Abstract : The ammonia stripping tank is a device in the iron oxide red production line used to strip ammonia from raw ammonia waste liquid. As a key piece of equipment in the iron oxide red ammonia stripping process, the quality of ammonia pressure control in the stripping tank directly affects the quality of the ammonia product and the subsequent neutralization and oxidation reactions. Therefore, the research and design of the ammonia pressure control system for the stripping tank can improve product quality, reduce labor intensity, and is of great significance for achieving full automation of the ammonia stripping process.
This paper details the production process of ammonia stripping from iron oxide, analyzes the time delay and other characteristics of the stripping process, and based on this, performs mechanistic modeling of the ammonia pressure control system in the stripping tank to obtain an approximate model of the system. Considering the time delay and inertia characteristics of the ammonia pressure control system, this paper proposes applying generalized predictive control (GDC) to the research on the ammonia pressure control system of the iron oxide stripping tank, which significantly improves the control performance of the ammonia pressure control system. Simulations of the GDC using the MATLAB platform show that GDC has excellent control performance even under time-varying parameters and disturbances, and can effectively eliminate the influence of disturbances on the system.
Keywords : Iron oxide red ammonia stripping tank; ammonia pressure control; PID control; generalized predictive control
Abstract : The ammonia evaporation tank is a device that used for the wastewater of aqua ammonia in the production line of iron oxide red to evaporate. As a key equipment of the process of ammonia evaporation, the pressure control on the ammonia evaporation tank has a direct influence on the quality of the ammonia and the follow-up neutralization and oxidation reaction. Therefore, the research and design of the pressure control system of the ammonia evaporation tank can improve the product quality, reduce the labor intensity, and it has very important significance of realizing the automation of the process of the ammonia evaporation.
The paper introduces the process of the ammonia evaporation in the production of the iron oxide red, and analyzes the characteristics of time delay and great inertia in the process of the ammonia evaporation.Besides, on the basis of that the approximation model of the pressure control system of the ammonia evaporation tank is deduced from the establishment of the mechanism model and the estimation of the model parameters.Through the simulation of the generalized predictive controller which based on the MATLAB platform, the The results show that even in the cases of parameter time-variance and disturbance, the generalized predictive control can have well control effects, and can be very good to eliminate interferences on the influence of the system.
Key words : ammonia evaporation; pressure control; PID control; generalized predictive control
1 Introduction
Titanium oxide powder is not only an important raw material for smelting metallic titanium and producing titanium dioxide via the chloride process, but also a high-performance chemical pigment. It possesses excellent acid and alkali resistance, high-temperature resistance, and is waterproof, non-toxic, and exhibits good wear resistance and anti-chalking properties. However, the waste liquid generated during its production process contains harmful substances such as ferric chloride and ferrous chloride, attracting increasing attention. Improper treatment of this waste liquid will pollute the environment and harm human health. For enterprises, it will lead to substandard wastewater discharge, hinder continuous production, and further impact business efficiency. From the perspective of titanium oxide powder production processes, ammonia stripping, as a crucial step in waste liquid treatment, requires strict control of its process parameters. This research focuses on the ammonia pressure control system in the ammonia stripping tank.
