Who has questions related to advanced process control APC? Where is it?

They even create *questions* for things like this? I’ve never heard of such a thing before! Are these exam questions designed for college students studying control?

We have used Honeywell’s RMPCT controllers here, but the advanced control performance was not satisfactory; they were stopped from being used after being in operation for a while. I’m not sure which company’s advanced controller you are using; different controllers have different operating methods.

I. Basic Knowledge 1.1 PID Control: Since the 1930s, PID controllers have been used for controlling production processes. Over time, various single-loop control methods have been developed, such as the process response curve method, the first-order with time delay model method, the integral error detection method, the direct synthesis method, internal model control, and robust PID control. These methods are all based on the control concept of single input (or multiple inputs) and single output. 1.2 Advanced Control and Optimal Control: As for Honeywell’s Advanced Process Control (APC) technology, we refer to it as Robust Multivariable Predictive Control Technology (RMPCT). It is a more advanced control strategy that has evolved from Model Predictive Control (MPC). Its basic idea is to use a dynamic process model to predict future changes in CV based on process inputs (control variables MV and disturbance variables DV), and to determine how to adjust the controller outputs so that all process variables (including the controlled variable CV and the control variables MV) reach their set values or remain within specified limits (this is what is known as advanced control). Then, if there are degrees of freedom (number of degrees of freedom = number of MVs not within the constraint limits – number of CVs that have reached the set value or the constraint limits or are outside those limits), the controller further adjusts the process to achieve operational optimization (so-called optimal control). 1.3 Several Basic Concepts 1. Model: A mathematical relationship established using measurable device data and empirical knowledge to reflect the dynamic characteristics of a process. 2. Predictive: The future value of a process variable that is calculated using dynamic process models and data from measurable devices. 3. Control: Dynamic process models and prediction techniques are used to calculate the changes in the control variables (MV) in order to minimize the deviation of the process variables (CV) from their target values (TARGET). 4. Robustness: This is the most distinctive feature of RMPCT; even in the presence of errors in the process model, it is able to maintain the reliability of process control involving numerous interrelated variables. 5. Multi-variables: The controller is used to handle process systems that involve multiple inputs and outputs, with highly complex dynamic and thermodynamic relationships. 6. Manipulated Variables (MV): These are the variables that the controller needs to adjust in order to keep the CV within its specified limits, or to optimize the process, without exceeding the constraints of that CV itself. The MV has a process value, and its valve position can be adjusted via the DCS to control the process variable value in real time. Such as feed temperature, side-stream withdrawal volume, etc. 7. Controlled Variable CV: A variable that the controller attempts to maintain at its set point or within a certain operational range; it is the primary target that the controller strives to control. Generally speaking, there are process values, or there may be no process values but they can be determined using certain methods (such as testing methods or empirical formulas); they are variables that cannot be controlled directly by changing the valve position. Such as the product’s dry point, flash point, etc. 8. Disturbance Variables: For these variables, the controller cannot control them (they may originate from upstream processes) or it is difficult to control them, yet they have an impact on the CV. Based on the DV’s prediction of the future impact on CV, the controller can take action before CV deviates from the constraints. This actually provides a feedforward signal to the controller. 9. Step Test and Model Identification: These are techniques used to determine the mathematical relationship between MV and DV and their impact on CV, by applying a step change in MV or DV (the appropriate magnitude, direction, and number of changes can be selected based on practical considerations). Record the response curves of the corresponding CVs to obtain the data required for model identification. Model identification is the process of using a set of data obtained from step tests, and through some mathematical method, deriving a mathematical relationship between CV and MV or DV. This project does not use the method of performing step tests followed by model identification; instead, it employs a method based on the analysis of the process mechanism to derive the mathematical model, thereby avoiding the disruption caused by step tests to the production process. 10. Key variables and non-key variables: Variables that have a significant impact on the production process are key variables, while those with a lesser impact are non-key variables. It is necessary to determine these based on the importance of each variable in the actual production setup. All variables of the RMPCT controller in this device are configured as non-critical variables. 11. Set-point control and interval control: For any CV, set-point control and interval control are possible. The greatest advantage of using interval control is that it increases the degree of freedom of the control system, as well as its robustness, and makes it easier to determine the control solutions. Given point control can be regarded as interval control with the lower limit equal to the upper limit. 12. Slope limitation: When the operator makes significant changes to the CV limit or set value, slope limitation is used to restrict the rate at which the old limit moves toward the new one, thereby allowing it to adjust gradually and preventing sharp changes in the device’s parameters. Both CV and MV can have a limited climbing rate set (configured by the engineer). II. Simplified Structure of Advanced Control Systems An advanced controller system is equipped with a dedicated computer for performing calculations by the advanced controller. The variables required for advanced control are collected via the DCS; after the calculations are completed, the results generated by the advanced controller (the set values for the control variables) are sent to the DCS as set values for the conventional controllers, thereby adjusting the operating conditions of the equipment to achieve the objectives of advanced control. Currently, the operating cycle of the advanced controller is initially set at 3 minutes. III. Advanced control aspects of this device: Since advanced control requires direct multivariable constraint control over product quality parameters that cannot be measured in real time online, it is first necessary to calculate these product quality parameters using the operating conditions of the device (i.e., so-called soft sensors). Therefore, the calculation of these soft sensors and the calibration using laboratory data form the basis for implementing advanced control. The advanced control of this unit includes the following components: the RMPCT multivariable predictive controller for the initial distillation tower ； Atmospheric pressure column RMPCT multivariable predictive controller ； Multivariable predictive controller for the vacuum distillation column RMPCT ； Atmospheric pressure furnace branch temperature difference balance controller (implemented via DCS configuration) ； Reduced-pressure furnace branch temperature difference balance controller (implemented via DCS configuration) ； Thermal efficiency controllers for atmospheric pressure furnaces and reduced-pressure furnaces (cannot be used at the moment due to issues with the oxygen concentration sensors in the flue gas; they will be put into use once the sensors are replaced) ； Soft instrument calculation and laboratory data correction.

Companies and products related to predictive control: Setpoint Inc.: SMC-IDCOM Setpoint Multivariable Control; DMC Corp.: DMC Dynamic Matrix Control; AspenTech: DMCplus, improved versions of DMC; Profimatics: PCT Predictive Control Technology; Honeywell: RMPCT, Robust Model Predictive Control; Adersa (France): HIECON, Hierarchical Constraint Control. Zhejiang University, Foxboro: Connoisseur Predictive Control Ltd.—Foxboro—Invensys; DOT Ltd.: STAR Dynamic Optimization Technology. Collaborations with ABB and others; MDC Tech.: SMOC; Shell Multivariable Optimizing Control; Yokogawa, Petroleum University: UPCC, Unified Predictive Coordinate Control; Qingdao Dongfei: SMART, similar to RMPCT; Zhejiang University Zhongkong: APC-Suit – an overall concept. Domestic applications: research and development; foreign product applications; engineering applications in development……

Concept of advanced control, components of advanced control systems, control strategies for advanced control systems, optimization modules of advanced control systems. Five questions each on the above topics, without focusing on any specific system. Who wants to give it a try? :lol

Are there any examples where it can be used for a long time in projects and is more effective than conventional controls? :)

Thinking too much about others – one must live for oneself!