it a Cri + O cascade, Feedforward, and Ratio Control 40.1 INTRODUCTION The classical feedback control has always been susceptible to dynamics of loop elements. Slowness of the process limits its speed, and large disturbances take the controlicd variable away from the desired operating point. Feedback is effec- tive, but a single loop feedback control is not efficient enough to meet the re- quirements of a large scale process having complex parameters. In this chapter, we will discuss some schemes other than the ones employing classical feedback loops. Asingle-loop feedback contro! configuration involves one measurement (out- put) and one manipulated variable in a single loop. There are, however, other control configurations with one manipulated variable and more than one mea- surement, or one measurement and more than one manipulated variable. In such cases, control systems with multiple loops are to be employed. Systems with multiple loops improve the controllability of process. We will cover four such configurations, namely cascade, feedforward, feedforward-feedback, and ratio Control in the following pages. : (or CASCADE CONTROL we have one manipulated variable and more ¢ to consider if direct feedback control a N cascade control configuration, than one measurement. It is an alternativ e sing the primary variable is not satisfactory and a secondary variable measure- “ent is available, Cascade control uses the output of primary controller to ma- -Mpulate the set point of secondary controller. ; _ The basic principle of cascade control is that if the secondary variable re ‘Ponds to the disturbance sooner than the primary variable, then there is a pos ity to capture and nullify the effect of the disturbance before it propagates into primary variable, The concept of cascade control is illustrated in Fig, 10.1.Final control] element Inner set point Taner consoller Inner variable Outer controller measurement ! Outer variable v. L_mesurement Outer set point Fig.10.1. Cascade control system The two measurements are taken from the system and used in their Tespective control loops. In the outer loop, the controller output is the set point of the inner loop. The outer loop is called primary loop and the inner loop is called secondary loop. Thus, if the outer loop dynamic variable changes, the error signal (i, input to the controller) affects a change in set point of the inner loop. Even though the measured value of the inner loop has not changed, the inner loop experiences an error signal and produces a new output by virtue of the set point change. Cascade control generally provides better control of the outer loop vari- able than is accomplished through a single variable system. The primary objec- tive of cascade control is to divide an otherwise difficult control process into two 10.2.1 Features of Cascade Control! The outer controller is otherwise termed as primary or master controller. In the Same way, the inner controller ig otherwise termed as ‘Secondary or slave controller. The main features of cascade control are: © More than one measurement, but one manipulated variable * Decreases variation in Primary variable © Enhances stability characteristics * Insensitive to modelling errorsCascade, Fe ‘Sedforward, and Ratio Control 223 The ondary loop before they affect the pana rbances arising within the sec variable. The secondary controller is mos she of this single proportional poPorional with high gain value. Tun- e ortional mode is simple, and moreover, the off a morcover, the offset associated with propo: th proportional mode can be easil present in primary controller, ly removed by integral action {9 Cascade Control of a Jacketed CSTR a ider the CSTR shown in Fig, 10.2, Tk « . 10.2, The reaction is exothermi s Crorated is removed by the coolant, which f ion is exothermic and the heat | objective is , lows in the jacket around the tank. The control objective is to keep the temperature of : ‘desired val t ature of the reacting mixture, 7, con- stant at a desired value. Possible disturbance i srature, T, and the coolant sturbances to the reactor include the feed temp an ee EU lant temperature, 7. The only manipulated variable is the coolal , F. Let us suppose that the major disturbance is change in coolant temperature, 7. Simple Feedback Control: Consider the simple feedback control of this jack- eted CSTR as shown in Fig.10.2, where reactor temperature, T, is the measured controlled variable and coolant flow rate, /, is the manipulated variable: Reactor temperature, 7, responds faster to changes in feed temperature, T,, than to changes incoolant temperature, 7,. Therefore, the simple feedback control is more effective in compensating for changes in 7;, and less effective in compensating for changes in 7,, since changes in T, are not getting instantly reflected on changes in T- Cascade Control: Figure.10.3 illustrates how cascade control is implemented on jacketed CSTR. The response of simple feedback control to the changes in coolant temperature can be improved by measuring T., and taking control action before its