Computer Process Control

COMPUTER PROCESS CONTROL

Computer Process Control

• Control Requirements

• Capabilities of Computer Control

• levels of Industrial Process Control 

The use of digital computers to control industrial processes had its origins in the continuous process industries in the late 1950s (Historical Note 4.1). Prior to then, analog controllers were used to implement continuous control, and relay systems were used to implement discrete control. At that time, computer technology was in its infancy, and the only computers available for process control were large, expensive mainframes. Compared with today's technology, the digital computers of the 195Ch;were slow, unreliable, and not well suited to process control applications. The computers that were installed sometimes cost more than the processes they controlled. Around 1960, digital computers started replacing analog controllers in continuous process control applications; and around 1970, programmable logic controllers started replacing relay hanks in discrete control applications. Advances III computer technology Since the 1960s and 1970.~have resulted in the de:elopm~nl of the microprocessor. Today, virtually all industrial processes, certainly new installations, are controlled by digital computers based on microprocessor technology. Microprocessorbased controllers are discussed in Section 4.4.6.

The available computers In the late I950s were not reliable, and most of the subsequent process control installations operated by euher printing out instructions for the operator or by making adi\lstments in the set points (If analog controllers, thereby reducing the risk of process downtime due to compute: problems. The latter mode of operation was called set point conlrol. By March 1%1, a total of 37 computer process control systems had been installed. Much experience was gamed from these early installations. The interrupt feature (Section 4.3.2). by which the computer suspends current program execution to quickly respond to a process need. was developed during this period.

The first direct digitai control (DOC) system (Section 4.4.2), in which certain analog de" vices are replaced by the computer, was installed by Imperial Chemical Industries in England in 1%2. In this onplementauon.zza process variables were measured, and 129actuators (valves) wcr~ controlledImprovements in DOC technology were made, and additional systems were installed during thc 1%0s. Advantages 01 DOC noted during this time included: (1) cost savings from elimination of an~l"g instrumentationfor large systems. (2) simplified operator dis_ play pencts, and ,:3) flexibility through reprogramming capability.

Computer technology was advancing. leading to the development  of the minicomputer

in thelate 1960s.Process cnntrol applications were easier to justify using theses maller, lessexpensive computers. Development of the microcomputer in the early 1970s continued this trend. Lower cost process conlrol hardware and Interface equipment (such as analogtodlgltal converters) were becoming available due to the larger markets made possible by lowcost computer controllers.

Most of the developments in computer process control up to this time were biased toward theprocessinduslriesrath",rthandiscretc part and product manufacturing. Just as analog devices had been used to automate process industry operations, relay banks were widely used to satisfy the ciscrcte process control (ON/OFF)  requirements in manufacturing automation.

Let us consider the requirements placed on the computer in industrial control applications. We then examine the capabilities that have been incorporated into the control computer to address these requirements. and finally we observe the hierarchical structure of the functions performed by the control computer.

       Control Requirements

Whether the application involves continuous control, discrete control. or both, there are certain basic requirements that tend to he common to nearly all process control applications. By and large, they are concerned with the need to communicate and interact with the process on arealtime basis.A reallime controller is able to respond to the process within a short enough time period that process performance is not degraded. Factors that determine whether a computer controller can operate in realtime include: (1) the speed of the controller's central processing unit (CPU) and its interfaces, (2) the controller's operating system, (3) the design of the application software, and (4) the number of different input/out. put events to which the controlier is designed to respond. Realtime control usually requires the controller to be capable of multitasking, which means coping with multiple tasks concurrently without the tasks interfering with one another.

There are two basic requirements that must be managed by the controller to achieve realtimeeontrol:

   Processinitiated interrupts. The controller must be able to respond to incoming signals from the process. Depending on the relative importance of the signals, the computer may need to interrupt execution of a current program to service a higher priority need of the process. A processinitiated interrupt is often triggered by abnormal operating conditions, indicating that some corrective action must be taken promptly.

    Timerinitialed actions. The controller must be capable of executing certain actions at specified points in time. Timerinitiated actions can be generated at regular time intervals,ranging from very low values (e.g., 100 j.ts) to several minutes. or they can be generated at distinct points in time. Typical timerinitiated actions in process control include: (1) scanning sensor values from the process at regular sampling intervals, (2) turning on and off switches, motors, and other binary devices associated with the process at discrete points in time during the work cycle, (3) displaying performance data on the operator's console at regular times during a production run, and (4) recomputing optimal process parameter values at specified times.

These two requirements correspond to the two types of changes mentioned previously in the context of discrete control systems: (l) eventdriven changes and (2) timedriven changes.

