Forms of Computer Process Control

        FORMS  OF COMPUTER  PROCESS  CONTROL

        There are various ways in which computers can be used to control a process. First, we can distinguish between process monitoring and process control as illustrated in Figure 4.8. In process monitoring. the computer is used to SImply collect data from the process, while in process control, the computer regulates the process, In some process control implementations, certain actions are implemented by the control computer that require no feedback data to be collected from the process. This is openloop control However, in most cases, some form of feedback or interlocking is required to ensure that the control instructions have been properly carried out. This more common situation is closedloop control.

In this section. we survey the various forms of computer process monitoring and control, all hut one of which <Hecommonly used in industry today, The survey covers the following categories. (1) computer process monitoring. (2) direct digital control. (3) numerical control and robotics. (4) programmable logic controllers.Iy) supervisory control, and (6) distributed control systems and personal computers. The second category, direct digital control, represents a transitory phase in the evolution of computer control technology, In its pure [OM, it is no longer used today. However. we briefly describe DDC to expose the opportunitiex it contributed. The sixth category. distributed control systems and personal computers. represents the most rCCCIl!means of implementing computer process control.

          Computer  Process  Monitoring

Computer process monitoring is one of the ways in which the computer can be interfaced wnh a process. Computer process monitoring involves the use of the computer to observe the process and associated equipment and to collect and record data from the operation The computer is not used to directly control the process. Control remains in the hands of humans who use the data to guide them in managing and operating the process

The data collected by the computer in computer process monitoring can generally be classifi<.;l! into three categories:

~               Process datil. These are measured values of input parameters and output variable, that indicate process performance, When the values are found to indicate a problem, the human operator takes corrective action.

~               Equipment data. These data indicate the status of the equipment in the work cell. Functions served by the data include monitoring machine utilization, scheduling tool changes, avoiding machine breakdowns. diagnosing equipment malfunctions, and planning preventive maintenance

~               Product data. Government regulations require certain manufacturing industries to collect and preserve production oeta on their products, The pharmaceutical and medical supply industries are prime examples. Computer monitoring is the most conveOlen, means of satisfying these regulations. A firm may also want 10 collect product data for its own use

Collecung data from factory operations can be accomplished by any of several means Shop data can be entered by workers through manual terminals located throughout the plant or can be collected automatically by means of limit switches, sensor systems, bar code readers, or other devices. Sensors arc described in Chapter 5 (Section 5.1). Bar codes and similar automatic identification technologies are discussed in Chapter 12. The collection and use of production data in factory operations fer scheduling and tracking purposes is called shop floor control. explained in Chapter 26

       Direct Digital Control

Direct digital control was certainly one uf the important steps in the development of computer proc~s.s cont.ro\. Let us briefly examine this computer control mode and its limitations, which motivated rrnprovernents leading to modern computer control technology. DIrect dlgrwl cmllror (DDC) IS a computer process control system in which certain components in a conventional analog control system are replaced by the digital computer. The regulation of the process is accomplished by the digital computer on a timeshared, sampleddata basis rather than by the many individual analog components working in a dedicated continuous manner. With DDe, the computer calculates the desired values of the input parameters and set points, and these values are applied through a direct link to the process; hence the name "direct digital" control

The difference between direct digital control and analog control can be seen by comparing Figures 4.9 and 4.10. The first figure shows the instrumentation for a typical analog control loop. The entire process would have many individual control loops, but only one is shown here. Typical hardware components of the analog control loop include the sensor and transducer, an instrument for displaying the output variable (such an instrument is not always included in the loop), some means for establishing the set point of the loop (shown us a dial in the figure, suggesting that the setting is determined by a human operator). a comparator (to compare set point with measured output variable), the analog controller. amplifier, and actuator that determines the input parameter to the process.

In the DDC system (Figure 4.10), some of the control loop components remain unchanged, including (probably) the sensor and transducer as well as the amplifier and actuator. Components likely to be replaced in DDC include the analog controller, recording

as a more efficient means of performing the same kinds of control analog components it replaced. However, the practice of simply using the digital computer to imitate the operauou of analog controllers seems to have been a transitional phase in computer process control. Additional opportunities for the control computer were soon recogru/ed, including:

   More control options than traditional analog. With digital computer control. it is pasvible to perform morecomplex control algorithms than with the conventional proportionalintegralderivative control modes used by analog controllers; for example.on/off control or nonllnearities in the control functions can he implemented.

   lntegrution and optimization of multiple loops, This is the ability to integrate feedback measurements from multiple loops and to implement optimizing strategies to improve overall process performance.

