Advanced process control and real-time optimization are techniques that can improve a plant’s profitability and efficiency by maintaining a process at desired operating conditions while taking process constraints into account.
by Ali Awais Amin, INTECH Process Automation
Process control is designed to minimize variations in processes and keep them within specified boundaries. The main purpose of a process control is to maintain a process at desired operating conditions while taking process constraints into account. Safety, environmental compliance, and process reliability must be maintained; optimal process controls help operations produce quality products, minimize costs, and respond to changing business conditions.
Process control strategies can be classified and organized in a hierarchy of process control activities. In the hierarchy shown in Figure 1, the required functions are at lower levels while the optional ones are at the higher levels. The time-scale of each activity is shown, as well.
The measurement and actuation functions (level 1) are the required part of any industrial control system. The measurement devices measure process variables, and the actuation equipment implements the calculated control actions. Process control plays a key role in ensuring process safety and protecting personnel, equipment, and the environment (level 2). Regulatory control techniques (level 3a) ensure that the control variables are maintained near the set point, and advanced control techniques such as multivariable and constraint control (level 3b) enable a process close to a limiting constraint. To determine the optimum controller set points for current operating conditions and constraints, real-time optimization (level 4) is employed. The highest level (level 5) is concerned with planning and scheduling operations for the entire plant. The successful implementation of these process control activities is a critical factor in making plant operation as profitable as possible.
Advanced process control (APC) is an umbrella term that covers a broad range of techniques and control methodologies, such as fuzzy logic and statistical control. The common objective is to manage complex interactions within a process in such a way to reduce process variability and allow the plant to run closer to the operating constraints. This results in higher energy efficiency and product quality.
One of the important advanced control techniques is model predictive control (MPC). A simplified block diagram of a model predictive control system is shown in Figure 2.
MPC is used to optimize hybrid systems with large multivariable constrained control issues. MPC is implemented to reduce variation in process variables in industrial applications, which in turn leads to an increased throughput and higher profit.
MPC is a multivariable strategy that encompasses constraints, handling of actuators, states, process outputs, and other variables. It brings a structured approach to solutions where the main aim is to minimize a performance criterion in the future. The criterion would possibly be subject to constraints on the manipulated inputs and outputs, and the future behavior is computed according to a model of the plant. MPC utilizes an internal dynamic model of the process, a history of past control moves, and an optimization cost function over the receding prediction horizon to calculate the optimum control needed.
The receding control horizon principle can be summarized as follows:
- Taking into account the current and future constraints, at time and for the current state xi, solve an optimal control problem over a fixed interval, say [ιi, i + N-1]
- Apply only the first step in the resulting optimal control sequence
- At time i + 1, measure the state reached
- At time i + 1, repeat the fixed horizon optimization over the fixed interval [i + 1, i + N], starting from the current state xi+1.
The optimum operating conditions for a plant are determined as part of the process design. However, during plant operations, due to changes in equipment availability, economic conditions, and process disturbances, for example, the optimum conditions change frequently over the course of time. Hence, the optimum operating conditions need to be re-calculated on a regular basis. This control activity is defined as real-time optimization (RTO) (level 4) in the hierarchy discussed earlier. RTO utilizes the plant operating conditions for variables to predict properties such as product characteristics. A suitable problem statement needs to be formulated and solved once a process has been selected. For RTO, optimization of set points requires two models-the economic model and the operating model. The economic model consists of an objective function that needs to be maximized or minimized. This includes costs and product values. The operating model is a steady-state process model and contains all process variable constraints.
Figure 3 shows the steps in solving any practical RTO problem.
The input and output variables for the process that are identified in this step are employed in the process model and the objective function. The next step is to select an objective function based on operating profit, product qualities and quantities, as well as plant configuration. The third step is to formulate steady-state process models and identify the operating limits for the process variable. To ensure compatibility with the most effective solution techniques, it is important to simplify the model as well as the objective function at this stage. The fourth step involves calculating the optimum set points after choosing an optimization technique. In the last step the most sensitive parameters in the optimization systems are identified through varying model and cost parameters.
In the example shown in Figure 4, implementing an advanced control strategy project can be carried out in four phases.
Phase 1: Benefits estimation
The decision to implement any advanced control strategy is based on a cost-benefit analysis. This phase is the most important one in implementing a control strategy for a plant process. Mistakes lead to incorrect cost estimates, which will have negative consequences on the project.
Phase 2: Process modeling and algorithm implementation
Modeling process dynamics and configuration of the real-time database and the controller are required before implementation of advanced control algorithms. First, plants tests are carried out to obtain a model for process dynamics. After the completion of plant tests, a real-time database is designed. The next step is to define the communication with the process, determining and establishing the protocol between a workstation and the DCS.
The process model is derived from process identification. For the controller configuration, the manipulated, controlled variables and constraints are defined. The modules designed and programmed are then integrated followed by program tests to ensure the integration of these modules with the real-time operating system.
Phase 3: Commissioning
Commission of an advanced control project should start after interfacing the advanced control system with the existing plant infrastructure. First, the communication with the existing plant control system is tested. Next, each multivariable controller module must be tested in an open-loop model.
For the next step, each control loop is tuned and the controller actions are tested and closed in an advisory model by the operator. It is then closed on automatic mode after each controller is successfully tested. The optimization package can then be commissioned after the multivariable controllers has exhibited the desired performance.
Phase 4: Maintenance
The last phase is maintenance, which is required to ensure continuous benefits from the implemented advanced controls. Advanced controller algorithms need regular analysis. When product specifications and process conditions change or new product specifications are added, re-tuning of the controller is required.
This article was originally published in Control Engineering; INTECH Process Automation’s content partner.