effect has been felt by the reacting mixture. If 7. goes up, the eon) action increases the flow rate of the coolant; and if T, goes down, then the control action decreases the coolant flow rate. Tigi, Reactant in Coolant out Coolant out “Coolant c in ant . Reactant Cool ss : Fig.10.3 Gascaded contrat scheme on ig.10- cans acste Fig.t0.2 Conventional feedback scheme ona CSTRThe two control loops using two different measurements, T and T,, share 4 common manipulated variable, F,. a. The loop that measures. T (controlled vanal of master control loop and uses 3 set poun! b. The loop that measures 7. is the secondary of slave loop. It uses the outpay of the primary controller as its set pomt. le) is the dominant, oF primary, 1 supplied by the operator i 40.2.3’ Cascade Control of Boiler Drum Level often used in industry is shown in matic diagram of a drum boiler which 1s Fig. 10.4(a). Holler drum fevel control is complex because specific v of bubbles is very large. and the bubbles displace water resulting in higher appar. cent level than the level due to water alone. If pressure in steam header suddenly drops because of increased demand for steam by the users. then certain quantity of water will flash into steam bubbles. The drop in pressure also forces the volume of the existing bubbles to expand—further increasing the apparent level, This surge in level resulting from decrease in pressure i led swell. Similarly, increase in steam header pressure due to decreased demand for steam by the users has the opposite effect on apparent level because of collapsing of bubbles, and this phenomenon is called shrink. Swelling and shrinking are the Wo as. pects which make the level control of boilers a challenging task, The scheme shown in Fig. 10.4(b) is popularly known as Three Element Con- trol since three elements, viz. (i) outflow from tank, (1) inflow to tank, and (iii) level in tank, are being controlled. The output of level control and steam flow control are added, and the resultant output acts as the set point for control of inflow to boiler drum in cascade Fuel —— fashion. Air 2] olume (volume mass) © Feedwater Fig, 10.4(a) Schematic of drum boiler Feed water Fig. 10.4(b) Three element control of boiter drum levelCascade, Feedforward, and Ratio Control 225 4024 Analysis of Cascade Control Loop rig. 10.5 depicts the open loop block diagram Tepresentation of the jacketed CSTR system shown in Fig.10.3. For the CSTR example, Process | is the fion in the tank and ils temperature T is the primary variable to be controlled. process Il is the jacket around CSTR and its output 7: (coolant temperature) is ihe secondary variable which affects Process I, and consequently (reactor temperature). The secondary process has an output which we are not interested in controlling but which affects the output we want to control tf Process Il + Process | + (secondary) (primary) Fig.10.5 Open-loop process For the jacketed CSTR, block diagram representation of simple feedback con- tol loop and cascade control loop is shown in Figs.10.6 and 10.7, respectively. Set lay id point ‘tn Controlled 4 Process I) | +, Process 1 |e ‘output a Controller (secondary) (primary) Measuring device Fig.10.6 Conventional feedback loop with primary and secondary processes fe far Go G Ra Rot +, Ghia is Ga me Ga F>[ Ge ip G, Gm Fig.10.7 Cascade controlloop ni and Gyy ~ Measuring device G.) ~ Primary controller Gy, and Gp — Disturbance gains Gq ~ Secondary controller L, and Ly ~ Disturbance inputs G,. — Primary process Ry ~ Set point of primary loop 2 ~ Set point of secondary loop Gy ~ Secondary process ’ C, ~ Primary controlled variable Cy ~ Secondary variable226 Process Control: Principles and Applications From the block diagram, the following transfer function expressions may be derived using Mason’s gain formula as given below. L$, was Akt where Y is the output variable, and X is the input variable. N= total number of forward paths P,= path gain of k" forward path A= 1 ~(sum of loop gains of all individual loops) + (sum of gain products of all possible combinations of two non-touching loops) ~ (sum of gain products of all possible combinations of three non-touching loops) + , Ak =the value of A for that part of the graph not touching the A"" forward path GaaGya Y & Gap (10.1) Ry 1+ GG Gna (2) = Gana (102) Ri Jesscate 1+ G2 Grea + Gy Gyr Ges Gant ( ) 6.6.26 6G, a = GGG Gyr simple eo ° Sn ae a Rages V4 GG 2GaGq ° “ Assuming that major disturbances enter the secondary loop: (<} _ GGG ‘ Pewee 1+ CaO Gq2 + GG. GsGaGm ue (<| = —GnG Gy /conventional 4G.G.0 07 Ptah 1+ GG GG, oe These expressions can be used (0 analyse si ed to analyse the effect of. i adding a secondary loop loo] ig a secondary and speed of response, “tS Of Performance, s uch as steady state err! Example 10.1 The transfer fancy Gy = a * Ssseade system are given as (254 Nase? Sn = Gan? Sn= mol Ga isa Poontoller,Gs=4 . G — Gs+1) (a) Calculate the ultimate value off. yp tt 0:95: Gyn = 0.2 feedback and cascade loop go inty oo ore NetY Controller for which simp!® . 