   Edumg the control prowams. Using a digital computer makes it relatively easy to change the control algorithm if that becomes necessary by simply reprogramming the computer. Reprogramming the analog control loop is likely to require hardware changes that are more costly and less convenient.

These enhancements have rendered the original concept of direct digital control more or less obsolete. In addition, computer technology itself has progressed dramatically so that much smaller and lessexpensive yet morepowerful computers are available for process control than the large mainframes available in the early 1960& This has allowed computer process control to be economically justified for much smaller scale processes and equipment. It has also motivated the use of distributed control systems, in which a network of rnicrocomputers is utilized to control a complex process consisting of multiple unit operations and/or machines.

       Numerical  Control and Robotics

Numerical C(>n1!01(NC) is another form or industrial computer control. I! involves the use of the computer (again, a microcomputer) to direct a machine tool through a sequence of processing slep~ defined by a program of instructions that specifies the details of each step and their sequence. The distinctive feature of NC is control of the relative position of a tool with respect to the object (workpart) being processed. Computations must be made to determine the trajectory that must be followed by the cutting tool to shape the part geometry. Hence. N(' requires the controller 10 execute not only seLJuence control but geometric calculations as well. Because of its importance in manufacturing automation and industrial control, we devote Chapter 6 to the topic of NC

Closely related to NC is industrial robotics, in which the joints of the manipulator (robot arm) are controlled to move the endofarm through a sequence of positions during the work cycle. As in NC the controller must perform calculations during the work cycle to implement motion interpolation. feedback control, and other functionsIn addition, a robotic work cell usually iuciudc.s. other equipment besides the robot, and the activities of the other equipment in the work cell must be coordinated with those of the robotThis coordination is achieved using interlocks. We discuss industrial robotics in Chapter 7

       Programmable  logic Controllers

Programmable logic controllers (FLCs) were introduced around 1970 as an improvement on the electromechanical relay controllers used at the time to implement discrete control in the discrete manufacturing industries. The evolution of PLCs has been facilitated by advances in computer technology, and presentday PLCs are capable of much more than the 1970sera controllers. We can define a modern programmable logic controller as a microprocessorbased controller that uses stored instructions in programmahle memory to implement logic, sequencing, timing, counting, and arithmetic control functions for controlling machines and processes. Today's PLCs are used for both continuous control and discrete control applications in both the process industries and discrete manufacturing. We cover PLCs and the kinds of control they are used to implement in Chapter R

       Supervisory Control

The term supervisory control is usually associated with the process industries. but the concept applies equally well to discrete manufacturing automation. where it corresponds 10 the cell or system level. Thus, supervisory control coincides closely with coordination control in the ANSIIISA·S88 Standard (Section 4.3.3). Supervisory control represents a higher level of control than the preceding forms of process control that we have surveyed in this section (i.e., DOC, ~C,and PLCs).ln general. these other types of control systems are interfaced directly to the process. By contrast. supervisory control is often superimposed on these processlevel control systems and directs their operations, The relationship between supervisory control and the processlevel control techniques is illustrated in Figure 4.11

In the context of the process industries,supervisory control denotes a control system that manages the activities of a number of integrated unit operations to achieve certain economic objectives for the process. In some applications, supervisory control is not much more than regulatory control or feedforward control. In other applications, the supervisory control system is designed to implement optimal or adaptive control. It seeks to optimize some welldefined objective function. which is usually based on economic criteria such as yield. production rate, cost, quality, or other objectives that pertain to process performance.

       Distributed Control Systems and Personal Computers

Development of the microprocessor has had a signifil:ant impact on the design of' control sysrcrns.tn this section. we consider [wo related aspects of this impact: (I) distributed control systems and (2) the use of personal computers in control systems. Before discussing these topics. let us provide a brtef background uf the microprocessor and iLl uses.

Microprocessors.      A microprucrssor   is an integrated  circuit  chip containing  the digital logic elements  needed  to perform  arithmetic  calculations,  execute  instructions  stored in memory.  and carry out other  data processing  tasks. The digital  logic clements  and their interconnections   in the circuit form  a builtin  set of instructions  that determines   the runemicroprocessor.   A vcrv common  function  is to serve  as the central  processing unit            microcomputer,   By definition.  a microcomputer   is simply  a small digital

computer whose CPU is a microprocessor and which performs the haxic functions of a computer. These basic functions consist of data manipulation and computation. carried out according to software stored in memory to accomplish user applications, The most familiar and widely used example of a microcomputer is the persona! computer (PC), usually programmed with software for business and personal applications.

