DESIGN OF AN INTELLIGENT SYSTEM TO CONTROL THE TECHNOLOGICAL SYSTEM OF AMMONIA PRODUCTION SECONDARY CONDENSATION

Modern ammonia production involves complex large-tonnage energy technological systems, which are built in almost all countries according to the unified ideology of the company “M. Kellogg & Co” (USA) [1, 2]. In the synthesis units of such production, a two-stage condensation system of production ammonia is most often adopted, which occurs due to the cooling of circulation gas (CG). For cooling at the stage of secondary condensation, the AM-1360 series assemblies employ economical heat-use absorption (ARU) and steam-ejector (SRU) refrigeration units [3, 4]. Their efficiency is ensured due to the possibility of utilization of heat of material flows of both low (about 140 °C) in ARU, and ultra-low (up to 90 °C) in SRU, temperature potential. However, the use at the preliminary stage of primary condensation of air cooling devices predetermines the functioning of the technological system of secondary condensation (TSSC) under the influence of constant changes (seasonal and daily) in the external heat load from CG. That leads to significant fluctuations in the TSSC temperature regime and deviations in the temperature of secondary condensation from the rated norm of −5 °C [5]. The increase in this temperature even by 1 °C in the synthesis units AM-1360 causes a decrease in the energy efficiency of production by increasing the annual consumption of natural gas by 307.3 thousand nm3 [6]. Therefore, it is necessary to stabilize the temperature regime of TSSC at the regulatory level. DESIGN OF AN INTELLIGENT SYSTEM TO CONTROL THE TECHNOLOGICAL SYSTEM OF AMMONIA PRODUCTION SECONDARY CONDENSATION


Introduction
Modern ammonia production involves complex large-tonnage energy technological systems, which are built in almost all countries according to the unified ideology of the company "M. Kellogg & Co" (USA) [1,2]. In the synthesis units of such production, a two-stage condensation system of production ammonia is most often adopted, which occurs due to the cooling of circulation gas (CG). For cooling at the stage of secondary condensation, the AM-1360 series assemblies employ economical heat-use absorption (ARU) and steam-ejector (SRU) refrigeration units [3,4]. Their efficiency is ensured due to the possibility of utilization of heat of material flows of both low (about 140 °C) in ARU, and ultra-low (up to 90 °C) in SRU, temperature potential. However, the use at the preliminary stage of primary condensation of air cooling devices predetermines the functioning of the technological system of secondary condensation (TSSC) under the influence of constant changes (seasonal and daily) in the external heat load from CG. That leads to significant fluctuations in the TSSC temperature regime and deviations in the temperature of secondary condensation from the rated norm of −5 °C [5]. The increase in this temperature even by 1 °C in the synthesis units AM-1360 causes a decrease in the energy efficiency of production by increasing the annual consumption of natural gas by 307.3 thousand nm 3 [6]. Therefore, it is necessary to stabilize the temperature regime of TSSC at the regulatory level.

DESIGN OF AN INTELLIGENT SYSTEM TO CONTROL THE TECHNOLOGICAL SYSTEM OF AMMONIA PRODUCTION SECONDARY CONDENSATION
need to predict the regime parameters of the settings of the control system regulators. This is especially true of SRU, in which the change in refrigeration capacity, and, therefore, the consumption of refrigerant to HTE is carried out, as a rule, by auto-adjusting the position of the nozzle and aircooled capacitors (ACC) [12,13]. However, such auto-configuration, which is considered in the cited works, is too complex, time-consuming, and expensive to implement. In addition, in such large-tonnage industries, reservation must be implemented, that is, the installation of another ejector with auto-adjustment to improve the reliability of operation. Therefore, it is more expedient to equip HTE evaporators with much simpler ejectors (for example, three) without auto-tuning. That could ensure their selective inclusion in work according to the predicted value of a certain situation by the operator under a supervisory mode, which would definitely improve the reliability of control technology. At the same time, the consumption by one of these three ACCs is 200 kWh, which provides for the condensation of 10 t/h of ammonia vapor. However, the issues related to the development of particularly algorithmic software remained unresolved in modern literature and in production conditions. Therefore, the task of improving the reliability of control and the ability to prepare the operator for such changes under a supervisory control mode requires the construction of a mathematical model and algorithmic software for a decision-support subsystem. This approach makes it possible to establish both patterns and quantitative dependences of the influence of external heat load on the effectiveness of heat transfer processes in TSSC, information on which is practically absent in periodicals. Owing to these studies, it would be possible to assess in advance the forecasts of possible changes in the temperature of CG at the input of HTE under the conditions of existing uncertainties and, therefore, numerical indicators (U settings) for control vectors.
Thus, the task of modeling and forecasting in order to automatically search for hidden patterns and relationships between variables in large data sets is a complex multi-stage process. All these tasks, as it is known [14], have all the features of the intelligent system.
The AM-1360 series assemblies employ the information and control system (ICS) TDC-3000 made by the United States [15]. This system covers both field and technological level of production management. However, its hardware and software were developed for the so-called "closed"-type systems. Such ICSs do not provide for the possibility of modernizing production based on the latest ammonia synthesis technologies without adding or integrating additional and auxiliary control systems. That is, the addition of new algorithms and automation tools is impossible. As a result, the modernization of such ICSs is significantly complicated.
Thus, the tasks of developing algorithmic software for the subsystem of decision-making support to the control system and retrofitting its hardware acquire special relevance in the overall process of increasing the efficiency of production by stabilizing the temperature of secondary condensation. There is an additional task to align existing control system with the specified subsystem.