0 oscilla Oe simple feedback and cascade loop Sen ee () Compare the afteCascade, , Feedfonvard, |, and Rati , snd Ratio Control. 227 fa) Finding ultimate value of ., n {i Simple feedback (without casende, 6; Characteristic equation (C.E. . G, Using the transfer functions in ce E, os ba G, we 8+ 147+ I5+1+K,=9 substituting = jay the CE — ~ 80 ~ 40h > Tjo+K, =0 Equating real and imaginary pans, 14K. -@=0 = Ky= 14-1 - 8ju' + Tio=0 we 8 = 0.9354 rad’see = K,, (ultimate) = 2-1 = 8 = 125 4 4 (i) With cascade control CE. = 1+ Ges Gyo Gpa* GG Using the transfer functions in C.E., we have: Ga 4 AK yy 3 2 14+—_+—_—- = 85) + 46s? + 31s +5 + 4K, = (stl) #254 45+) Substituting s = j@, the C.E, becomes, 8 jal — 46a? + 317045 * 4K Equating real and imaginary parts, 544K, — 4607 = 0 and 310-80 =0; => w= 0 mdisec or @= 1.96 radisee 4602-5 _ (46%384)-5 - 42.91 = Ky, (ultimate) = yat which oscillations will start is 42.91. er imate value of Ky - back loop. (b) Finding Offset ascade) d finding steady {9 Simple feedback Cera casefer functions wih s= 02 Substituting the values © State value of error, “4 GG Gu__ = > a8 Geo = TE Ky 1225 ( } GaGa ft 2 J syavetona 0.0816 Te Gs Gp2228 Process Control: Principles and Applications: (ii) With cascade control ey H Ginn Ta oscase 1G 2G Gua + Gyn G26. G2 The offset is less in éascade control. Therefore, we come t0 the conclusion that ca si cl ulti 10.2.5 Thumb Rules for Designing and Tuning Cascade Contro, 1. Let the secondary loop be the input point of the most serious disturbance Secondary loop should reduce the effect of one or more disturbances, Make the secondary loop fast by including only minor lags of complete control system. The secondary loop must be at least 3 times faster than the master loop. 3. Choose a secondary variable which will provide stable performance with narrow proportional band. The correct sequence of operations while tuning the controllers that work in cascade is: 1. Set the primary controller to manual 2. Tune the secondary controller 3. Put the secondary controller in automatic mode 4. Tune the primary controller 10.2.6 Advantages of Cascade Control Cascade control is most advantageous in applications where the secondary closed loop can include the major disturbance, and only the major lag is included in the primary loop. Following are some of the advantages of cascade control. 1. Better control of the primary variable 2. Primary variable is less affected by disturbances 3. Faster recovery from disturbances 4. 5, 6. 1. Increases the natural frequency of the system Reduces the effective magnitude of a time-lag Improves dynamic performance Provides limits on the secondary variable FEEDFORWARD CONTROL. Conventional feedback control loops can never achieve perfect control. It8 difficult forthe conventional loops to keep the process output continuously atthe desired set point value in the presence of load or set point changes. This iS because a feedback controller reacts only after it has detected a deviation in tht value of the output from the desired set point. Unlike feedback systems. # feedforward control configuration measures the disturbance directly and tikes control action to eliminate its impact on the process output. Feedforward £0" trollers have the theoretical potential for perfect control,'aSCade, * Fesdtonwar, €Nd Ratio Control 229 Disturbance [Pisurors fot ial Controlled [Fess j_Sutpnat inipulated PPeces} Comoe ali Fig.10.8 Basic block diagram i Fig.10.: oftediioved ony 1910.9 Basic block diagram of feedback contro! lated signals and sends those to th variable remains unaffected in spite ofa feedforward control system Figs. 10.8 and 10.9, respectively. final control element, Therefore, the control of load changes. The generalised block diagram and a feedback control system are shown in 103.1 Feedforward Control of a Distillation Column Figure 10.10 depicts the use of feedforward contro in distillation column Distl- lation is defined as a process in which a liquid or vapor mixture of two or more reboiler Lo of distillation column ro! Fig. 10.10 Feedforward controlBQ Proosss Contos Panciniss and Ansications | * Condensate L | Feedforward $1 FT. ieaaal Desired Liquid | set point stream Fig. 10.14 Feedfonvard contro! of heat exchanger 10.3.3 Feedforward Control of a Drum Boiler \Gieme 10.12 depicts the use of feedforward control in drum boiler, Het ® i isturbas® are the steam flow from the boiler, which is dictated by varying demand e feed water, The flow rate of feed is ae" principal manipulated variable.Waeeavioy Feeciqnvard, ant Ratio Conuol 291 Siam, Ke l ‘ i pt < Feotwatye FE Foatinnardooatiotat kun bodee AQAA Aleachy Stato anc Dynanig Feectonvare Control Design Sa tRurwanad vamioat gatos he PIAEAKE AMAL, Tho mnadtol EOULE Do a atestaty WHE HWS AL A AQ UTILS Haat HA AME SOOT, We WiLL lisetiss the chstan at AWM Sadler Hse ade sAHHE MEG ANAL AMvatIE MAIL. The OS TH QUAI AL BEAK OR A AT ALO WME bs dlaseuabest fy tho follow tag: HAN] AMAA ah A ae afl. aes ae Xs Fed ve aectiie (hat Otis OE ettatie (ue PLS A then dee = A and we NONE TH HAAAL Beate The VATE HSMUERARAH, 2) 8 the MAAK TARRLAMIES aaNet SAAMI AER