Sie sind auf Seite 1von 36


Unit 2: Concepts of Computer Control Introduction, Sequence control, Loop control, Supervisory control, Centralized computer control, Distributed system, Human-computer interface, Benefits of computer control systems 6 Hours

Computer control system fields in, the applications are typically classified as follows: 1. Batch 2. Continuous 3. Laboratory (test) categories are not mutually exclusive particular process may involve activities which fall into more the above categories 1. Batch batch process term used to describe processes in which a sequence of operations are carried out to produce a quantity of a product (i.e., a batch) that sequence is repeated to produce further batches specification of the product or exact composition may be changed between the different runs Ex.: Rolling of sheet steel Ingot is passed through the rolling mill to produce a particular gauge of steel Next ingot may be either of a different composition or, rolled to a different thickness so, require different settings of rolling mill Set-up (Change-over)time: Time taken to prepare the equipment for the next production batch than one of

Its the wasted time time in which no output is being produced Operation time: Time during which the product is being produced Operation time/ Set-up time ratio, is important for batch size determination NC (Numerically Controlled) machine tool set up in a much shorter time and led to the reduction in the size of the batch and considered to be economic 2. Continuous Continuous term used for systems in which, the production is maintained for long periods of time without interruption typically over several months or even years Ex.: Catalytic cracking of oil Crude oil enters at one end Various products fractionates are removed as the process continues Ration of different fractions can be changed without halting the process May produce batches in which, the product compositions can be changed without halting the current production process Problem: During change-over from one specification to another output of the plant to be to of the required product tolerance and scarp output must be lessened So, change of composition to be made as quickly and smoothly, as possible A process can be converted into a continuous process (whole or a part of the process) Ex.: Baking industry in which, the bread dough is produced in batches - through continuous ovens Continuous ovens frequently used to bake here where the loaves are placed on a conveyor - which moves through the oven Tracking of material through the process is a problem in the mixed mode systems (Ex.: Systems in which batches are produced on a continuous basis) and also identification of a particular batch at all times is also needed

3. Laboratory Systems Are of the operator-initiated type in which the computer is used to control complex experimental test Some complex equipment used for routine testing Ex.: 1. Control and analysis of data from a vapor phase chromatograph 2. Testing of an audiometer Audiometer a device used to test hearing It has to produce sound levels at different frequencies It is complex the actual level produced is a function of frequency since, the sensitivity of the human ear varies with frequency It has to be tested against a sound-level meter a test certificate to be produced It is done by using a sound-level meter connected to a computer using the output form the computer to drive the audiometer through its frequency using the output from the computer - to drive the audiometer through the frequency range Results are printed out from the test computer Test certificate will also produced Difficult to categorize the systems into batch, continuous, and laboratory as

Ex.: production of steel using an electric arc furnace involves complex calculations to determine the appropriate mix of scrap, raw materials and alloying additives also, as the melt progresses samples of the steel are taken and analyzed using a spectrometer which, is connected to the computer which analyzes the results and calculates the necessary adjustments to the additives - computer for this purpose can be the computer using the arc furnace itself

Applications classified activities what ever the way will include: 1. Data acquisition 2. Sequence control 3. Loop control (DDC) 4. Supervisory control 5. Data analysis 6. Data storage 7. Human-Computer Interface (HCI) Objectives of using computer to control the process are: 1. Efficiency of operation 2. Ease of operation 3. Safety 4. Improved products 5. Reduction in wastes 6. Reduced environmental impact 7. Reduction in direct labor General Embedded Systems Embedded computers usage: 1. Domestic appliances 2. Hi-fi systems 3. Automobile management systems 4. Intelligent instruments 5. Active control of structures to large flexible manufacturing systems 6. Aircraft control systems, and etc., Activities that are carried out by the computers and the objectives of using a computer are same. But, activities - differ in balance between different activities, time-scales involved, emphasis given to the various objectives