The aim and objectives of the study
The purpose of this research is to design an algorithmic and hardware and software subsystem for decision support The large-tonnage ammonia production also predetermines the significant metal consumption in the technological equipment, and, therefore, the excessive inertia of heat exchange processes in TSSC. All this greatly complicates the process of control. Therefore, research into the development of a high-quality and reliable control system for TSSC under the conditions of external disturbances is a relevant task for the overall process of improving the energy efficiency of ammonia production.

Literature review and problem statement
Paper [7] shows that the construction of high-quality computer-integrated control technology under the existing conditions of TSSC functioning is most effectively executed using a system approach. One of the main aspects of this approach is system-control. It should be aimed at studying TSSC under the conditions of external and internal disturbances to make decisions on stabilization of the secondary condensation temperature at the regulatory level of no more than −5 °C.
It is known from [8] that the main element of the functional structure of such control technology for decision-making should be an identifier with an embedded TSSC mathematical model. The TSSC assembly is quite complex and contains a condensation column (CC), an auxiliary heat exchanger (AHE) a high-temperature evaporator (HTE) with SRU, and two low-temperature evaporators (LTE) with water-ammonia ARU [3]. Therefore, in the process of building a mathematical model that could predict the result or value of certain events (forecasts) of such a complex system, there are certain difficulties. They are associated with the processing and execution of multiple operations with a significant array of current information [9]. In addition, model construction is further complicated by the presence of uncertainties in the functioning of TSSC. This is due to the influence of external heat load from CG. The uncertainty, as found in [6], relates to the impossibility of continuous automatic control over ammonia concentration in CG, both at the TSSC input and output. There is also a change in the heat transfer coefficients in the system's devices. As shown in [10], this is caused by a change in the condensation thermal resistance and, mainly, by its dependence on the concentration of ammonia in CG at the input [6].
The range of change, at the TSSC input, of CG temperature and the concentration of ammonia in CG is 35÷45 °C and 8.6÷13 % by volume, respectively. In such circumstances, the temperature of CG at the input of HTE, which is included in the scheme of ammonia SRU, would also change. Therefore, as shown in [11], it is necessary to use a system for stabilizing the temperature of CG at the level of 30 °C in order to maintain the temperature of CG at the input of LTE at the level of 9 °C. Such a control system could also make it possible to stabilize the temperature of secondary condensation (CG at the output of LTE) at the regulatory level of no more than −5 °C, regardless of the change in the external thermal load at the TSSC input.
The significant metal consumption by the equipment and the complexity of technological structure contribute to the excessive inertia of heat exchange processes. This is especially true of AHE and CC, the total weight of which is more than 400 tons and is characterized by feedback on CG. In such circumstances, to an even greater extent there is a under the conditions of uncertainty in the computer-integrated control technology of TSSC regarding the synthesis units of the AM-1360 series. This could ensure improved reliability and quality of process management under the conditions of uncertainty. Owing to that, the energy efficiency of production would increase due to the stabilization of the secondary condensation temperature at the regulatory level of no more than −5 °C, regardless of external disturbances acting on TSSC.
To accomplish the aim, the following tasks have been set: -to develop the algorithmic and program support for a decision-support subsystem given the existing uncertainties regarding TSSC; -to determine the parametric sensitivity and coordinates of the control vector on the basis of the developed algorithmic software for real changes in the external heat load at the TSSC input; -to define the functional structure and design the software and technical structure of computer technology for managing the contours of refrigerant flow rate to HTE and MEA solution to the SRU steam generator; -to devise a structural-logical scheme of information flows of computer-integrated TSSC control technology of the "open" type.