SUE AMR YAHIE NERY SWAT, GL: te MERIDA AL ataatto, The eure SIENA EWU Rp Rte La ttit Got Eat, EAE MO MWoottERd Soe POLAT Valine fo. Seaceatate Feactianvard Contrattor he AY MAT At Gata ail eartit tor eine Be aovelapyal Eve eagentor tho: WNL fini Uhvaak aban Ov Ht anya 222 TRU aaripanbatest vatantato, OY Wi L etait apy With) 1 Eg. CULE geen sest AL FOES AYE ATEEAV AIAN VATA O° Ke h & WWo \ AN ave ceagnnsedgaane ba Use Abeta CAMEL AAG AHA) nae Lapel cand eames NSH AL sav ernie tty CE atinnlst y tate: AE ARE PVSNAIEY AE NMA OH NE HOC Nanas Ey nivel Stains ARN TAUAHAAE CATANTTRE UATE ALWAYS aeteye hy we NWT Aft atthe poor dayyaanniten vA Ade Byte AAW WAL aE Be ae, a gontenal st Wavntoinayd dvsqurans Pagieiye WOT stunts Hane stay: Wetland eqantiat ‘Wteayweayinst tataat sa abyasty Shy Wht Ms232 Process Control: Principles and Applications e Heedfon wind controller | \ | “>. nF, i (' nya dine | Q C <= Controller Fig.10.13 Static feodforward control of stirred tank heater Dynamic Feedforward Controller To improve the quality of control during the transient response, a feedforvag controller is designed using the dynamic heat balance. ‘ney La rage | Edt Fp, (103) 4h = liquid volume of the tank, Expressing 1 variables and taking Laplace transform: (10.17) in deviation (108) ve . es + where 7= — = retention time of liquid in tank, Since the feedforward conta i ler should keep the temperature at the set point despite disturbances, 7(s) = Toy(s). From Eq, (10.8), expression for he: ls) = FpC,lw + Ny — Fs) (108) This is the design equation for the dynamic feedforward controller. As eat © seen from the equations, the only difference between the steady state and‘ namic feedforward controllers for the tank heater is that the set point is ml plied by the transfer function (ts+1), Therefore, it is expected that for ! (disturbance) changes, the two controllers will be equivalent, On the cont dynamic feedforward control will be better for set point changes. Figure shows how dynamic feedforward control aie model. input is given by: implemented based on the dy" |Case, ade, Feedtonvard, and Ratio Control 233 Feedforward controller ~ a Fig.10.14 Dynamic feedforward control ofthe stired tank heater The response to step changes in load and set Point are plotted in Figs. 10.15 and 10.16, respectively. The response of both static and dynamic feedforward con- toller is same for load changes as coefficient of disturbance [(s) is same in both Eqs. (10.6) and (10.9). The response of dynamic feedforward controller to set poit change is faster than static feedforward controller because of the term (25+1) Kp. The better a model represents the behaviour of a process, the better the resulting feedforward controller will be. 7 FF s Response to load Dynamic change (static = dynamic FF) 1=0see 1 1 = Osec t 7.10.15 Response of static and dynamic Fig. 10.16 Response of static and dynamic : feedforward controller to set ller to load ao controller ec e 103.5 Generalized Design Procedure for Feedforward Control A teneralised design procedure for feedforward control of an peels pro- °€S5 Whose block diagram is shown in Fig. 10.17(a) is discussed in this section, Process equation in terms of deviation variables is given by: 7 (10.10) KS) = G(s)ii(s) + Gy9)As)tions 234 Procosa Control: Principles and AppHCA Disturbance/ioad Mamuputatedd ‘eedforward] controller GAs) ins) Fig.10.17(b) Feedforward control loop with unity transfer function for measurng =< final control element, Let Fp(s) be the desired set point for the process output. Then, th equation changes to: Fons) = G, (s)Fi(s) + Gy(s)dls) (att Interchanging terms in the above equation, we have: _ ' . : A Fils) = [é woo fue «10l- Md G,(s) The above equation determines the form that the feedforward control 395" should have, which is represented in Fig.10.17(b). It also determines t8¢ ™ transfer functions G(s) and G,(s), which completes the design of the feedio™™* control mechanism,Cascade, Feedforward, and Ratio Control 235 Gils) 2 GAS) = transfer functi Gy * GOs) TMunction of the main feedforward controller. (10.13) Gl) Gos = transfer function of the set Point clement, (10.14) From the design equations, it is clear gifferent from the conventional feedback ¢ should be viewed as a special purpose why feedforward controller is sometimes thermore, it is clear that the effectiven thorough knowledge of the process mod drawback of feedforward control. that a feedforward controller is quite ontrollers (P, P/, and PID). Instead, it ‘omputing machine. This is the reason termed as feedforward computer. Fur ess of feedforward control requires a lel (G,, G,), which seems to be the main ‘A more general feedforward control system which includes a final control element and a sensor that measures the disturbance is shown in Fig, 10.17(c), Feedforward control mechanism Set point : clement Gen = Sm sr = Ge Measuring device Disturbance/load ‘eedforward| controller Process output Hs) Final control element Fig.10.17(c) Feedforward control loop with measuring device and final control element From Fig. 10.17(c), we can derive the