In addition to these basic requirements. the control computer must also deal with other types of interruptions and events. These include:

Computer commands to process. In addition to incoming signals from the process, the control computer must be able to send control signals to the process to accomor readjust a 5ct point in 5iyrtem and programmitiated IT('nts. These are events related to the computer systerri itself 'lhcv arc similar to the kinds of corr.puter operations associated with business and eng:ineering applications of computers. A systeminitiated event involves communications among ~ompl!ten and peripher<ll devices linked together in a network. In these multiple computer networks, feedback signals, control commands, and other data must be transferred back and forth among the computers in the overall control of the process. A programinitiated {'vent is when some nonprocessrelated action is called for in the program.such as the pr.ntiug or display of reports on a printer or monitor. In process coutrut, vvstern and programinitiated events generally occupya low level of priority compared with process interrupts, commands to the process. and timerinitiated events

Capabilities of Computer Control

The above requirements can he satisfied by providing the controller with certain capabttitics that allow it 10 interact on a realtime basis with the process and the operator, The capabilities are: (1) polling, (2) interlocks. (3) interrupt system, and (4) exception handling.

Polling (Data Sampling). In computer process control,polllNg refers 10 the periodic ~ampling of data that indicates the status of the process. When the data consist of a continuous analog ~ignal. sampling means that the continuous slgnalrs substituted with a series of numerical values that represent the continuous signal at discrete moments in time. The same kind of substitution holds for discrete data. except that the number of possible numerical values the data can take on i~more limitedccertainly the case with binary data We discuss the techniques by which continuous and discrete datil are entered into and transmitted from the computer in Chapter 5. Other names used for polling include sampling and scanning

In some polling procedure simply requests whether any changes have the last polling cycle and then collects only the new data from shorten the cycle time required for polling. Issues related to

    Polling order. The polling order is the sequence in which the different data collection points of the process are sampled.

Polling formal. This refers to the manner in which the sampling procedure is designed. The alternatives include: (a) entering all new data from all sensors and other devices ~very polling cycle; (b) updating the control system only with data that have changed since the last polling cycle; or (c) using highlevel and towlevdscanntng.ot conditional scanning, in which only certain key data are normally collected each polling cycle (highlevel scanning), but if the data indicates some irregularity in the process. a lowlevel scan i~undertaken to collect morecomplete data to ascertain the source of the irregularity.

These issues become increasingly critical with very dynamic processes in which changes in process S(;JIII, occur rapidly

Interlocks. An inrerlock is a safeguard mechanism for coordinating the activities of two or more devices and preventing vue device from interfering with the otherts). In process control. interlocks provide a means by which the controller is able to sequence the activities in a work cell, ensuring that the actions of one piece of equipment arc completed before the next piece of equipment begins its activity. Interlocks work by regulating the flow of control signals back and forth between the controller and the external devices.

There are two types of interlocks, input interlocks and output interlocks, where input and output are defined relative to the controller. An input interlock is a signal that originates from an external device (e.g., a limit switch, sensor, or production machine) and is sent to the controller. Input interlocks can be used for either of the following functions:

    To proceed with the execution of the work cycle program. For example, the production machine communicates a signal to the controller that it has completed its processing of the part. This signal constitutes an input interlock indicating that the controller can now proceed to the next step in the work cycle, which is to unload the part.

    To interrupt the execution of the work cycle program. For example, while unloading the part from the machine, the robot accidentally drops the part. The sensor in its gripper transmits an interlock signal to the controller indicating that the regular work cycle sequence should be interrupted until corrective action is taken.

An output interlock is a signal sent from the controller to some external device. It is used to control the activities of each external device and to coordinate its operation with that of the other equipment in the cell. For example, an output interlock can be used to send a control signal to a production machine to begin its automatic cycle after the workpart has been loaded into it.

lnterrupt System. Closely related to interlocks is the interrupt system. As suggested by our discussion of input interlocks, there are occasions when it becomes necessary for the process or operator to interrupt the regular controller operation to deal with morepressing matters. All computer systems are capable of being interrupted; if nothing else, by turning off the power. A moresophisticated interrupt system is required for process control applications. An interrupt system is a computer control feature that permits the execution of the current program to be suspended to execute another program or subroutine in response to an incoming signal indicating a higher priority event. Upon receipt of an interrupt signal, the computer system transfers to a predetermined subroutine designed to deal with the specific interrupt. The status of the current program is remembered so that its execution can be resumed when servicing of the interrupt has been completed.

Interrupt conditions can be classified as internal or externaL Internal interrupts are generated hy the computer system itself. These include timerinitiated events, such as polling

of data from sensors connected to the process, or sending commands to the process at specific points 10 clock time. System and programinitiated interrupts are also classified as

important programs higher priority) Be executed before Jessimportant programs (ones with lower prioruiesj.Thc system designer must decide what level of priority should be attached to each control function. A higher priority function can interrupt a lower primity function. A function at a given priority level cannot interrupt a function at the same priority level. The number of priority levels and the relative importance of the functiom depend on the l<:'[uir<:I1n;nl>of the individual process control situation. For example, emergency shutdown of a process because of safety hazards would occupy a very high priority level, even though it may be an operator. initiated interrupt. Most operator in" puts would have low priorities.