Figure 1. Relationship between paper tension and traction force
The ammonia stripping process not only provides ammonia gas, the reactant raw material for the neutralization oxidation tank, but also affects the biochemical wastewater treatment process, making it a crucial link in the entire wastewater treatment process. To ensure the normal operation of the ammonia stripping production process, the most important aspect is controlling the pressure of the ammonia stripping tank. Excessive ammonia outlet pressure will cause excessive ammonia gas to be injected and neutralized with the atomized titanium yellow wastewater during the subsequent neutralization process, increasing the pH value of the neutralized liquid and making it difficult to meet emission standards. Conversely, excessively high ammonia pressure increases the solubility of ammonia in water, hindering its volatilization and evaporation; conversely, excessively low ammonia pressure makes it difficult to reach the required temperature for ammonia stripping, resulting in poor stripping efficiency, incomplete ammonia removal, and excessively high ammonia nitrogen content in the wastewater at the bottom of the tank, affecting the production of the biochemical treatment process. Maintaining a constant ammonia pressure is essential for the continuous and normal production of the ammonia stripping process and subsequent neutralization oxidation reactions. For the iron oxide red ammonia stripping tank, the main control is the ammonia pressure at the top of the tank. Specifically, according to the iron oxide red production process requirements, the pressure at the top of the tank is generally controlled at around 20 kPa. Since the ammonia pressure control system of the ammonia stripping tank is a complex control system with nonlinearity, time-varying characteristics and large time delay, it is difficult to establish an accurate mathematical model. Under such circumstances, traditional control algorithms have many shortcomings, such as poor anti-interference ability and difficulty in adjusting parameters online in real time. Therefore, an algorithm suitable for ammonia pressure control in the ammonia stripping tank needs to be studied.
The ammonia stripping process takes place in the ammonia stripping tank. The raw ammonia solution fed into the neutralization and oxidation tank reacts with the added alkali solution via a feed plate, generating an ammonia solution mixture within the tank. This mixture then encounters superheated steam introduced from the bottom of the tank, undergoing a two-phase mass and heat transfer process until equilibrium is reached. After repeated operation through multiple layers of trays within the tank, the gas and liquid components of the ammonia solution mixture are relatively completely separated, yielding concentrated ammonia gas. The ammonia vapor at the top of the ammonia stripping tank passes through a separator, and the resulting condensate is directly returned to the ammonia stripping tank. The concentrated ammonia vapor is then sent to the next stage, the neutralization and oxidation tank, for neutralization and oxidation reactions.
Figure 2. Process flow diagram of iron oxide red ammonia stripping process
In general, the ammonia stripping process in the iron red ammonia stripping tank can be divided into two parts: the first is the reaction stage, in which the added alkali solution and the raw ammonia solution provided by the neutralization oxidation tank undergo a chemical reaction in the feeding plate device, and the reaction stage generates ammonia solution to be distilled; the second is the distillation stage, in the iron red ammonia stripping tank, the superheated steam and ammonia solution mixture are in countercurrent contact, and mass and heat transfer are carried out multiple times on each tray. After each tray exchange, the gas phase with increased ammonia content rises from the tray, while the liquid phase with increased water content descends along the tray. The difference in temperature and concentration between the reaction mixtures is the main reason for the mass and heat transfer process. The steam and ammonia solution mixture can theoretically achieve two-phase equilibrium in the tank after repeated exchange processes [1] .
Therefore, the principle of iron oxide red ammonia stripping reaction can be summarized as reactive distillation. Superheated steam is one of the crucial conditions for the normal operation of the ammonia stripping reaction. Simultaneously, to prevent gas-liquid imbalance in the tank after the reaction from affecting subsequent ammonia stripping processes and energy consumption, the condensate generated during the ammonia stripping production process must be returned to the stripping tank, thereby ensuring continuous operation of the ammonia stripping production.
2. Ammonia pressure control system model for ammonia stripping tank
Considering the complexity of ammonia pressure control in the ammonia stripping tank, a simplified mathematical model is needed to establish a suitable model for ammonia pressure control. Here, the system is simplified to four parts: a feed pump, a reaction vessel, a pressure vessel, and piping. The reaction vessel refers to the physicochemical reaction in the ammonia stripping process; the raw ammonia water is fed into the stripping tank by the feed pump, undergoes sufficient reaction, and is then distilled to obtain concentrated ammonia gas. The pressure vessel is a hypothetical buffer volume container for measuring the ammonia gas product. The piping model refers to the gas transmission pipelines, including those for reaction lag factors.
Figure 3. Simplified schematic diagram of the ammonia pressure control system for the iron oxide red ammonia stripping tank.