following equation:, P= (GG, Gey Gop sp + (Gy ~ GpGy Gey Gul (10.15) From this generalised model, it is straightforward to determine the conditions “hich would totally eliminate the disturbance. We have to get rid of the term Containing din Eq. (10.15) to eliminate the disturbance, which implies that the “efficient of @ should be zero. G, Therefore, Gy GG; Gy Gu = 9 => (10.16) I” 66,6, Similarly, for set point tracking, the coefficient of ysp in Eq. (10.14) should °:'. Using the above relation for Gy we obtain 1 8 ee (10.17 Cop Gy GG, =1 => G0 SEG ) bb238 Process Control: Principles and Applications: Substituting the value of G,, from Bq, (10.16) in Eq. 10.17), (01, 10.3.6 Practical Aspects on the Design of Feedforward Controllers: ‘The design equations discussed catlicr indicate that the feedforward controlfe 4 speeial purpose computational machine, Its practical implementation 15 raj. easy if we use a digital controtler instead of an analog controller which is gui expensive and difficult to implement, In this section, we examine some simpy, cations that lead to a comparatively casy implementation of the feedforwary concept Assume G,, = G,= 1. Now, cach of the process transfer functions ~ G, (5) ang GAs)—has two elements, viz. © The static clement which corresponds to the static gain, and * The purely dynamic element which is a function of s. Therefore, G,(s) and G,{s) can be represented as: j GJs)=K, Gis) and Gs) = Ky Gs) For instance, in case of CSTR, we can easily identify the static and dynamic Paris of the process transfer functions from the following equation Gs) = +1. which indicates that Fpe, 341 1 1 = and G4(s) = — Fp, #) Bt 1 GAs) = Sas The steady state feedforward controller is the ment, At steady state, we retain only functions 7 indicates that K, = 1 and G(s) = Bel simplest and the easiest to imple the static elements of the process transtét and Ggy= Ky G.j and Gp are simple constants anc way as a proportional controller, which hi elements given by the above Lead-Lag Element Instead of using the exact transfer functions of G i te , bi # GAs) and GAs), it is possible approximate them by first-order lags and still obtain Very ae results, Insucd? case, l+ Gfsy= A Le nys Ling 69 d can be constructed easily in the ts only proportional gain, So, the des" equations are called gain-only elements. o ) nv Veav 1 tus” Xs) w 4238 Process Control: Principles and Appfications Substituting the value of G., from Eq, (10.16) in Eq. (10.17), G, Gp * Gj (1015 J 10.3.6 Practical Aspects on the Design of Feedforwarg Controllers ‘The design equations discussed earlier indicate that the Feedtorward conto is 8 special purpose computational machine, Its practical implementation ig hey casy if we use a digital controller instead of an analog controller which is que expensive and difficult to implement. In this section, We examine some simpli. Cations that lead to a comparatively casy implementation of the teedfor ay concept. Assume G,, » G,« 1. Now, cach of the process transfer functions G4s)—has two elements, viz * The static clement which corresponds to the static gain, and * The purely dynamic element which is a function of s. Therefore, G,(s) and Gs) can be represented as: GAs)" KG.) and Gs) = K, Gifs) For instance, in case of CSTR, we can easil Pants of the process transfer functions from thi rou GI) = ———— ; whict e Epc, 341 Gals) ang y identify the static and dynamic ie following equation bh indicates that i 1 K, = —— and Gs) = 1 Fp, ctl Bel Instead of using the exact transfer fun id yet approximate them by first-order lags and oye, “ol and Gifs), it is possi : : wee $6 and still obtain very good results. Inst! “4B tee, = 2B lets Hs) Cds" es a Ts Xs) aCascade, Foodtorward, and Riatia Contech 237 ww ty ime (min) Fig. 10.18 Step response of lead-lag element oe! (10.20) For step input x(t), y(t) = 1+ (10.21) The unit step response of lead-lag element is shown in Fig. 10.18. As ce all curves approach unity. At t= 0, y = Tg/Tig The controller given by Garis called a lead-lag element because (B=) introduces a phase lead and the term [7 introduces phase lag. The lead- clement is the most commonly used dynamic feedforward controller. It is quite versatile because the two adjustable parameters @ and B allow its use as a lead clement (when B> @) or as a lag element (when a> ). Physical Realizability of Feedforward Controllers Let us consider a process with the following transfer functions: -0 Gs) = eee and G(s) = ne Astle et aol Then, the transfer function of the main feedforward controller is given by: G(s) = Gu - 1,03 Gs) 5 The positive exponential term implies that we need to know the disturbance Values in advance in order to compute the current value of the manipulated variable. But such future values of the disturbance cannot be made available in real time, tyaueh a case, the feedforward controller terised as physically "realisable and cannot be applied in real situations. Le238 Process Control: Principles and Applications Weazs Comparison of Feedforward and Feedback Cont tro} Ted forward and feedback control schemes have their