Development of an online monitor for condenser dirtiness

Abstract: A new method for online monitoring of condenser dirtiness is proposed. This method takes the heat transfer end difference as the research object, comprehensively considers the impact of various factors on the end difference, and uses neural network modeling technology to successfully separate the effects of condenser dirt and changes in operating parameters on the end difference, which can accurately Monitor the degree of condensation contamination online. The monitor with DSP as the core developed according to this method was introduced, and a field test was conducted. The test results proved the effectiveness of the instrument. Keywords: condenser dirty heat transfer end difference online monitoring neural network DSP The condenser is a large heat exchange equipment in a thermal power plant. Its function is to condense the low-temperature steam into water after the steam turbine performs work to improve the efficiency of the thermodynamic cycle. . Figure 1 is a schematic structural diagram of a surface condenser. When the condenser is running, the cooling water comes in from the lower half of the front water chamber, enters the rear water chamber through the cooling water pipe (heat exchange pipe), turns upward, then flows into the front water chamber through the upper half of the cooling water pipe, and is finally discharged. Low-temperature steam comes in from the steam inlet, flows downward through the gap between the cooling water pipes, releases heat to the pipe wall, and condenses into water. During this working process, due to the unclean quality of the cooling water, some solid mixtures (called dirt) that are not conducive to heat transfer accumulate on the inner wall of the copper tube. The presence of dirt reduces the heat transfer capacity of the heat exchange surface, thereby reducing the turbine efficiency, so it must be cleaned. How to quantitatively measure the degree of contamination of the condenser in order to provide a basis for reasonable cleaning of the condenser is a problem that many scholars are discussing. To sum up, the methods that have been proposed are roughly as follows: (1) Determine the degree of dirtiness of the condenser by measuring the thermal resistance of the dirt. (2) Determine the degree of contamination of the condenser by measuring the water flow resistance between the condenser outlet and the inlet water chamber. (3) Judge the degree of dirtiness of the condenser by calculating the heat transfer coefficient. Thermal resistance method can more accurately determine the degree of contamination of the condenser, but it needs to bury armored heat on the heat exchange tubes to detect the tube wall temperature. There are many heat exchange tubes in the condenser, which is difficult to implement in engineering; water flow The resistance can reflect the amount of dirt, but it cannot reflect the thermal conductivity of the dirt. Using this method to determine the degree of dirt in the condenser is not accurate enough. The heat transfer coefficient reflects the heat transfer performance of the condenser, but the current calculated heat transfer coefficient is not accurate enough. The traditional empirical formula is adopted and the influence of non-condensable gas (air) in the steam on the heat transfer effect is not considered. Therefore, there is a large error when the condenser operates under different operating conditions. The heat transfer end difference is an important performance index that reflects the heat exchange status of the condenser. Compared with the heat transfer coefficient, this parameter is easy to measure, and its changes can be continuously observed to accumulate data. Therefore, this article chooses it to reflect the dirt of the condenser. state. However, in addition to the degree of contamination of the heat transfer surface, the heat transfer end difference is also closely related to the operating parameters of the condenser such as steam flow, cooling water volume, etc. Therefore, how to separate the contamination of the heat exchange surface from many parameters? The impact of dirt on the end difference has become the key to accurately determine the degree of dirt in the condenser. 