The study materials and methods
The research was carried out by mathematical modeling. To this end, we used equations for the mathematical notation of the heat exchanger AHE, the evaporator HTE, the subprogram for calculating the coefficients of heat transfer and the concentration of ammonia in CG at the TSSC input 3

IN а
and output 3 NH AP а according to the algorithms developed and tested for adequacy. The algorithms that are in STAB, STOCH, and DANI files were built on the basis of earlier studies [6,16].
We determined the quantitative dependences of the setpoints for flow rate regulators of refrigerant consumption to the HTE and MEA solution to the PCO steam generator from the existing external perturbations at the TSSC input using a specially developed algorithm. The software implementation of the algorithm underlies the decision-support subsystem.

1. Development of the algorithmic software for a decision-support subsystem
The generalized flowchart of the developed algorithm is shown in Fig. 1; the software implementation was realized in the MATLAB R2014a package (The MathWorks, USA). Fig. 1 correspond to the following physical quantities:  The algorithm contains convergence cycles that provide measurement of heat flows from the pipe space Ф , A P and intertube space Ф A IT and in the process of heat exchange Ф A HE of the AHE apparatus to determine the temperature of the CG at the output of its intertube space 1

Θ
In this case, the STAB subroutine algorithm makes it possible to form a stable information array of current data and exclude transition modes. That provides the possibility of calculating the actual heat transfer coefficient , A E K Further, using the STOCH stochastic approximation subroutine, the conditions of stationarity, the reproducibility of the process, and the hypothesis of the normality of the empirical distribution are checked. According to the results of such a check, functional dependences are established for the numerical assessment of the uncertainties of volumetric concentrations of ammonia in the CG at the input 3

IN а
and output 3 NH AP а of TSSC. At the same time, the numerical assessment of the uncertainty of the total coefficient of thermal resistance , EA T R taking into consideration condensation, was carried out according to the resulting array of experimental data. In the process of approbation of the algorithm, the following functional dependences were used to numerically assess these uncertain-ties obtained according to experimental data of industrial operation of TSSC from earlier works [3,6] is the flow rate of nitrogen-hydrogen mixture and CG at the TSSC input, nm 3 /s, respectively; P CG -CG pressure at the TSSC input, MPa; AME C М -consumption of condensed ammonia, t/h.
At the end, based on the set temperature 1 ; , where , GCG P С HTE L С -respectively, the average heat capacity of the gas mixture CG and liquid ammonia of the pipe space of the HTE, kJ/(kg·K); r C , r R -heat of condensation and vaporization of ammonia in the pipe and intertube space of HTE, kJ/kg;

2. Determining the parametric sensitivity and coordinates of the control vector under real external disturbances
The assessment of parametric sensitivity regarding the influence of such perturbing factors as the temperature of the primary condensation and the flow rate of the CG to the parameters of the regulators' settings was carried out using the dimensionless coefficient K Zi , which is determined from the following formula: where M i1 , M i2 is the coordinate of the control vector, respectively, under the perturbing factor Z i1 and Z i2 . Mathematical modeling according to the developed algorithm makes it possible to investigate the patterns of influence of the most characteristic for industrial conditions of variable vectors of external perturbations Z(t) on the vector of control Y(t) in order to stabilize the temperature of secondary condensation at the regulatory level of -5 °C. Proceeding to the variable space, these vectors take the following form: It should be noted that the limitations in research are due to the range of changes in coordinates in the process of developing a mathematical model of TSSC. Fig. 2 shows separate results of studies on the effect of the temperature of the primary condensation Θ PC on the efficiency of the heat transfer process in TSSC on the coordinate indicators of the control vector under the following restrictions: nm 3  °С -accordingly, the temperature of CG at the input of the pipe space of AHE, at the input of CC, at the output of LTE, and at the outlet of the intertube space of CC [3]. Fig. 3 shows the results of studies on the impact of CG consumption at the input of AHE CG IT V on the efficiency of the heat transfer process in TSSC and the coordinate indicators of the control vector. The results of the studies are valid under the above restrictions, temperature Θ PC =36 °C, pressure P PC =22 MPa. Table 1 gives the results of the assessment of parametric sensitivity of coordinates of the control vector to the change in perturbing factors, performed using (8).