advantages : an j tations, Table 10.1 compares the schemes with respect to various Param and Table 10.2 illustrates advantages and disadvantages of these Configure Ms, Table 10.1 Comparison of feedforward and feedback Contro) Property FFC FBC Design principle. | Balance/conservation | Feedback control ean principle Measured variable Disturbance Controlled variable” Controller system Open loop Closed loop configuration Action| Anticipatory Compensatory Typical controller ‘Mathematical On-off, P, PI, PID (usually) relationship Table 10.2 Advantages and disadvantages of feedforward and feedback configurations FEEDFORWARD CONFIGURATION Advantages Disadvantages 1. Acts before the effect of a disturb- 1. Requires identification of all possible ance has been felt by the system, disturbances and their direct measurement, 2.Cannot cope with unmeasured distur bances, 2. Is good for slow systems or with significant dead time 3. It does not introduce instability in 3. Requires good knowledge of the the closed loop response. Process knowledge so is sensitive to Process parameter variations. FEEDBACK CONFIGURATION Advantages Disadvantages |. Itwaits until the effect of the disturbance has been felt by the syste before control action is taken. 2. Does not require good knowledge |3.1tis unsatisfactory for slow process oF of the process so itis insensitive ignificant dead time. to modelling errors. 1. It does not require identification and measurement of disturbances 3. IL is insensitive to process 3. [may create instability in the closed parameter changes. loop response. Simple guidelines for choosing between ke bel feedforward or feedback cont alone are given below.Cascade, Fe Sedforward, and Ratio Control 239 use only feedforward contro! if: 4 The physical and chemical pro 4 The variables in the equations can be o, There are no significant process disrbarg ae : ances, The accuracy of the me; : ¥ 9 ‘surements used is an order of than the application specifications, eet Use only feedback control if: erties S of the process are well known. « Significant un-modelled process disturbances exist. ‘annot measure it ; Can Fatih or calculate variables in the equations describing the physical and chemical properties of the application, Ifthe sialon oes not come under either of the above, it is advisable to use feedforward-feedback control configuration which supplement and complement cach other. 10.4 FEEDFORWARD-FEEDBACK CONTROL CONFIGURATION Clearly, if one can measure up-stream disturbances, then one can take anticipa- action that pre-empts the disturbance affecting the process. A word of caution is that feedforward control depends on the use of open-loop inverses, and hence, it is susceptible to the impact of model errors. Thus, one can usually supplement feedforward control by some form of feedback control, so as to correct any miscalculation involved in the anticipatory control action inherent in feedforward control. The feedforward control cancels the effect of the mea- sured disturbance. Since feedback acts as the system’s watchdog, the process model need not be exact. The effect of load changes other than the measured disturbance will be corrected by the feedback system. tory control fae To steam _——> users ar os Ge Make-up water {forward control Fig. 10.19(b) Feed To steam users Fig. 10.19{a)_ Feedback contro! water forward-feedback control240 Process Control: Principles and Applications Figure 10.19(a) shows application of feedback control to maintain the a drum boiler by changing inflow to the boiler. Feedback alone must g eis variations in steam usage (load variations) by feedback action only, Figur, 1 ft hy shows how feedforward alone can handle variations in steam usage by ne) ing the steam flow rate and then accordingly changing the inflow 1 tid level. Figure 10.19(c) illustrates application of feedforward plus Feedba” maintaining the level in a drum boiler. It measures both controlled variabjey.® as well as disturbance (steam flow rate) and then manipulates the inflow, “) eee Example 10.2 For the heat exchanger shown in Fig. 10.2), draw tgp matic diagram for a combined feedforward and feedback controller in ykig inlet feed temperature is the major load variable, and outlet temperature gy. controlled variable. The combined controller output is the set point for the Stean, pressure controller. Solution Steam PT) rea DF Fig. £10.2(a) Schematic of heat exchanger Condensate Steam rr Feed RSP = Remote set point Fig. £10.2(b) Combined feedforward-feedback control of heat exchange Condensate Se seee ashe aee Cascade, Fo, * Foedtoryae, 2M Analysis of | ind Ratio Control 40.4-1 "oetlorwerdFocdbagy Cc ised Confipur ‘ontrol Loop HO OF 4 combined gigure-10.20 depicts the peng feedback control system, us now develop an eqy cedforward- Let tion forth Ne close system. ed 1009 response of th compenite J? Gri + Gd ‘The value of the manipulated Variable jy Biven b en by; a =O +2 )=6 6 4G Gore (10,22) m = G,Ga(5 ~Gid)4GG0 + Gy Ga(Goy Hsp — Gygil) (10,23) Using the above equations, a 5 = DGG +646) 1+ GG/GAG,, From Eq. (10.24), we can obs response under feedforward-feedback control, 1, The stability of the closed-loo characteri Gy ~GG,G..G, aba 5 4G,G,0,0-4 (10.24) ve the followin 18 about the closed-loop process tesponse is determined by the roots of the tic equation: 1+G,GG4G,,=0 ipGGGyy This depends on the transfer functions of the feedback loop only. There- fore, the stability characteristics ofa feedback system will not change with the addition of feedforward loop, Sensor measuring disturbance for ' ‘. feedforward controller . sp a : Gsp(s) Gat) Set As) point 25) Gals] Feedforward Gu) < controller eats) . Process ip. + Tsp 4 Bs) =|) 1G output Ge) i) Gals) +-(0)— i) * Ta) OA) oO Feedback controller Gin (5) ‘Sensor measuring the process output for feedback control back contro! 