Process control algorithms

In order for the data generated by the analyzer to be useful, it must be transferred to the operation s centralized control or host computer and made available to process control algorithms. Vendor packages manage instrument control and can do spectral interpretation and prediction or pass the data to another software package that will make predictions. Most vendors support a variety of the most common communications... 

Computers continued to get smaller, faster, and cheaper. The accuracy and speed of chemical analysis continued to improve. With real-time chemical analysis of process streams and a mathematical model for each unit in the process, the computer was programmed to set controllers on process streams using the new process control algorithms. With some additional programming and with the process control set points adjusted by the computer control system, the quality of the product improved. The physical property data for the chemicals in the process streams required for these model calculations were built into the process control program. 

In most areas of production, the output of the plant must be monitored. The simplest monitoring occurs when random or systematic samples are pulled and taken offline to get the analysis results in a quality control (QC) laboratory. These results are then fed into a statistical process control algorithm or software in order to check the consistency of the plant output. 

To implement Ml image process control, an IR line scanner is placed at the oven exit, scaiming across each blank as it leaves the oven. The forward movement of the part, combined with the fast response of the line scanner, provides the Ml thermal map to the computer. A Ml statistical control chart is software-generated, based on mean temperature and standard deviation over the entire surface. The oven heater is zoned and the computer interfaced to a relay control board with one relay managing a set-point controller for each zone. The statistics for each control zone are calculated based on the corresponding section on the part, and a closed-loop statistical process control algorithm is implemented. The result is seen on the 3D strip map presentation shown in Fig. 10.8(b). The color hues indicate temperatures in accordance with the scale at the left. The control level temperature is about 300°F. 

Some of the inherent advantages of the feedback control strategy are as follows regardless of the source or nature of the disturbance, the manipulated variable(s) adjusts to correct for the deviation from the setpoint when the deviation is detected the proper values of the manipulated variables are continually sought to balance the system by a trial-and-error approach no mathematical model of the process is required and the most often used feedback control algorithm (some form of proportional—integral—derivative control) is both robust and versatile. 

Classical Feedback Control. The majority of controllers ia a continuous process plant is of the linear feedback controller type. These controllers utilize one or more of three basic modes of control proportional (P), iategral (I), and derivative (D) action (1,2,6,7). In the days of pneumatic or electrical analogue controllers, these modes were implemented ia the controller by hardware devices. These controllers implemented all or parts of the foUowiag control algorithm ... 

The second area, the implementation of a modem process monitoring and control system, is the most dramatic current appHcation of CAD/CAM technology to the chemical process industry. The state of the art is the use of computer graphics to display the process flow diagram for sections of the process, current operating conditions, and controUer-set points. The process operator can interact directly with the control algorithms through the... 

This is a control algorithm that attempts to eliminate the offset (caused by proportional control) between the measurement and the setpoint of the controlled process variable. This control mode remembers how long the measurement has been off the setpoint. 

Process control and safety shutdowns must be provided during all modes of operation, not only in the RUN mode. Other modes will require a BPCS configured for the mode operating algorithm and very likely a different set of safety interlocks must provide appropriate protection. Hardwire devices, like timers or software logic, can be used to actuate the SIS pertinent to the operating mode. 

The complexity and reduced time constants of modern processes imply the adoption of high performance programmable controllers. This requires not only higher processing speed but also more advanced control algorithms that can optimize the process operation in real time. 

One final note While the techniques used here were applied to control temperature In large, semi-batch polymerization reactors, they are by no means limited to such processes. The Ideas employed here --designing pilot plant control trials to be scalable, calculating transfer functions by time series analysis, and determining the stochastic control algorithm appropriate to the process -- can be applied In a variety of chemical and polymerization process applications. 

Multiscale process identification and control. Most of the insightful analytical results in systems identification and control have been derived in the frequency domain. The design and implementation, though, of identification and control algorithms occurs in the time domain, where little of the analytical results in truly operational. The time-frequency decomposition of process models would seem to offer a natural bridge, which would allow the use of analytical results in the time-domain deployment of multiscale, model-based estimation and control. 

---