3. Development of the functional and software-technical structure of the control system for a technological system
The implementation of computer control technology of TSSC realized on the basis of a three-level hierarchical structure. The general structure of the contour management system for the flow rate of material flows and electricity is shown in Fig. 4.
The hierarchical structure of the system to control the contour of the flow rate of material flows and electricity predetermines the use of existing hardware and software at the first and second levels. At the third level, the human-machine interface unit consists of three modules. These modules provide process correction and modeling and are connected to a real-time database. Fig. 5 shows the software and technical structure of the automated control system of TSSC with a correction subsystem.
Data exchange is ensured through the use of "open" information technologies OPC and ODBC. This makes it possible to built a subsystem for correcting the modes of functioning and integrating it into a single information and control system.
Thus, the software-technical structure of a typical automated control system for TSSC with a correction subsystem takes the form depicted in Fig. 5.

4. Development of the structural-logical scheme of information flows for a computer control system of the "open" type
The functioning of the information interaction between the main nodes that ensure the quality of the product is shown in Fig. 6 in the form of a diagram of network information flows of a fragment of computer-integrated technology for TSSC. The structure of the scheme, which is illustrated in Fig. 6  network, as well as the mechanisms and technologies used in its implementation. According to Fig. 6, information streams connect the nodes of the correction subsystem, database, and laboratory to the upper level using the Ethernet local area network. At the same time, the subsystem receives primary data from the TDC-3000 control system and an additional VIPA controller using a serial RS-485 interface using special network equipment.  Fig. 6. Diagram of network information flows in a fragment of the computer-integrated technology for secondary condensation system

Discussion of results of studying the influence of the coordinates of the perturbation vector on the coordinates of the control vector
According to the results of our study by the method of mathematical modeling based on the developed algorithm (Fig. 1), we derived dependences, shown in Fig. 2, which characterize the effect of the temperature of the primary condensation Θ PC of on the TSSC thermal load. At the same time, the increase in Θ PC simultaneously affects both the increase in the concentration of ammonia 3 NH IN a in CG according to equation (1) and the temperature of CG at the input .