1-feed! Fig. 10.20 Generalized block diagram for feedforward:-teoN 2. The transfer functions of the feedforward loop, G.> and Gp witt be . 2. The fe oo 242 Process Control: Principles and Applicatio ey ” G, Ca" Seer mt, SiG, (0034 re G Gand Gyg are known exactly, the perciale ‘ard loop Compe, sates completely for disturbance or set point changes and ihe feat controller remains idle. 3. Iany of the four parameters. GG,» G; OF Gy is known only weg mately, then Gy~G,GjG2Guz#0 and/or G,G,G.sGgp #1 se food! vi erfeet ¢ In such a case, the feedforward loop does not provide perfect contn Ol. They € #0 and the feedback loop is activated and offers the nece Compensation So ey __ aa _—— Example 10.3 inflow to a tank having area 4 is m, (manipulated variable) ang outflow gg is through a resistance R. Load is downstream head 1, Design g feedforward-feedback control configuration for this system, The ‘Measuring ole, ments have first-order transfer functions with character ING parameters ag yy final contro! element has unity transfer function. A P-controller having in place to provide feedback. Draw the block diagram, and express in terms of set point and disturbance, Solution Mass conservation equation around tank is: dh m~g)= AS a h-u dh > m—{A=") = dh R dt Taking Laplace tansfoy nipulated (nm) and disturbar nity, Sain, K, is height in tank ™m, controlled variable j S expressed in terms of ma- ince Variable (12) as: -_® 1 Mts) = 75") * 63), ls) is disturbance = d(s), T= AR Block diagram of feedforward-feedbac! Fig. E10.3 Block diagram ot eedtonvary ecoacconro control contCnocndo, 6, Foedtorward, ane Hawi Gonteot 243 RATIO CONTROL peonttol is used fo ensure that two orn OF Mote flaws are kept at fi Je kept at a constant 1160 ward ¢ Hows are changing. Ratio control iv a H f Af sw special type of feedfore eon th eto where two disturbances are measured and held at t th each other, Sometines, the control of one of the streat ‘i ae si I i gol NET flow fs tenstre streams becomes difficult. im hich case. US How is meastred: and then, the How rate of the ot! Ki ¥ Bee ieecantwhiieel oe low rate af the other stream 18 gonrotled: The stream w hose flow rite is uncontrollable is called wild street goine of te COMIN applications of ratio control ate listed below ol ate listed belov « To keep a cons ant ratio between the feed flow-rate and the steany in the reboiler of # distillation column, fy hold the reflux ratio constant in a distillation column Jo control the ratio of two reactants entering a reactor at a desired ratio « To hold the ratio of two blended streams constant, in order to maintain the composition of the blend at a desired value. To maintain correct air and fuel mixture for optimal combustion. 40.5.1 Ratio Control Configurations jin which ratio control may be implemented. Some illustra- ‘There are many Way! Jn ratio-control is employed are tive examples showing the cont described below: igurations in whicl flow rates, and take their int, as shown In the d to generate the Configuration 1 In this control confi ratio, This ratio is compare Fig, 10.21), and the 1g signal for the ratio cont we measure both the red ratio (the set por two is use jguration, .d with the desi deviation (crror) between the actuatin} roller. Divider Measured ratio Controllable strea™ figuration ") Fig.10.21 Ratio contro! (Com! teON 244 Process Controt: Principles and Applications 4 Wild Configuration 2 & to this configuration. 9s ay 7 ofthe wi steam. 3. pray (Daiedraie] forthe controllable tow serge ‘fe &. A Fe ition 3 Configura’ O-O——- © A This configuration is plicable yy B oe both the streams are controtagye is implemented with a double cedback jy” Fig40.22, Ratiocontrol (Configuration2) a5 shown in Fig. 10.23. Bach fog? maintained by individual feedback loop. Here, the error si s gencrated with a given set point signal SP: whereas for s ; Othe set ping kenerated using the signal from 4 multiplied by the desired ratio. ignal for sty Fig. 10.23 Ratio control (Contiguration 3) 10.5.2 End-point Control Acidic flowCascade, » Feedior . rw wet point of flow controller of basi fard, and Ratio Control 245 fi pHT are flow transmitter and eae tr fa or. ran: controle smitter respectively; F 4053 Total Quantity Ratio Contro! "ype of ratio control, total s TYPE | , totalisers antities from bot wee Place it . h the transmitters