Sequence control
Occur in some part of the most systems Often it predominates in batch systems hence, batch system is used to illustrate it Batch systems widely used in food processing and chemical industries where the following operations involve usually 1. mixing of raw materials 2. carrying out some process, then 3. discharging the product Fig. shows the typical reactor vessel A chemical is produced by the reaction of two other chemicals at a specified temperature Chemicals are mixed together in a sealed vessel ( a reactor) Temperature of the reaction is controlled by feeding hot or cold water through the water jacket which surrounds the vessel Water flow is controlled by adjustment of the valves C and D Flow of materials into and out of the vessel is regulated by using the valves A, B, and E Temperature of the contents of the vessel, and the pressure in the vessel is monitored Procedure for the operation: 1. Open valve A to charge the vessel with chemical 1 2. check the level of the chemical in the vessel (by monitoring the pressure in the vessel); when the correct amount of chemical has been admitted, close valve A 3. Start the stirrer to mix the chemicals together 4. repeat stages 1 and 2 with valve B in order to admit the second chemical

5. Switch on the three-term controller and supply a set point so that the chemical mix is heated up to the required reaction temperature 6. Monitor the reaction temperature, when it reaches the set point, start a timer to time the duration of the reaction 7. When the timer indicates that the reaction is complete, switch off the controller and open valve C to cool down the reactor contents. Switch off the stirrer 8. Monitor the temperature; when the contents have cooled, open valve E to remove the product from the reactor

Based upon the software, all of the above actions and timings - are implemented by the computer. Fig. shows the typical chemical batch process Processes shown above for single vessel case can often only part of a larger process Two reactor vessels R1 and R2 are used alternatively so, that the processes of preparing for the next batch and cleaning up after a batch can be carried out in parallel with the actual production Assume R1 has been filled with the mixture and the catalyst reaction is in progress there will be for R1 loop control of the temperature and pressure, operation of the stirrer, timing of the reaction In parallel with this, in reactor R2 will be cleaned the wash-down sequence and the next batch of the raw materials will be measured and mixed in the mixing tank Meanwhile, the previous batch is thinned down and transferred to the appropriate storage tank - and, if there is to be a change of product or a change n product quality the thin-down tank will be cleaned Once, this is done next batch can be laded into R2 and then, assuming that the reaction in R1 is complete the contents of R1 will be transferred to the thin-down tank and the wash-down procedure for R1 initiated Complex decisions to be made - like when to begin the sequence Sequence initiation may be left to a human operator of the computer Decision to use human or computer supervision is often very difficult to make the aim is to minimize the time during which the reaction vessels are idle since, it is the unproductive time

Computers usage for the process is good but it is not enough flexible that of the human operator

In most batch process in addition to the sequence control some continuous feedback control are used Ex.: control of temperature, pressure, flows, speeds or currents Process control terminology: Continuous feedback control can be termed as the Loop control or modulating control Ex.: For loop and computer through sequence control is as follows: Float glass process Fig. shows the Float glass process

Raw materials sand, powdered glass, fluxes (the frit) mixed together in batches and fed into the furnace It melts rapidly to form the a molten mixture which flows through the furnace As the molten glass moves through the furnace it is refined Process requires accurate control of temperature - in order to maintain quality and to keep fuel costs to a minimum (heating the furnace to a higher temperature than, required waste of energy to lead and to the increased cost)

Molten glass flows out of the furnace forms a ribbon on the float bath (here, also the temperature control is required as the glass ribbon has to cool sufficiently so, that it can pass over rollers without damaging its surface)

Continuous ribbon passes into the lehr where, it is annealed, and where temperature control is again needed Glass ribbon moves from lehr to cut-up stations at a speed which is too great for manual inspection - so, automatic inspection is used here faults being marked by spraying paints onto the ribbon It then passes under the cutters which cut it into sheets of the required size Automatic stackers then, lift the sheets from the production line Whole of this process is controlled by several computers and involves loop, sequence, and supervisory control Sequence control systems vary from large start-up of a large boiler turbine unit 9n a power station where, 20,000 operations and checks may have to be made to the small starting a domestic washing machine Those systems in past, controlled by the relays, discrete logics, or IC logic units nowadays, been controlled by the sequence control systems Ex.: simple presses where the sequence might be locate blank, spray lubricant, lower press, raise press, remove article, and spray lubricant. PLC Programmable Logic Controllers (computers) - used for simple sequence systems with specific programming languages to work for

