal inspection maintenance, effect on expected profit of optimal inspection maintenance is suggested. (6) It is suggested that reliability-centered maintenance process of HVAC system using original MTBF, existing reliability of equipment can be assessed, and based on original MTBF, expected profit model by CBM can be made in order to improve reliability.