Microprocessors arc also widely used as controllers in industrial control systems, An important distinction between a PC and a controlleris that the controller must be capable of interacting with the process being controlled. as discussed in Section 4.3.1, It must be able to accept data from sensors connected to the process, and it mUSI be able to send command signals to actuators attached to the process. These transactions arc made possible by providing the controller with an extensive input/output (lIO) capability and by designing its microprocessor so that it can make use of this 110 capability. The number and typc of 110 ports arc important specifications of a microprocessorbased controller. By type of I/O ports. we arc referring to whether the type of data and signals communicated between the controller and the process arc continuous Or discrete. We discuss liD techniques in Chapter 5, In contrast. pes are usually specified on the basis of rncrnorv size and execution speed. and the microprocessors used in them are designed with this in mind.

Distributed Control Systems. With the development of the microprocessor, it became feasible to connect multiple microcomputers together to share and distribute the process control workload, The term distributed control system (DeS) is used 10 describe such a configuration. which consists of the following components and features [13]:

   Multiple process control suuions located throughout the plant to control the individualloops and devices of the process.

   A central control room equipped with operator stations, where supervisory control of the plant is accomplished.

   Local operator stations distributed throughout the plant.This provides the DeS with redundancy. If a control failure occurs in the central control room. the local operator stations take over the central control functions. If a local operator station fails, the other local operator stations assume the functions of the failed station.

   All process and operator stations interact with each other by means of a communications network, or data highwav, as it is often called.

These component, arc illustrated in a typical configuration of a distributed process control system presented in Figure 4.12. There arc a number of benefits and advantages of the nes~ (1) A DeS can be installed for a given application in a very basic configuration, then enhanced and expanded as needed in the future; (2) since the system consists of multiple computers. this facilitates parallel multitasking; (3) because of its multiple computers, a DeS has builtin redundancy; (4) control cabling is reduced compared with a central computer control configuration; and (S"fnetworking provides process information throughout the enterprise for moreefficient plant and process management.

Development of DeS, started around 1970. One of the first commercial systems was Honeywell's TDC 2000, introduced in 1975 [2]. The first DeS applications were in the process industries. In the discrete manufacturing industries, programmable logic controllers were introduced about the same time. The concept of distributed control applies equally well to PLCs; that is, multiple PLCs located throughout a factory to control individual

pieces ~f equipment hut integrated by means of a common communicati?ns netw?rk. TntroduclJon of the PC shortlv after the Des and PLC, and LIS subsequent mcrease In computing power and rcductio~ in cost over the years, have stimulated a significant growth in the adoption of PCbased DCSs for process control applications.

PCt; in PrUl;fjt;S Control, Today, pes dominate the computer world.They have become the standard 1001 by which business is conducted. whether in manufacturing or in the service sector. Thus, it is no surprise that PCs are being used in growing numbers in process control applications. Two basic categories ofPe applications in process control can be distinguished: (1) operator interface and (2) direct control. Whether used as the operator interface or for direct control. PCs arc likely to be networked with other computers to create DCSs

When used as the operator interface, the PC is interfaced to one or more PLCs or other devices (possibly other microcomputers) that directly control the process. Personal computers have been used to perform the operator interface function since the early 1980s. In this function, the computer performs certain monitoring and supervisory control functions, hut it does not directly control the process. Advantages of using a PC as only the operator interface include: (1) The PC' provides a userfriendly interface for the operator;

(2) the PC can be used for all of the conventional computing and data processing functions that PCs traditionally perform: (3) the PLC or other device that is directly controlling the process is isolated from the Pc. so a PC failure will not disrupt control of the process; and (4) the computer can be easily upgraded as PC technology advances and capabilities improve. while the PLC control software and connections with the process can remain in place.

Direct control means that the rc is interfaced directly to the process and controls its operations in real time. The traditional thinking has been that it is too risky to permit the PC to directly control the production operation. If the computer were to fail, the uncontrolled operation might stop working, produce a defective product, or become unsafe.Another factor i~ that conventional PCs, equipped with the usual businessoriented operating system and applications software, are designed for computing and data processing functions, not for process control. They are not intended to be interfaced with an external process in

the manner necessary for realtime process control. Finally, most PCs are designed to be used in an office environment, not in the harsh factory atmosphere.