Θ
In addition, the latter increases even more due to the compression of CG with a compressor located between the stages of primary and secondary condensation.
For example, with an increase in the temperature Θ ПК from 32 °C to 36 °C, the temperature CG IT Θ increases from 41 °C to 45 °C, and the concentration of ammonia in CGfrom 9.95 % by volume to 10.16 % by volume The increase in the latter causes, according to equation (2), an increase in the concentration 3 NH AP а at the output of the AHE pipe space from 3.69 % by volume to 3.71 % by volume. In turn, the increase in  the consumption of working vapor M V should be increased from 7.41 t/h to 21 t/h per ejector, which requires an increase in the consumption of the MEA solution M MEA from 191 t/h to 541.34 t/h to the SRU steam generator. At the same time, the total load of ammonia vapor M TOT on SRU air capacitors would increase from 10.37 t/h to 29.4 t/h, which makes it necessary to increase the number of ACC fans operating from one to three. As a result, electricity consumption N should be increased from 200 kWh to 600 kWh, that is, all three ACC air fans would be enabled. In case of lowering the temperature Θ PC to 28 °C, which is observed at an atmospheric air temperature of less than 5 °C, the operation of SRU should be stopped at all. In such circumstances, cooling the CG to the operating temperature 5 LTE P = − Θ°C would be provided only in the evaporators of LTE with ARU.
The consumption of CG at the input CG IT V due to possible production requirements for changes in the performance of the ammonia synthesis unit also affects the thermal load of TSSC. As shown in Fig. 3 ). The nature of the impact of these indicators has been sufficiently justified by a preliminary explanation, and, therefore, Fig. 3 does not show these dependences. In such circumstances, the operation of SRU should be carried out under a maximum load; this is especially true for fans with ACC.
The analysis of the study's results, shown in Fig. 2, 3 and given in Table 1, explains the excessive amount of parametric sensitivity of the control vector under the conditions of changes in the temperature of the primary condensation, which, compared with the CG consumption at the input of TSSC, exceeds more than six times.
Designed to predict the modes of operation of such metal-intensive, and, therefore, too inertial objects as TSSC under the conditions of changing the coordinates of the perturbation vector allow it to be used in the subsystem of decision support under the existing uncertainties. This subsystem is one of the main components in the general technical structure of computer technology for controlling the process of secondary condensation of ammonia production [11]. However, the implementation of this structure requires the improvement of the existing control system based on the TDC-3000 with available hardware and software of the "open" type.
The unit of HMI (Fig. 4) includes three modules. The module for entering new data makes it possible to modify data tables for new components in a convenient form for the operator. The model configuration module is designed to manually enter the necessary structural data on evaporator, condensation column, and heat exchanger and start the process of checking the complete adequacy of models. Graphic screens provide a more detailed analysis of all information on the secondary condensation process intended only for the subsystem operator.
The separation of correction units and modeling units in the structure of the correction subsystem takes into consideration their final implementation since the model software is developed in the form of user libraries. The human-machine interface unit should provide the configuration of the subsystem and analysis of its work, as well as the construction and maintenance of the process database -the collection, processing, storage, and management of data on the technological process.
That has made it possible to develop a functional scheme of computer-integrated technology for TSSC, which is identical to the one built and reported in [6]. At the same time, a client-side module (OPC client) is embedded using the MATLAB programming environment, which provides free software access to the process data (Fig. 5). Regulators of technological parameters and classical regulators in cascading circuits are implemented as a separate control subsystem. The values of settings for control vectors are formed by the correction subsystem. However, the operator can independently make decisions and change them due to the level of supervisory control. This mode is especially important in the process of changing the flow rate of working steam M V and electricity N by changing the number of working ejectors and ACC fans.
The proposed correction subsystem is based on the functional structure to control the quality and amount of resulting product. Its structure includes the following components: -real-time databases (functionally, it is a process database); -algorithmic part (functionally, it is a correction unit and a modeling unit); -human-machine interface.
The configuration windows of all components, as well as display mnemonics schemes for the operator's work, vary depending on the structure of the entire secondary condensation system. At the same time, the first and third components are the main elements of the SCADA system, which makes it possible to implement a subsystem based on an existing tool.
Thus, the correction subsystem can be implemented by a separate network node within the system or integrated into the SCADA node of the secondary condensation system, as shown in Fig. 6. This subsystem is developed on the basis of the SCADA-program Zenon [17,18]. Connections of data sources depend on the method of connection of real-time databases (ACS of secondary condensation, laboratory workstation, etc.). In this case, OPC technology and ODBC technology are used [19].
Taking into consideration the requirements for the openness of the system, it is proposed to use the most common open communication technologies. These are the so-called OPC technologies.
The result of the use of the developed computer control technology for TSSC provides a decrease and stabilization of the secondary condensation temperature to −5 °C. Such a decrease causes a decrease in natural gas consumption, as already noted, in an additional steam boiler by almost 1 million nm 3 per year. Further work will aim at the further improvement of the technology to control condensation processes in order to reduce the load on the compressor system of ammonia synthesis department.

1.
We have developed an algorithm for the subsystem of decision support to the computer technology of TSSC control under the conditions of uncertainty, whose software implementation was carried out using the special MATLAB programming environment. This approach, owing to the client part embedded using the MATLAB environment (OPC client), provides free software access to the current data on the technological process.
2. A method of mathematical modeling based on the developed algorithmic software helped determine the regularity and quantitative dependences of the influence of external heat load on the performance indicators of heat exchange processes of TSSC under the established restrictions. These quantitative indicators make it possible to forecast the coordinates of the control vector in order to stabilize the temperature of secondary condensation at the regulatory level. Among these indicators, we distinguished the heat flows, a heat transfer coefficient, and the coordinates of the control vector. The regularity of increasing heat flows with an increase in the temperature of primary condensation has been established. As a result, there is a significant increase in the temperature load and coordinates of the control vector, despite some increase in ammonia in CG at the input of TSSC and a decrease in the heat transfer coefficient. The parametric sensitivity of the coordinates of the control vector from the change in the temperature of the primary condensation has been determined, which, compared with the change in the CG flow rate at the input of TSSC, exceeds by more than six times. We have built dependences of the coordinates of the control vector on the change in the external thermal load at the input of TSSC and determined the limit value of the primary condensation temperature at the level of 28 °C. At this temperature, it is possible for the operator to decide to stop the operation of SRU within TSSC.
3. We have designed a functional structure of the computer technology to control the contours of refrigerant flow rate to HTE and MEA solution to SRU steam generator, which uses the controller of free programming VIPA and SCADA-system Zenon.
4. Based on the use of "open" technologies, the hardware and software support to the computer-integrated control technology of TSSC was developed. This approach provided an opportunity to integrate it into the structure of the existing TDC-3000 control system and implement it in the form of automated operator's workplace of the decision-support subsystem.