2.1 Feed Pump Control Model
Traditional methods of adjusting flow rate by regulating valve opening result in unnecessary power loss and are not energy-efficient. From a power source perspective, using a frequency converter to regulate the feed pump speed can change the flow rate. During speed adjustment, as the pump output head decreases, power previously wasted on resistance in the regulating valve is saved. Therefore, for feed pumps with frequently changing operating conditions, such as in ammonia stripping processes, frequency conversion speed regulation should be considered to adjust the flow rate, adapting to the changing conditions of the ammonia stripping process while achieving energy savings.
Under certain pressure conditions, ignoring the internal and external leakage of the pump and the volumetric flow rate of the compressed raw liquid, the relationship between rotational speed and flow rate is as follows:
In the formula, Q represents the actual flow rate; V represents the pump displacement, which is the volume of medium discharged per revolution of the pump. The displacement value is provided by the manufacturer, or it can be calculated using geometric relationships by measuring the flow rate under no-load conditions and then dividing it by the rotational speed. From the above formula, we know that the pump displacement and efficiency are constant values, indicating that the speed and flow control model of the feed pump is a linear proportional model, which can be simplified as follows:
2.2 Iron Red Ammonia Evaporation Reaction Model
The ammonia stripping reaction is a complex two-phase gas-liquid mass and heat transfer process. When establishing its reaction model, necessary simplifications must be made to the ammonia stripping operating conditions and characteristics. When establishing the vapor-liquid equilibrium, material equilibrium, and energy equilibrium of the ammonia stripping process, the following assumptions are made:
The mixture in the ammonia stripping tank is approximated as a vapor phase and a liquid phase with ideal characteristics, and the Wilson equation is used to derive and solve the problem. On each tray in the tank, the vapor and liquid phases are completely mixed and the temperature and concentration are uniformly distributed during the countercurrent mass and heat transfer process. The heat consumed during the reaction process and the dynamic characteristics of the superheated steam are ignored [2] .
Assuming the conditions are met, and given the process parameters of the ammonia stripping process, such as tank specifications, load, feed flow rate, and concentration, the MSEH equations for each tray in the tank can be established, namely the material conservation equation, energy conservation equation, vapor-liquid balance equation, and normalization equation. Since the gas flow rate is calculated based on the pressure difference between trays, calculating the pressure on each tray individually and then determining the flow rate would lead to unstable pressure values, making it difficult to obtain the transfer function for the reaction model control. To address the control objective of the ammonia pressure control system in the ammonia stripping tank, the calculation process needs to be simplified. Since the final controlled object of the system is ammonia pressure, the ammonia flow rate can be calculated using the conservation of chemical molecules, specifically ammonium ions, within the material conservation principle, and then converted into ammonia pressure to obtain the control effect. Let the feed flow rate of the ammonia stripping tank be F, the raw ammonia concentration be C, and the overall tray efficiency be given. Based on the conservation of the amount of ammonium ions, the following equation is obtained:
Ammonia volume and mass flow rate:
Combining the above equations (3), (4), and (5) and performing a Laplace transform, the mass flow rate of ammonia gas at the inlet of the ammonia stripping tank is obtained as follows:
Here, n(s) is the speed of the feed pump.
2.3 Outlet regulating valve and pressure vessel model
The ammonia gas outlet generated during the ammonia stripping process is controlled by a regulating valve; therefore, modeling is performed specifically for the outlet regulating valve. The mass flow rate of the linear regulating valve is:
Linearization and Laplace transform of its equilibrium point yields:
Where is the proportional constant parameter of the control valve; is the pressure at the equilibrium point across the valve; is the density of the gas; is the initial valve opening; is the valve opening degree; and is the pressure across the valve.
The relationship between the mass flow rate changes at the inlet and outlet of the ammonia stripping tank is as follows:
Linearization around its equilibrium point and Laplace transform with respect to time yield:
Where V is the volume of the pressure vessel; is the density of the gas; R is the gas constant; is the gas temperature; and is the ammonia pressure in the ammonia stripping tank.