One possible organization of priority rankings for process control functions is shown IIITable 4.4. Of course, the priority system may have more or less than the number of levcis shown here. depending on the control situation. For example, some process interrupts may be more important than others. and some system interrupts may take precedence over certain process interrupts, thus requiring more than the six levels indicated in our table

To respond 10 the various levels of priority defined for a given control application, an interrupt system can have one or more interrupt levels. A singlelevel interrupt system has only two modes of operation: normal mode and interrupt mode. The normal mode can be interrupted, but the interrupt mode cannot. This means that overlapping interrupts arc serviced on a firstcome. firstserved basis, which cOllld have potentially hazardous consequences it an important process interrupt was forced to wait its tum while a series of lessimportant operator and system interrupts were serviced. A multilevel interrupt system has a normal operating mode plus more than one interrupt level. The normal mode can be interrupted hy any interrupt level. hut the interrupt levels have relative priorities that determine which functions can interrupt others. EXample 4.1 illustrates the difference between the singlelevel and multilevel interrupt systems.

EXAMPLE     4.1        SingleLevel Versus Multilevel Interrupt Systems

Three Interrupts representing tasks of three different priority levels arrive for service in the reverse order of (heir respective priorities. Task 1with the lowest priority, arrives first. Shortly later.higher priority Task 2 arrives. And shortly later,

highest pnority Task 3 arrives. How would the computer control system respond under (a) a suvglcle ce! interrupt system and (b) a rnurt'ilevcl interrupt system?

Exception Handling. In process control, an exception is an even! that is outside the norma! or desired operation of the process or control system. Dealing with the exception is an essential function in industrial process control and generally occupies a major portion of the control algorithm. The need for exception handling may be indicated through the normal polling procedure or by the interrupt system. Examples of events that may invoke exception handling routines include:

   product  quality  problem

   process  variables  operating  outside  their  normal  ranges

   shortage  of raw materials  or supplies  necessary  to sustain  the process

   hazardous  conditions  such as a fire

   controller  malfunction

In effect, exception handling is a form of error detection and recovery, discussed in the context of advanced automation capabilities (Section 3.2.3).

          Levels of Industrial  Process Control

In general. industrial control systems possess a hierarchical structure consisting of multiple levels of functions, similar to our levels of automation described in the previous chap. ter [Table 4.2). ANSIIISAS88.011995^J [1] divides process control functions into three

levels: (1) basic control, (2) procedural control, and (3) coordination contro1.These control levels map into our automation hierarchy as shown in Figure 4.7, We now describe the three control levels, perhaps adapting the standard to fit our own models of continuous and discrete control (the reader is referred to the original standard [1], available from the Instrument Society of America)

Basic Control. This is the lowest level of control defined in the standard, corresponding to the device level in our automation hierarchy. In the process industries, this level is concerned with feedback control in the basic control loops. In the discrete manufacturing industries, basic control is concerned with directing the servomotors and other actuators of the production machines. Basic control includes functions such as feedback control. polling, interlocking, interrupts. and certain exception handling actions. Basic control functions may be activated.deactivated, or modified by either of the higher control levcis (procedural or coordination control) or by operator commands.

Procedural Control. This intermediate level of control maps into regulatory control of unit operations in the process industries and into the machine level in discrete manufacturing automation (Table 4.2). In continuous control, procedural control functions include using data collected during polling to compute some process parameter value, changing serpoints and other process parameters in basic control, and changing controller gain constants, In discrete control, the functions are concerned with executing the work cycle program, that is.directing the machine to perform actions in an ordered sequence to accomplish sornc productive task. PI ocedural control may also involve executing error detection and recovery procedures and making decisions regarding safety hazards that occur during the process,

Coordination Control. This is the highest level in the control hierarchy in the A'JSI/ISA standard. It corresponds to the supervisor} level in the process industries and the cell or system level in discrete manufacturing. It is also likely to involve the plant and possibly the enterprise levels of automation. Coordination control initiates, directs, or alters the execution of programs at the procedural control level. Its actions and outcomes change over rime. as in procedural control. but its control algorithms are not stnleturoo for a specific processoriented task. It IS more reactive and adaptive. Functions of coordination control at the celllevel include.coordinating the actions of groups of equipment or machines.coordinating material handling activities between machines in a cell or system, allocating production orders to machines in the cell, and selecting among alternative work cycle programs.

At the plant and enterprise levels. coordination control is concerned with manufacturing support functions, including production planning and scheduling; coordinating common resources. such as equipment used in more than one production cell; and supervising availability, utilization. and capacity of equipment. These control functions are accomplished through the company's integrated computer and information system.