1 Measurement principle The heat transfer end difference is defined as: δt=ts-two (1) In the formula, δt——the heat transfer end difference of the condenser ts——the saturated steam temperature corresponding to the condenser pressure two——cooling water It can be seen from the analysis of the heat transfer process of the outlet temperature that when the cooling area of ​​the condenser is constant, δt can be expressed as: δt=f(Dc, Dw, c, ε, twi) (2) In the formula, Dc——steam flow rate Dw—— Cooling water flow rate c - contamination coefficient of the condenser ε - content of non-condensable gas (air) in the steam twi - cooling water inlet temperature Assume that after the condenser is thoroughly cleaned, at a given steam flow rate The end difference measured under Dc, cooling water flow rate Dw, cooling water inlet temperature twi, and air content ε is δtd (δtd can be regarded as the end difference corresponding to the working condition in the clean state). After changing the working condition and running for a period of time, the end difference is measured. The end difference obtained is δtf. Obviously, the difference Δδ between δtd and δtf is caused by dirt on the heat exchange surface and changes in working condition parameters. It can be expressed as: Δδ=Δδc+Δδg (3 ) In the formula, Δδc - the change in end difference caused by dirt on the heat exchange surface, is called the dirt end difference. Δδg - the change in end difference caused by changing working conditions, is called the end difference under variable working conditions. The dirt coefficient is defined as: c=(Δδc)/δtd=(Δδ-Δδg)/δtd (4) It can be seen from the above formula that to determine c, Δδg needs to be found. Since Δδg = f (ΔDs, ΔDw, Δtwi, Δε) describes a very complex heat transfer process, its accurate mathematical model is difficult to obtain. Therefore, this article uses neural networks to establish variable working condition terminals based on input and output measurement data. The difference model achieves accurate measurement of the degree of condensation contamination. 2 Neural network modeling Variable working condition end difference Δδg=f (ΔDs, ΔDw, Δtwi, Δε) can be approximated by a three-layer feedforward neural network, as shown in Figure 2. The Sigmoid function is selected as the excitation function of the hidden layer neurons: in the formula, a=1.716 b=2/3. The test data of different working conditions of the condenser in the clean state are used as training data, and the BP algorithm is used to train the neural network. The learning objective function is: In the formula, n - the number of samples yi - model output di - expected output The weight correction of the neural network adopts the fast gradient descent method. After the neural network is trained, it can be put into application. According to the variable working condition end difference obtained by the neural network and equation (4), the dirt coefficient can be calculated. 3 Instrument structure 3.1 Hardware design The online monitor takes DSP as the core, collects relevant parameters in real time, calculates the pollution coefficient and displays it dynamically. Its hardware structure is shown in Figure 3. In the figure, tp is the temperature of the steam-gas mixture at the measurement point; p is the pressure of the steam-gas mixture at the measurement temperature. The air content is obtained by the following method: Measure the pressure of the steam-water mixture at the outlet of the condenser extraction equipment, and also measure the temperature of the steam-water mixture. The air content in the steam-water mixture is measured by the following formula: ε=( p-ps)/(p-0.378ps) (7) Among them, ps——the water vapor saturation pressure corresponding to the outlet temperature of the steam-gas mixture, which can be obtained by looking up the table. The DSP uses TMS320F240, whose structure is: (1) 32-bit CPU; (2) 554-word dual-port RAM, 16K-word FLASH EEPROM; (3) two 10-bit A/D converters; (4) serial Communication Interface. The chip can exchange data with the control room host through a serial communication interface. 3.2 Software design The software design adopts a modular structure, which mainly includes: (1) data acquisition and processing module; (2) neural network calculation module; (3) display module; (4) communication module. 