Loop Control (Direct Digital Control) Fig. shows: DDC in computer is in the feedback loop

Assumption: System involves several control loops all of which are handled within one computer Consequence of computer being in feedback loop is that it forms a critical component in terms of the reliability of the system

Care is needed in the event of failure of loop control plant remains to be in the safe condition usual means of ensuring safety are to : limit the DDC unit to making incremental changes to the actuators on the plant limit the rate of change of the actuator settings

Advantages of DDC over analog controllers 1. Cost: Single digital computer can control a large number of loops Past times, the break-even points were between 50 and 100 loops but, form the introduction of computers a single-loop DDC unit can be cheaper than an analog unit 2. Performance: It offers 1. simpler implementation of a wide range of control algorithms 2. Improved controller accuracy 3. Reduced drift 3. Safety: Modern digital hardware is highly reliable with long MTTF (Mean Time To Failure time interval between to successive failures) Software used in the control systems going to less reliable than the hardware

Development of ICs and microprocessors made less cost, economical operation than that of analog controllers Single-loop controllers used as stand-alone controllers are now based on the use of digital techniques and contain one or more microprocessors chips which are used to implement DDC algorithms

Adoption of improved control algorithms usually slow, even then, analog PID Proportional Integral Derivative controllers are being taken over by the computer control implementations

PID control General form of PID control algorithm: m(t) = Kp[e(t) + 1/Ti e(t)dt + Td de(t)/dt] Where, e(t) = r(t)-c(t) error r(t) reference value/set point c(t) measured variable Kp Overall controller gain Ti Integral action time Td Derivative action time Above, algorithm can be expressed in many other forms

PI (Proportional Integral)/PID (Proportional Integral Derivative) algorithms are widely used - since, it is difficult to improve on the control performance that can be obtained by these (except, at considerable expense) For majority of the systems PI control is all that is necessary

1. Error signal Control signal is made proportional to the error signal (difference between the desired value of an output and the actual value of the output) is a reasonable strategy Ratio between the control signal and the error signal is adjusted using Kp (gain, proportionality constant) Choosing the right value of the Kp is a compromise: Kp if high steady-state error e(t) is less but, fast response of the system occurs which, might be of the form oscillatory and hence, might unacceptable in many applications Kp if low steady-state error e(t) is high but, slow response of the system occurs 2. Integral term (Second term): Integral sign integrates the error signals wrt the time so, steady-state error can be reduced to zero Integral action time,(Ti): determines - for a given error value the rate at which the integral term increases Advantage: of incorporating the integral term it compensates for changes that occur in the process being controlled Integral term compensates for changes in process conditions like load, environmental conditions and others hence, reduces the error to zero

3. Derivative term (Third term): Provides the improved performance: for the systems which affects by the sudden disturbance Produces control signal that is, proportional to the rate of change of the error signal It anticipates the error - and hence, acts to reduce the error that occur form the disturbances PID control algorithms are not the only type of the algorithms for DDC available but, many others available but, PID control algorithms gives the well-behaved controlling for almost 90% of all control problems - so, they provide the strong deterrent to the adoption of the new control system design techniques

DDC Applications DDC applications can be use for a single-loop system or for the large system where 100s of loops are present The DDC usage loops might be cascaded also Output or actuation signal of one loop acting as the set point for another loop, signals may be added together (ratio loops) and conditional switches may be use to alter signal connections Ex.: Typical industrial system a boiler control scheme

Its the steam boiler system where, the steam pressure is controlled by regulating the supply of fuel oil to the burner To comply with the pollution regulations a particular mix of air and fuel is required which, is done by the control system Working: 1. Steam pressure control system generates an actuation signal which is fed to an auto/manual bias station If the station is switched to auto actuation signal is transferred If the station is switched to manual signal entered manually (through a keyboard) is transferred 2. Bias station: signal form here is connected to the two units 1. High signal selector transmits higher of 2 i/p signals to o/p o/p signal provides the set point for the air flow controller (two cascade loop) 2. Low signal selector transmits lower of 2 i/p signals to o/p o/p signal provides the set point for the DDC loop controlling the oil flow

