You are probably reading this indoors. It is probably comfortable because somewhere on the wall is a thermostat. This device measures the temperature in the room and strives to keep it at a comfortable setting.
Is this Advanced Process Control (APC)?
Most of us would not regard this simple duty cycle algorithm as advanced. If this is not advanced, where do we draw the line?
The next step up is a PID loop. Unlike the on-off duty cycle control, PID is a second order differential equation that requires some degree of expertise to tune it properly. Despite the fact that it is ubiquitous and ancient, many engineers have not mastered the configuration and tuning of PID loops. While many would not regard PID as advanced, it seems to be more advanced than the skillset required by many that use it.
The reason I am bringing this up in the context of a newsletter devoted to Industry 4.0 is that there is an impression that APC is a new capability in the Industry 4.0 portfolio. We need to define advanced control to know whether there is anything new here.
If PID is not advanced, the next level would be variants of single loop algorithms. By this, I mean a control loop like PID that only has one measurement, one setpoint, and one actuator for control. Examples would be deadtime compensation algorithms like IMC, Smith Predictor, and Dahlin. Another example is a Fuzzy Logic controller, which uses fuzzy membership functions instead of differential equations to determine control actions. Also, a Neural Network can be used to provide a virtual sensor prediction of a property that cannot be measured online, such as tensile strength, and this could be used as the controlled measurement in a PID loop. All of these could be considered advanced control.
At a more advanced level are algorithms that control multiple measurements and actuators with an optimizer to select the best settings subject to economic or physical constraints. The Dynamic Matrix Control (DMC) algorithm that originated in the 1980s sparked the development and application of Multivariable Predictive Control (MPC). Clearly, these are advanced control techniques.
Of note is that all of these techniques predate Industry 4.0. Therefore, we must ask if Industry 4.0 provides anything new in advanced control.
If you go above MPC, there can be batch and Manufacturing Execution Systems (MES). These applications can automate the scheduling of production from the corporate level to the plant operation. Setpoints and economic constraints in MPC can be dynamically and automatically updated. The December issue of InTech magazine included the article "Integrating Production Planning using APC and Other Technologies" by Simon Rogers, in which he describes such an approach.
Is this APC?
Well, it certainly is advanced. It is very pertinent to Industry 4.0 since it relies upon connectivity to get data from one point to another in a cost effective and secure manner.
However, the skillset involved here has to do with data communication and databases. It does not require the math of differential equations, optimization, and neural networks. A control engineer that is fully competent in PID through MPC may be out of their element when they try to put MES on top of it. It seems to be a different discipline that might not fit the definition of APC.
Industry 4.0 does indeed bring us into an era when automation can extend beyond the unit operation for a plantwide and corporate wide closed loop control. The fundamental math of APC is not part of that offering since it was born long before this era. However, Industry 4.0 does offer the possibility of a more advanced means of control that might enable someone to achieve optimum performance from the comfort of their climate controlled corporate office.
