Optimization of Interval Between Overhaul on Steam Power Plant with Risk Based On Human Error and Profi

Power plant companies have many different standard interval between overhauls. The difference may be due to the different methods used by the company. However, these methods do not take into account aspects of risk, aspects of human error and financial aspects simultaneously. The purpose of this study is to determine the optimal interval overhaul by considering aspects of risk, human error aspects and financial aspects simultaneously. The propsed method to calculate reliability plant model using the Criticality Risk Matrix tool and elimination of equipment that can be overhauled at the time the plant under operating conditions, succeeds in reducing the number of equipment that needs to be analyzed, from 210 to 30 equipment. Parameter reliability plant obtained β: 0.9755, η: 602.0508, γ: 7.5942. The reliability plant model is combined with reliability constant affected by human error resulting in a combined reliability model. The combination of combined unreliability model, multiply to Economical Consequences so that the Total Cost model can be obtained. Genetic algorithm is an effective method to be used in the optimization process of a non linear function. The difference between net income model and total cost will produce profit model, so the optimal overhaul interval can be known by doing optimization on the model so that obtained top1 = 7698 hours. Optimization of total cost model can be done to find out the latest time the plant must be shutdown to do overhaul in order to avoid cost inefficiency. Optimal time total cost obtained top2 = 17645 hours. Keywords―Interval, Overhaul, Risk, Human Error, Profit.


I. INTRODUCTION 1
A power plant company in West Java, Indonesia as the object of this study, applies overhaul management standards to its power plant installations where overhaul intervals are set for 8000 hours of cumulative operation calculated from the first synchrone on grid to the next overhaul. The interval overhaul is stated for 4-8 months in the manual book and with additional notes if the equipment is in good condition, it can be extended gradually in order to optimize function and lower maintenance costs [1]. Every institution has different steam turbine overhaul intervals [2]. GE's steam turbine service official guide line recommends a 5-year interval, while the Electric Power Research Institute (EPRI) says the overhaul interval time on the turbine is when Nett Present Value calculation method is negative. 1 Emilius Wahyu is with Java-Bali Power Generating Company (PT. PJB), West-Java, Indonesia. E-mail: emilius@ptpjb.com. 2 Arif Wahjudi is with Department of Mechanical Engineering, Institut Teknologi Sepuluh Nopember (ITS), Surabaya, 60111, Indonesia. E-mail: arybach@me.its.ac.id.
Not far from GE, the power generation company in the USA uses a 5-6 year interval for its generating unit and the VGB states the interval between the turbine overhaul is 25.000 hours of operation. The method of determining the inspection interval at a LNG-based processing plant based on risk is formulated [3]. This method assists the process of optimization of inspection intervals (overhaul) by considering the risks associated with production, safety and the environment and combining them with the calculation of system reliability by using previous operating data. The study was developed by adding a correction factor derived from the calculation of probability due to human error so that the calculation of the combination of risk with the reliability of the equipment will change [4]. This human error factor can be calculated using the Success Likehood Index Method (SLIM). The interval between overhauls is obtained by finding the difference between income potential and the level of risk, so that the quality of the overhaul results greatly affects the interval after [5].

II. METHOD
The calculation of the optimization of the overhaul interval uses the following equations: A. Reliability equations with a 3-parameter Weibull distribution [6] ( ) = ( ) . (1)

III. RESULT AND DISCUSSION
Modeling is implemented in 5 modules : 1. selection of equipment 2. calculation of plant, human error and combined reliability model 3. calculation of risk model 4. calculation of profit model 5. optimization of profit function and total cost Selection of equipment is done with the aim of simplifying the process of analysis by reducing equipment with certain categories in the next analysis process. Reliabilty Block Diagram (RBD) system and plant level is structured to facilitate the analysis of modeling. Equipment selection requires input data downtime equipment to be able to calculate probability failure. Probability failure is used to determine the probability value. Calculating the impact value if a down equipment can be done using RBD instructions. Determination of risk categories, based on Criticality Risk Matrix in the figure 1 can be implemented so that it can be done eleminasi on equipment that has low risk category. The process of elemination at this stage succeeded in reducing 210 to 51 equipment.
The next process is to calculate the Mean Time Between Failure (MTBF) of the redundant equipment that passed the previous elemination. An equipment can be eliminated to be removed from subsequent analysis if the MTBF minus equipment system is greater than the duration of the overhaul equipment. Eliminated equipment will be overhauled when the plant in operating condition. The advanced elemination process successfully reduces 51 to 30. RBD is revised by eliminating eliminated equipment.
The compilation of plant reliability model is done by fitting data down time using final revision RBD and Blocksim software tool, yielding parameter β: 0.9755, η: 602,0508, γ: 7,5942 on the equation (1) From the above ECT model can be punctured risk model per unit time (RISKe "), as follows: Net income can be obtained from the gross income difference with the fuel cost. Fuel costs will change according to NPHR changes, so it is necessary to model NPHR. Based on NPHR data analysis with linear regression can be obtained the model as follows:  Maximization optimization of profit function is done to get the time point (top1) on the highest profit per hour. Optimization of minimization of total cost function is done to get the time point (top2) at the lowest total cost value. Non linear optimization is done using Genetic Algorithm method, with top.
In the graph V "can be obtained