3. Ratio unit is installed in the air flow measurement line Signal from the controller which monitors the combustion flames directly (using an optical pyrometer) is added to the air flow signal to provide the input to the air flow controller

DDC implementation techniques Types of DDC implementation techniques 1. Inferential control 2. Feed-forward control 3. Adaptive control i. Pre-programmed adaptive control a. Auxiliary process measurements b. External environment(open loop) ii. iii. Self-tuning adaptive control Model-reference adaptive control

1. Inferential control Fig. shows the inferential control system

Inferential: term applied to control where the variables on which the feedback control is to be based cannot be measured directly but have to be inferred from measurements of some other quantity Proposed by Edwards, 1982 Some of the outputs can be measured and used directly in the feedback control other, outputs required by the controller cannot be measured directly, for them, some other process measurement is made and for this the value of the controlled variable is inferred Ex.: Distillation column control

Fig. shows the schematic of the binary distillation column Four independent variables are controlled here 1. Ha and Hb: liquid level indicators Ha: liquid level in accumulator Hb: liquid level in reboiler 2. Xa and Xb: Compositions in products can be measured directly by spectrographic techniques Xa: compositons in the accumulator (top of the column) Xb: compositons in the reboiler (bottom of the column) It is more usual to measure Temperatures (of the boiling points of the mixtures at the specific positions)at Ya and Yb, and pressure,P in the column Ya: temperature in the accumulator (top of the column) Yb: temperature in the reboiler (bottom of the column) P: pressure in the column

2. Feedforward control Frequently used in the process industries Involves measuring of the disturbances on the system rather than measuring the outputs

Speeds up the response of the system to the disturbances However, Can used only for disturbances that can be measured Can be used only in plants where the effects of the disturbances can be predicted accurately Fig. shows the feedforward control system Ex.: Hot rolling of sheet steel If billet temperature is known as it approaches the first-stage mill 1. initial setting of the roll gap can be calculated accurately 2. estimation of the reduction at each stage of the mill can be made, so, initial gap for subsequent stages can be calculated So, the time taken to get the gauge of the stele within tolerance can be much reduced so, quantity of scrap steel (out of tolerance steel) can be reduced

3. Adaptive control Different types of adaptive control systems: i. Preprogrammed adaptive control (Gain scheduled control) a. Auxiliary process b. External environment ii. iii. Self-tuning adaptive control Model-reference adaptive control


Preprogrammed adaptive control

a. Auxiliary process Fig. shows: the auxiliary process control system Adaptive/adjustment mechanism makes the preset changes on the basis of changes in auxiliary process measurements Ex.: 1. In a reaction vessel a measurement of the level of liquid in the vessel (an indicator of the volume of liquid in the vessel) is used to change the gain of the temperature controller 2. In aircraft controls the measured air speed is used to select controller parameters according to a preset schedule

b. External environment Fig. shows: the external environment Measurements of changes in the external environment are used to select the gain and other controller parameters Ex.: 1. Aircraft auto-stabilizer control parameters may be changed according to the external air pressure 2. To adjust control parameters for building environment control system measurement of external temperature and wind velocity are done and used


Self-tuning adaptive control

Fig. shows: self-tuning adaptive control system Uses identification techniques to achieve continual determination of the parameters - of the process being controlled Changes in the process parameters are then, used to adjust the actual parameters

Alternative form of frequently use for self-tuning control is the use of - commercial PID controllers (usually, called auto-tuning) In operator-initiated intervals or periodically controller injects a small disturbances to the process measures the response response is then, compared to some desired response controller parameters are adjusted to bring the response close to the desired response comparison may be based on a simple measure such as percentage overshoot or some more complex comparators


Model-reference adaptive control

Fig. shows: model-reference adaptive control system It relies on the ability to construct an accurate model of the process, and to measure the disturbances which affect the process