2.4 Pipeline Model
Ignoring pipeline losses during gas transport, the gas transport process inevitably results in significant pure time delays due to the long pipeline, the reaction of the raw material ammonia, and the ammonia vaporization process. For ease of analysis, the pipeline model can be simplified and treated as an independent link in the overall generalized object's transport delay, mainly including measurement lag, reaction lag, and transport lag, which are pure time delays, and each unit model is different.
2.5 Ammonia Pressure System Model for Ammonia Stripping Tank
Based on the models established in the preceding units, the model of the ammonia pressure system in the ammonia stripping tank is derived:
Wherein, is the ammonia pressure in the ammonia stripping tank, is the gas mass flow rate regulated by the feed pump, is the gas mass flow rate through the regulating valve, and is the valve opening degree.
make
The mathematical model of the ammonia pressure control system for the ammonia stripping tank, derived from the above formula, is as follows:
Introducing the system pipeline model, i.e., the delay element, we get:
From the above formula, it can be seen that the ammonia pressure in the ammonia stripping tank is mainly controlled by the speed of the feed pump. The valve opening of the outlet regulating valve and the outlet pressure remain constant within a certain range. Therefore, the transfer function of the speed on the ammonia pressure in the ammonia stripping tank is in the form of a first-order inertial element with pure time lag.
is the proportional gain, is the inertia time constant, and represents the pure time delay. These three parameters together describe the control characteristics of the controlled object. Their specific meanings are as follows:
Inertial time constant T: The time required for the controlled object to reach a new steady-state value after receiving an external input, without the intervention of a controller. It characterizes the dynamic response of the controlled object. If the time constant increases, the time for the system output to recover to a new steady-state value after responding will also increase.
Lag time: In actual system control, when the input variable to the controlled object changes, the system output needs a certain amount of time to respond and change. This is called time lag. Time lag characteristics exist in the production of iron oxide and ammonia stripping, as well as in the control of many complex industrial processes. Its time constant determines how fast the controlled object lags.
Generalized proportional gain factor: Here, "generalized object" refers to the part of the control system excluding the controller, specifically including the ammonia pressure control object, actuators, and pressure transmitters. The proportional gain factor is a static gain parameter and is independent of time. Static gain is the ratio of the change in input to the change in output before the system output stabilizes. Under the same input, a larger value indicates a greater change in the system output, making the system output more sensitive to input changes, and resulting in poorer stability of the controlled object. Conversely, a smaller value indicates better stability of the controlled object.
3. Generalized predictive control of ammonia pressure in ammonia stripping tank
Based on the previous analysis of the characteristics of the ammonia pressure control process in the ammonia stripping tank, the controlled object is a first-order inertial element with pure time delay. Due to the change in external environmental disturbance parameters during the ammonia stripping process, the model is mismatched, and conventional control algorithms are obviously not effective in controlling the ammonia stripping process.
In view of the time delay of the controlled object in the iron red ammonia stripping process control and the characteristics of the precise mathematical model of the control process, the predictive control algorithm GPC with low model dependence, adaptive ability and strong robustness is selected. In the iron red ammonia stripping tank ammonia pressure control system, the multi-step prediction and control time domain compensation time delay of the generalized predictive control are used to judge the future control action trend, and the current best control action is obtained by rolling optimization. At the same time, due to the online model identification and feedback correction function, it has a strong adaptive ability to parameter changes and environmental disturbance model mismatch, so the iron red ammonia stripping tank ammonia pressure system is flexible and convenient to design, and has good control performance and robustness [3] .
Figure 4. Structure diagram of the generalized predictive control system for ammonia pressure in the ammonia stripping tank.
3.1 Prediction Model
Assume the predicted model vector of the controlled object based on the step response is given, and N is the modeling time domain. When there are M control increments at time k , its future output value can be calculated:
3.2 Scrolling Optimization
Equation (21) can be written in vector form:
Considering the desire to avoid excessively drastic changes in the control increment, the vector form of the optimization performance index at time k can be taken as:
Substituting equation (3) into equation (4), and obtaining the result through the necessary condition of extreme values dJ ( k )/ dΔuM = 0, we can construct the actual control action applied to the object. At the next moment, it proposes a similar optimization problem to solve, namely the "rolling optimization" strategy.