Recent advances in both PC technology and available software have challenged this traditional thinking. Starting in the early 1990s, PCs have been installed at an accelerating pace for direct control of industrial processes. Several factors can be identified that have enabled this trend:

   widespread  [amtliartty  with pes

   availability  of highperformance   PC's

    trend  toward  open  architecture  philosophy  in control  systems  design

   Microsoft's Windows NTH' (the latest version is Windows 2000^lM) as the operating system of choice.

The peis widely known to the general population in the United States and other industrialized nations, A large and growing number of individuals own them. Many others who do not personally own them use them at work, Userfriendly software for the home and business has certainly contributed \0 the popularity of pes. There is a growing expectation by workers that they be provided with a computer in their workplace, even if that workplace isin the factory,

Highperformance CPUs are available in the rarest pes, and the next generation of pes will be even more powerful. For the last 20 years,it has been observed that processor speed doubles every 12 1&months. This trend. caned Moore's 1_i1w. is expected to continue for at least another 15 years. At the same time. processor costs have decreased by several orders of magnitude, and this trend is expected to continue as welL The projected results are seen in Table 4.5, in which performance is measured in millions of instructions per second (mips), and cost is measured in dollars per mips. In the earlytomid 1990s. PC performance surpassed that of most digital SIgnal processors and other components used in proprietary controllers [16]. New generations of pes are currently being introduced marc rapidly than PLCs are, allowing cycle speeds of PCs to exceed those of the latest Pl.Cs,

Another important factor in the use of PCs for control applications is the availability of control products designed with an open architecture philosophy, in which vendors of control hardware and software agree to comply with published standards that allow their products to be interoperable. This means that components from different vendors can be interconnected in the same control system. The traditional philosophy had been for each

vendor to design proprietary systems. requiring the user to purchase the complete hardware and software package from one supplier. Open architecture allows the user a wider choice of products in the design of a given process contra! system, including the Pes used inthc system.

For process control applications, the PC's operating system must facilitate realtime control and networking. At time of writing, Microsoft's Windows NT™ (now Windows 2000™) is being adopted increasingly as the operating system of choice for control and networking applications. Windows NT provides a multitasking environment with sufficient security, reliability. and fault tolerance for many if not most process control applications.

At the same time. it provides the user friendliness of the desktop PC and most of the power of an engineering workstation. Installed in the factory, a PC equipped with Windows NT can perform multiple functions simultaneously, such as data logging, tread analysis. tool life monitoring, and displaying an animated view of the process as it proceeds, all while reserving a portion of its CPU capacity for direct control of the process.

Not all control engineers agree that Windows NT can be used for critical process control tasks. For applications requiring microsecond response times, such as realtime motion control for machine tools, many control engineers are reluctant to rely on Windows NT. A common solution to this dilemma is 10 install a dedicated coprocessor in the PC The motion servo loops are controlled in real time using the coprocessor motion control card, but the overall operating system is Windows NT.

Regarding (he factory environment issue, this can be addressed by using industrialgrade: PC~. which arc equipped with enclosures designed for the rugged plant environment. Compared with the previously discussed PClPLC configuration,in which the PCis used only a~ the operator interface, there is a cost saving, Irom installing one PC for direct control rather than a PC plus a PLC. A related issue i~ data integration: Setting up a data link between a PC and a PLC is more complex than when the data a re all in nne PC

Enterprise·Wide Integration of Factory Data. The most recent progression in PCbased distributed control is enterprisewide integration uf factory operations data, as depicted IJl figure 4.13. This is a trend that is consistent with modem information management and worker empowerment philosophies. These philosophies assume fewer levels of company management and greater responsibilities for frontline workers in sales, order scheduling. and prudu!.:tion, The networking technologies that allow such integration are available. Windows 2000¹\1 provides a number of builtin and optional features for connecting the industrial control system in the factory to enterprisewide business systems and supporting data exchange between various applications (e.g .. allowing data collected in the nlant to be used in analysis packages. such as Excel spreadsheets), Following arc some of the capabilities that are enabled by making process data available throughout the enterprise

    Sales personnel can provide realistic estimates on delivery dates to customers, based on current shop loading.

    Order trackers are able to provide inquiring customers with current status information on their orders.

    Quality control personnel arc made aware of real or potential quality problems on current orders, based on access to quality performance histories from previous orders.

    Cost accounting  has access to the most  recent  production   cost data.

Production personnel can access part and product design details to clarify ambiguities and do their job more effectively.