Supervisory Control Computers through control has increased range of activities that can be done for controlling Supervisory control methods involves controlling as well as providing that picture to the in-charge of the controlling Computers are the best devices for supervising method of controlling DDC is in most of the cases simply the computer implementation of the techniques used of the traditional analog controllers Earlier 1. computers were not always very reliable and caution dictated that the plant should still be able to run in the event of a computer failure 2. computers were very expensive and it was not economically viable to use a computer to replace the analog control equipment in current use

computer system that was used to adjust the set points of the existing analog control system in an optimum manner to minimize energy and maximize production could perhaps economically be justified were Fig. shows: the basic idea of the supervisory control C, Controllers: represent individual controllers in the feedback loop Can themselves be the digital computers (or some other form of controller) Operations are supervised by the digital computers Actuators: actuates the controller signal and process input signals to the process/plant Ex.: Evaporation plant Fig. shows the evaporation plant

Two evaporators are connected in parallel Material in solution is fed to each unit Purpose of the plant: to evaporate as much water as possible form the solution Steam is supplied to a heat exchanges linked to the first evaporator Stem for the second evaporator is supplied form the vapors boiled off form the first stage Pressure inside the evaporators must be as high as safety permits to achieve the maximum evaporation Its necessary to achieve a balance between the two evaporators if the first evaporator is driven at its maximum rate it may generate so mc steam that the safety thresholds for the second evaporator are exceeded Supervisory control scheme can b e designed to balance the operation of the two evaporations to obtain the best overall evaporation rate Most applications of supervisory methods are simple and based upon the knowledge of the steady-state characteristics of the plant

Complex control algorithms in some applications gives more profitability Techniques: used typically include optimization based on 1. hill climbing 2. linear programming 3. simulations involving complex non-linear models of plant dynamics and economics Here, the complex algorithms have to be computed in real time in parallel with plant operation Expert systems: specialized systems for providing the supervisory control

Centralized Computer Control 1960s all through: computer control implied single computer usage for the control of the whole plant because Computers were expensive Typical computer-operated process involves the computer in performing man different types of operations and tasks General purpose computer can be programmed to perform all of the required tasks the differing time-scales and security requirements for the various categories of task make the programming job difficult particularly with regard to the testing of software Ex.: Feedback loops in a process may require calculations at intervals measured in seconds Some of the alarm and switching systems may require a response in less than 1 second

Supervisory control calculations may have to be repeated at intervals of several minutes or even hours Production management will want summaries at shift or daily intervals Works management will require weekly or monthly analysis Interrelating all of the above different time-scales can cause serious difficulties these tasks done efficiently by centralized computer control Centralized computer control Contains one central computer in the feedback loop Failure of the computer results in the loss of control of the whole plant Earlier computers - were not reliable they were with the MTTF of several hours only later computers with MTTF several months to years came in - they made the usage of defensive programming not of in the control systems

Defensive programming programs which ensure that the system could continue running in a safe condition, while if the computer is being repaired In mid 1960s digital controllers, with the analog back up came in There were based on the standard analog controllers but allowed a digital control signal from the computer to be passed through the controller to the actuator Analog system tracked the signal and if the computer did not update the controller within a specified (adjustable) interval the unit dropped on to the local analog control It enabled DDC to use, with the confidence that if the computer failed the plant could still be operated Cost of operation is high because two control systems had to be installed Dual computer scheme By 1970 cost of computer hardware reduced use of dual computer scheme been done

Fig. shows: the dual computer scheme

In time of failure of one of the computers the other takes over Change-over - can be manual or automatic Disadvantages of dual computer scheme: 1. many cases in cabling and interface equipment is not usually duplicated and also neither the software in the sense of having independently designed and constructed programs so that the lack of duplication become crucial 2. Automatic failure and change-over equipment when used becomes itself a critical component 3. problems of designing, programming, testing, and maintaining the software are more if any change to be in them means provision for monitoring ready for change-over has to be provided Reduction in the cost of hardware and the development of the microprocessor has made multi-computer systems feasible

It is of 2 types: 1. Hierarchical Tasks are divided according to function Ex.: one computer performs DDC calculations and being subservient to another which performs supervisory control 2. Distributed Many computers perform essentially similar tasks in parallel

