Thread Content
Design of Fault Diagnosis Scheme for Distillation Columns [Abstract] This paper briefly introduces the necessity of fault diagnosis in distillation columns. It mainly discusses the expert system for fault diagnosis of distillation towers and the specific steps for fault diagnosis (which primarily include establishing equipment records, determining inspection methods and intervals as well as maintenance times, performing performance predictions, and managing records). Only by integrating maintenance techniques, fault diagnosis techniques, and computer-based remote monitoring technologies can accidents be effectively prevented and avoided, operating conditions optimized, and the operational lifespan extended, thereby bringing greater economic and social benefits to enterprises. ________________________________________ Browse font settings: - 10pt + 10pt, 12pt, 14pt, 16pt. 1 Introduction: Distillation towers play a very important role in the petrochemical industry; they are mass and heat transfer devices that are widely used in the production processes of chemical and refining companies. Their purpose is to separate the various components in a mixture to achieve the desired level of purity. The essence of the distillation process is to take advantage of the different volatilities of the various components in a mixture – that is, the fact that at the same temperature, these components have different saturated vapor pressures. This allows the lighter components in the liquid phase to transfer to the vapor phase, while the heavier components in the vapor phase transfer to the liquid phase, thereby achieving separation. Therefore, the efficiency of distillation operations has a direct impact on the economic performance of petrochemical companies. Fault diagnosis of distillation towers has become an important aspect of chemical production. Since there are many factors that affect the operational performance of distillation towers as well as the quality of the products, the relationships involved are particularly complex, which in turn impacts the safe production of petrochemical enterprises. In recent years, with the development of computers, artificial intelligence, and neural network technologies, as well as the growing maturity of the theories related to distillation processes (such as process calculations for distillation columns, mechanical calculations for tray design, and calculations regarding auxiliary equipment in these columns), how to integrate computers and these distillation theory principles into the fault diagnosis of distillation columns has become one of the issues of concern for oil refining companies. For this reason, extensive research on fault diagnosis of distillation processes has been carried out both domestically and internationally, with the method of establishing mathematical models being commonly used. This paper discusses in detail the design approach for fault diagnosis of distillation columns. 2. Design of schemes for distillation column fault diagnosis 2.1 Establishing an expert system for distillation column fault diagnosis Common faults in distillation columns are shown in Figure 1. A schematic diagram of the method structure for building a knowledge base for distillation column fault diagnosis is shown in Figure 2. This method consists of four main components: fault input, an hermeneutic* model, a quantitative deep knowledge base, and result output to the domain knowledge base. First, a fault is entered, then the quantitative deep knowledge base is invoked to activate a learning model based on an hermeneutic* model, thereby obtaining a learning rule that is used directly for fault diagnosis. Finally, this rule is stored in the domain knowledge base; the quantitative deep knowledge base needs to be continuously updated with new fault diagnosis methods. 2.2 Diagnostic Steps 2.2.1 Establishing equipment records Figure 1 shows common faults in distillation towers. Every piece of equipment has potential faults, and the process of using such equipment involves a transformation from potential faults to actual functional failures; the progression of these faults is shown in Figure 3. To carry out economical and effective maintenance of the distillation tower, it is necessary to establish equipment records and keep track of its operating conditions. By understanding the historical performance of various components of the distillation tower as well as their current operating status, regular inspections along with appropriate fault-diagnosis methods can be employed to detect potential faults before they cause operational problems, thereby preventing such failures from occurring. 2.2.2 Determination of testing methods Figure 1 summarizes the faults in distillation columns; the main faults include low tray efficiency, low bottom temperature, high reflux temperature, and flooding. To address the aforementioned faults, the 7-ray field scanning detection technology can be used to diagnose and eliminate them, optimize operating conditions, and extend the operational cycle; it also provides important data or a scientific basis for guiding production in oil refining and chemical industries. 2.2.3 Determining the inspection cycle and maintenance time for the distillation tower: The determination of the monitoring cycle for various components of the distillation tower depends on the tower’s own service life. Here, the average ultimate service life of the distillation tower is taken as T; thus, the inspection cycle Tc = T/nd, where nd is the dispersion coefficient. The determination of nd depends on the operating conditions of the equipment. The maintenance time for the distillation tower is determined based on the condition of the tower after inspection. Here, the maximum operating condition of the distillation tower refers to the highest values that its various parameters can reach before a failure occurs; when one or more of these parameters exceed these values, it is considered that the tower has reached its limit of service life and requires repair. 2.2.4 Establishing performance prediction and record management: After each inspection of the distillation tower, it is necessary to keep records of the results of those inspections. At the same time, it is important to determine the availability of the equipment (i.e., its remaining useful life) as well as the timing of the next inspection; therefore, performance prediction is essential. The numerical modeling for predicting equipment performance is as follows: Let X{X1, X2, …, Xn}. } is a time series of length N that reflects the operating conditions of the distillation tower at N different moments in time; X ； To describe a certain characteristic parameter of the operating condition of mechanical equipment, Xi∈Rn(i_1,…,N); now it is necessary to predict the operating condition of the mechanical equipment k steps ahead, that is, to determine xN+k. The intrinsic relationship between the state at time (N+k), xN+k, and the first N states is expressed by a functional relationship F: XN+k = F(X1, X2, …, Xn). ) (1) Then the life prediction of the distillation column can be expressed as finding F by establishing a mathematical model, such that for any ∈>0, we have ||FF(X) – FF(X)|| = ||xN+k – XXN+k|| ||