4 Test results 4.1 Acquisition of neural network model The field test was conducted on the N-3500-2 condenser of Xiangtan Power Plant. Under the condition of ensuring that the condenser is clean, Dc=135t/h, Dw=9400t/h, twi=15℃, ε=0.015% are used as the set working conditions to obtain the test data of the condenser under different working conditions. to train the neural network. Table 1 shows the comparison results between the output of the neural network and the measured data under some working conditions during condenser cleaning. It can be seen from the comparison results that the output of the neural network is basically consistent with the measured end difference, indicating that the modeling method based on the neural network can obtain a variable working condition end difference model with higher accuracy. Table 1 Comparison between neural network model output and measured data under different working conditions Steam flow rate Dc (t/h) Cooling water volume Dw (t/h) Inlet water temperature tsi (℃) Air leakage amount ε (%) Measured end difference (℃) Model output difference (℃) Error (℃) 135 81.6 54.1 188.5 108.2 108.2 81.6 161.3 9400 9400 9400 9400 12350 12350 6800 6800 15.0 10.2 5.5 20.8 22. 2 17.4 7.8 13.3 0.015 0.015 0.015 0.054 0.033 0.075 0.015 0.015 6.1 4.8 4.4 10.3 6.8 12.1 5.0 6.3 6.1 4.7 4.4 10.4 6.9 11.9 5.0 6.2 0 -0.1 0 0.1 0.1 -0.2 0 -0.1 4.2 After the neural network model for online monitoring of the degree of contamination is determined, online monitoring can be carried out. In order to verify the accuracy of this method, 16 armored thermocouples were buried in different positions of the condenser for comparison with the thermal resistance method. The test is divided into two parts: (1) Thoroughly clean the condenser and measure the change in dirt coefficient within 24 hours after cleaning. (2) Re-operate the cleaning device and measure the change in dirt coefficient during cleaning. The test results are shown in Table 2 and Table 3. Among them, Table 2 shows the changes in the contamination coefficient of the condenser after the cleaning device is shut down; Table 3 shows the changes in the contamination coefficient of the condenser after the cleaning device is put back into operation. Dw=9400/h and ε=0.015% remained unchanged during the test. In the clean state, the end difference measured under the set working conditions is δtd=6.1°C. Table 2 Changes in contamination coefficient of the condenser after the device is out of service Time after cleaning (h) Steam flow rate Dc (t/h) Inlet water temperature twi (℃) Outlet water temperature two (℃) Steam temperature ts (℃) End difference δtf (℃) Fouling end difference Δδ (℃) Fouling coefficient Thermal resistance method in this article 0 2 4 6 8 10 12 14 16 20 24 108.2 108.2 108.2 108.2 108.2 108.2 108.2 108.2 108.2 135.0 135.0 13.5 13.5 12.8 12.3 12.0 11.6 11.1 11.1 10.9 12.3 14.2 23.5 23.5 22.6 22.1 21.7 21.3 20.7 20.6 20.4 23.7 25.5 29.0 29.4 29.3 29.1 29.0 28.8 28.4 28.3 28.3 32.3 33.8 5.5 5.9 6.7 7.0 7.3 7.5 7.7 7.7 7.9 8.6 8.3 0.00 0.48 1.02 1.44 1.52 1.71 1.78 1.84 1.93 1.99 2.15 0.00 0.079 0.167 0.236 0.25 0.28 0.291 0.302 0.316 0.326 0.352 0.000 0.063 0.164 0.225 0.265 0.282 0.288 0.318 0.327 0.339 0.341 Table 3 Changes in the dirt coefficient of the condenser after the double-flap cleaning device is put into operation Condition cleaning time (h) Steam flow rate Dc (t/h) Inlet Water temperature twi (℃) Outlet water temperature two (℃) Steam temperature ts (℃) End difference δtf (℃) Dirt end difference Δδc (℃) Dirt coefficient The method in this article Thermal resistance method 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 135.0 135.0 135.0 135.0 135.0 135.0 135.0 135.0 14.2 14.3 14.2 14.1 14.0 14.0 13.8 13.7 25.5 25.8 25.9 25.9 25.8 25.8 25.6 25.5 33.8 33.3 32.9 32.5 32.2 32.2 32.0 32.0 8.3 7.5 7.0 6.6 6.4 6.4 6.4 6.5 2.15 1.22 0.71 0.29 0.11 0.06 0.03 0.03 0.352 0.2 0.20.116 0.047 0.018 0.01 0.005 0.005 0.341 0.223 0.116 0.053 0.027 0.005 0.004 0.005 It can be seen from Table 2 and Table 3 that the dirt coefficient obtained by the method introduced in this article is basically consistent with the thermal resistance method, and the changing trend of the dirt coefficient is consistent with the condensation coefficient. The accumulation and cleaning characteristics of dirt in heat exchanger tubes show that the dirt coefficient obtained by this method is reliable. In view of the shortcomings of existing methods for monitoring the degree of dirtiness of condenser heat exchange tubes, this paper proposes a new method for online monitoring of the degree of dirtiness. Based on this method, an online monitor was developed and field tests were conducted. The test results proved that the instrument can accurately monitor condenser dirt online. Since many of the signals required by the measuring instrument (such as steam flow, cooling water inlet temperature, etc.) are already available on site and can be introduced directly or through communication, the measuring instrument is low-cost, easy to install, and has good application prospects. This monitor is also suitable for large heat exchange equipment in other industries.