3.3 Feedback Correction
When the control quantity u ( k ) is applied to the object at time k , the predicted output value at future time can be calculated using the prediction model (2). However, due to the actual model mismatch and environmental interference, the predicted value may deviate from the actual value [4] . The output error is corrected by weighting e ( k +1) to adjust the prediction of the future output:
In the formula: is an N- dimensional vector composed of weight coefficients [4] ,
This is the corrected output prediction vector. After shifting, it can be used as the initial prediction value at time k +1, which can be represented in vector form as follows:
4 System Simulation and Analysis
Using MATLAB software to implement the least squares method for parameter identification and inputting actual recorded data, the transfer function of the ammonia pressure control object in the ammonia steaming tank can be obtained as [5] .
Using MATLAB as the platform, the obtained ammonia pressure control system model was simulated and compared using PID and generalized predictive control through Simulink and M-file function programming. The PID parameters, after tuning, are as follows:
The simulation parameters of the generalized predictive control algorithm after adjustment are as follows: softening coefficient; prediction time domain; control time domain; control weighting constant; the simulation is shown in the figure below, including simulation under the standard case without interference, simulation under the interference signal, and simulation under the model mismatch.
Figure 5. Simulation of conventional PID control. Figure 6. Simulation of generalized predictive control.
Figure 7 Simulation of step disturbance in generalized predictive control Figure 8 Simulation of random noise disturbance in generalized predictive control
Figure 9 Simulation of PID static gain mismatch Figure 10 Simulation of generalized predictive control static gain mismatch
Comparing Figures 5 and 6, under the standard condition without interference, the PID control system has a large overshoot and a long dynamic response time to steady state, approximately 250 seconds. In contrast, the generalized predictive control system has a very fast response speed, almost no overshoot, and a settling time of less than 50 seconds. Its control performance is significantly better than that of the PID controller, and it well meets the control accuracy and control requirements of the ammonia pressure control system for the ammonia steaming tank.
Figures 7 and 8 show that a step signal disturbance and random white noise interference were added for 150s, respectively. The simulation results show that the generalized predictive control has an adaptive effect, which can eliminate the influence of external disturbances on the control system within a certain range, so that the error can approach 0 quickly and the system can reach a stable state.
Figure 10 shows that when the model is mismatched and the static gain is mismatched, the output performance of the generalized predictive controller changes very little. Overall, the generalized predictive control algorithm has strong anti-disturbance performance and robustness, and is fully applicable to the ammonia pressure control of the iron red ammonia steaming tank.
5. Conclusion
After determining the model parameters of the ammonia pressure control system for the iron oxide red ammonia stripping tank, and by selecting appropriate parameters, the object model of the ammonia pressure control system for the iron oxide red ammonia stripping tank was simulated using a PID controller and a generalized predictive controller, respectively. Through analysis and comparison, it was found that the generalized predictive controller not only has a good suppression effect on interference signals, but also can respond quickly under model mismatch conditions to achieve a stable system state, and has good control effect, adaptability, and robustness.
References:
- Wang Kun. Modification and Operation Analysis of Ammonia Stripping Unit [J]. Henan Chemical Industry, 2005, 22(4): 46-47.
- Zuo Aiwu. An intelligent control system for ammonia stripping production [D]. Wuhan University of Science and Technology, 2007.
- Zhang Jiaying, Wang Wenlan. Cascade generalized predictive control of boiler water level control system [J]. Electric Power Automation Equipment, 2010, 30(11): 75-77.
- Fu Xiaolin. Research and simulation of an implicit generalized predictive self-tuning control algorithm [J]. Industrial Instrumentation and Automation Devices, 2011, 24(2): 7-8.
- Li Guoyong. Intelligent Predictive Control and Its MATLAB Implementation [M]. Beijing: Electronic Press, 2010.