Hierarchical Systems Its the most natural development in that it follows the typical company decisionmaking structure shown in the pyramid as in Fig. Each decision element Receives commands from the level above Sends information back to that level on the basis of information received from the element or elements below and from constraints imposed by elements at the same level sends commands to the element(s) below and information to element(s) a the same level It follows a natural division of the production process in terms of the time-response requirements of the different levels

Bottom of the pyramid or hierarchy fast response (ms or s) will present for the simple problems As progressing up the hierarchy the complexity of the calculations increases and also the time allowed for the response

Fig. shows the batch system which is typical form of the hierarchical system Assumption: Single computer are used for manager and supervisor functions and that for each processing unit a single unit control computer is used It has 3 levels 1. Manager level Functions: resource allocation production scheduling production costs (or profit margins) on each product

Operating costs for each process unit and scheduled maintenance plans for each operating unit, and current state of production units

Production schedule it is the list of products to be produced and the quantities and process unit to be used calculation It can be done daily, or at some other interval depending on production times, etc., It is calculated and transferred to the supervisor 2. Supervisor level Assumption: supervisor has a store containing the product recipe ( that is, how to make a particular product) and a store of operation sequences for making the product Supervisor When the unit selected by the production scheduling ready the information on the product (like set points, alarm conditions, tolerances, etc.,) it is loaded into the unit controller as are the particular operations to be carried out during each stage of production receive regular reports of the progress of each process unit resolve any conflicts in the demand for resources such as those that can be where the units share a weigher (then, the supervisor will decide which is to use it at a particular time) or where the process requires energy input and the total availability is limited, either because of boiler capacity, or because penalties are imposed for exceeding a specified kVA rating 3. Unit controller level Its the lowest level Responsible for operating the plant opening and closing valves Controlling temperatures, pressures, speeds, flows, Monitoring alarms, Reporting plant conditions

Distributed Systems Assumptions: 1. each unit is carrying out essentially similar tasks to all the other units 2. in the event of failure or overloading of a particular unit all or some of the work can be transferred to other units Total work is divided up and spread across several computers Conceptually, it is simple and attractive approach Methods of allocation of work for each computer 1. Dynamic method Work to be done to be accessed dynamically Passing on of that work load each computer has to be done

Its not simple and requires high-bandwidth data highways for information movement Faster microprocessors are used for this purpose 2. ad hoc method Its the more practical approach for distributing the computing load No attempt is made to provide the load by dynamic allocation method Prior allocation of the work to each computer is done here Ex.: one computer performing non-plant input and output Another computer performing all DDC calculations Another computer performing data acquisition Yet another computer performing control of the actuators Modern implementations of distributed systems usually, along with hierarchical systems, with going to

Fig. shows: the above said system Tasks of measurement, DDC, operator communications, etc., are distributed among a number of computers which are linked together via a common serial communications highway and are configured in a hierarchical command structure Broad divisions of function: Level 1: all computations and plant interfacing associated with measurement and actuation Provides a measurement and actuation database for the whole system Level 2: all DDC calculations Level 3: all sequence calculations Level 4: operator communications Level 5: supervisory control Level 6: communications with other computer systems Rigid boundaries not necessarily be maintained Ex.: DDC unit may perform some sequencing or ma interface directly to plant

Major advantages of this approach: 1. System capabilities are greatly enhanced by the sharing of taks between processors Burden of computations for a single processor becomes very great if all of the described control features are included One of the main computing loads is that of measurement scanning, filtering, and scaling all can be done easily These less burdens the DDC system - so, number of computers needed for controlling will reduce DDC computer will collect measurements, already processed, via the communications link at a much lower frequency than that at which the measurement computer operates 2. Much more flexible than the use of a single processor Loops, operator stations increasing is very easy Allows standardizations its easier to develop standard units for well-defined single tasks for overall control schemes 3. failure of a unit causes much less disruption since, only a small portion of the overall system will not be working 4. provision for automatic or semi-automatic transfer to a back-up system is much easier 5. much easier to make changes to the system for hardware replacements or software changes Changing large programs is hazardous because of the possibility of unforeseen sideeffects with the use of small modules, such effects are less likely to occur and are more easily detected and corrected 6. Computer units can be widely dispersed by using the serial highway so, it is unnecessary to bring cables carrying transducer signals to a central control rooms