Diagnosis of feedwater heater performance degradation using fuzzy inference system

Highlights

• We show how fuzzy inference systems can be used in plant diagnosis when it is necessary.

• FIS are able to integrate expertise and rule learning from data into a single framework.

• FIS has expandability to diagnosis widely. This means that Fuzzy inference system can be applied to all power plant models in the same way.

• Consequently, more measurement inputs, accurate knowledge and elaborate composition of Fuzzy sets can allow for more accurate diagnosis of performance degradation.

Abstract

Power generation facilities cannot avoid performance degradation caused by severe operating conditions such as high temperature and high pressure, as well as the aging of facilities. Since the performance degradation of facilities can inflict economic on power generation plants, a systematic method is required to accurately diagnose the conditions of the facilities.

This paper introduces the fuzzy inference system, which applies fuzzy theory in order to diagnose performance degradation in feedwater heaters among power generation facilities. The reason for selecting only feedwater heaters as the object of analysis is that it plays an important role in the performance degradation of power generation plants, which have recently been reported with failures. In addition, feedwater heaters have the advantage of using many data types that can be used in fuzzy inference because of low measurement limits compared to other facilities. Fuzzy inference systems consists of fuzzy sets and rules with linguistic variables based on expert knowledge, experience and simulation results to efficiently handle various uncertainties of the target facility. We proposed a method for establishing a more elaborate system. According to the experimental results, inference can be made with consideration on uncertainties by quantifying the target based on fuzzy theory. Based on this study, implementation of a fuzzy inference system for diagnosis of feedwater heater performance degradation is expected to contribute to the efficient management of power generation plants.

Introduction

Performance degradation refers to a phenomenon in which a device fails to exhibit its intended performance. Power generation facilities, such as Nuclear Power Plants (NPPs), operated under severe conditions of high temperature and high pressure for long periods cannot avoid performance degradation. In fact, such issues have been frequently reported with the recent aging of facilities. As the performance degradation of facilities lead to economic, it is necessary to maintain and repair facilities based on accurate diagnosis before such serious conditions occur.

A feedwater heater is a device that preheats water supplied to a steam generator to maintain appropriate temperature conditions, and has the advantages of improving the cycle efficiency, minimizing the thermal stress of the steam generator, and preventing the reduction of life. When performance degradation occurs in such a facility, the heat transfer capacity of the feedwater channel is reduced to directly affect generation efficiency.

Through the past research, we developed a methodology using a regression model (Jee, Heo, Jang, & Lee, 2011) and a methodology using diagnosis tables (Kim & Heo, 2012) to diagnose the thermal performance degradation in feedwater heaters. The methodology using the regression model is a numerical analysis method that expresses the performance degradation of the facility and the performance degradation caused by surrounding facilities as mathematical formulae and computes the results through matrix calculations corresponding to the number of facilities. Reliability of this methodology was reduced when noise occurred in the data measured at the actual site, and there was difficulty in uniform application of the methodology to power generation plants operating under different conditions. The methodology using diagnosis tables is widely used at the actual sites because it can make an inference about the type of performance degradation present based on fluctuation of the data. Nonetheless, this methodology cannot make an inference about the severity of performance degradation that is found. The purpose of this study is to apply fuzzy logic as a method to make up for the shortcomings of these prior studies (Kothamasu & Huang, 2007; Wang & Hu, 2006; Wang & Elhag, 2008).

The body of this paper introduces a simulation of the performance degradation using simulation software, preparation of a diagnosis table using data obtained from the simulation, and a diagnosis using a fuzzy inference system.

There are different types of fuzzy inference methods, and the Mamdani fuzzy inference method is the most widely used method of direct inference. The advantage of the Mamdani fuzzy inference method using an If-Then rule is that it first determines whether an appropriate control input value of the user belongs to the membership function using ‘If’ and then converts the value into a number by calculating the degree to which the value belongs to the fuzzy set using `Then'. This inference method can make use of qualitative advantages of the methodology using a diagnosis table because it prioritizes fluctuation of the variables. Further, because it is a simple operation, the instability of the numerical analysis used in the regression model can be improved. Detailed inference and operation processes will be introduced in the following section (Abraham, 2005).

In this paper, the feedwater heaters of NPPs were selected as the objects of study. When performance degradation occurs in a feedwater heater, its measurement values fluctuate. Here, while degradation of thermal performance identically occurs on the macroscopic level, internal phenomena differ in each case and show different trends in the fluctuation of the variables. This study intends to make an inference about the type and degree of performance degradation using the fluctuation of measurement values (Guimara & Lapa, 2007). First of all, the limited condition described below is taken into account for simplification of the design of the performance degradation detection system. 1) Among the diverse performance degradations that may occur in a feedwater heater, five representative single performance degradation phenomena are used, including excessive increase in drain water level due to malfunctioning of the drain valve or from other causes, reduced pressure of the heater shell due to alien substances, clogging of the tube, phenomenon in which the feedwater does not pass by the heating part due to a defect in the pass partition plate, and leakage in the feedwater tube. 2) Double performance degradation phenomena assume cases in which two of the phenomena mentioned above overlap. An assumption was made that performance degradation phenomena occur sequentially instead of simultaneously. Therefore, triple performance degradation is not considered, as it is possible to detect it before the overlapping of three phenomena (Hadjimichael, 2009).