are ne el lan so that the asured, and then ratios of yantity are calculated and fl and flows are WS are appropriately controlled using a con- al pamsyement. In all the ratio Mer Masmitters fOr t Ho control configuratio s otal quantity ratio control. totalsers may be placed y rol. 1S set accorc iccording to ratio controller. the flow jah total afer 3 40.5.4 Relation Control inthis OPE of control, relation between two variabl mer opera Race arene a i wailed to ensure ressure and flow, temperature and concentration, an ei pee) Bre. For example, in a compressor, pressure and flow Ra a eae can no longer be maintained due to an increase in pressure ee ome car This results in 2 momentary Jone how reversal which will put inter- der stress and lead to nal parts un failure of compressor if control ac- ion is not taken. 'A surge controller typically mea- ction of pressure rise ver oller operates & ti sures a fun’ sus flow. The contr surge valve to maintain sufficient for- ward flow via recycle to prevent hows the sche- Fig. 10.25 Su ol of compressor using rrge contr surge. Figure 10.25 s| matic diagram of surge control us- relation control ing relation control. din Fig. E104 Example 10.4 Explain the control strategy246 Process Control: Principles and Applications Solution ‘ epel The wild flow (FT-101) is not controlled by this loop. Depending upon, utp FT-102 and the ratio required between wild and controlled flow, FE-1g3 deat the remote set point for controller (FIC-102) for the wild flows . Example 10.5 Explain the control strategy used in Fig. E10.5, Remote ratio adjustment Wild flow’ property Mixed flow Controlled flow Fig. E10.5. Solution The mixed flow is given to Analyser indicator-controller ( AIC-103). The output of AIC-103 and FT-101 put on wild flow decide the output of flow ratio (FF) which acts as remote set point for the flow controller (FIC-102) placed on con- trolled flow. This is a three level cascade controller. The controlled flow controller (FIC- 102) is the secondary controller. FF-102 is primary to FIC-102 as it provides set Point to FIC-102, but is secondary to AIC-103 accepting output of AIC-103 3s its setpoint, The top level primary controller is AIC-103, Example 10.6 Figure E10.6 shows the diagram of a ratio controller used for Wastewater neutralisation. Explain its control strategy. t{_| wastewater NaOH solution EMluent Fig. E106Cas cado, Feedforvard, and Patio Control 247 oil” cneme shown in Fig, E10 yn the Soren asured aci 6, the product of the aad he Fhe measure acidic wastewater lo ate roa cooler ou rovide the set point of! controller (FC) pla (sr) Pe oH biel ig a on basic solution (NaOH),which controls how ae oF NS n NaOH flowrate (FF2) to neutralize wastewater sm RECAPITULATION == = contol configurations with one manipulate sual systems with multe Toops. anipulated variable, Form 4, Salient Features of cascade control system are: More than one measurezoert but Salietanipulated variable te output of master eontlle or es ae sot point for one Meontroller; two feedback loops ae nested together. res 2 oy woeful siavjnating the effect of disturbances that move throush ere very slowly: Geereases variation in primary variable; enhances stability ‘heracterisics- Feedforward control measures disturbances and takes corrective action t Fessmise deviation of controled variable sransforvard contol acts before the effet ofa disturbance tas bese FI 6 tbe system. It is good for slow systems and does not introduce instability in the Syste op response bul i equres identification ofall possible dis ances close direct measurement, cannot cape with unmeasured diswefanceS and ane ansitive to process parameter variation. « esSpack control does not requte good knowledge of the Proest Goes not reer identification and measurement of disturbances insensitive to mod- reat grors and parameter changes, Dut it waits unt th effect of the distur caine as been felt by the system. Feedback contel tnsatisfactory for stow process and may create instability n 1 plement and complement each « Feedforward plus feedback configuration sv Saher, giving superior performance. Addition fof feedforward does not affect the stability characteristics. « Ratio control ensures that two or more flows sf Kept at a constat ig a special type of feedforward contol ‘here two disturbances are meas ‘and held at constant ratio. There are many control configurations. like end-point Control, total quantity control ete. Relation control maintains the ratio between any two variables. the closed-loop response. at. Ratio control Review Questions 1. Show that the stability of tion of feedforward mode. 2. “Ratio control is a type of feedforward control.” 3. List the advantages and disadvantage configuration. 4. Expl the phenomena of ‘Shei and “swell” in boiler drum | 5. Explain ‘three element control ‘or boiler drum level control feedback configuration Is not disturbed by the addi- the statement of feedforward and feedback control jevel control.248 Process Control; Principles and Applications seade control configuration? 6 What are the advantages of” 7. Uyplain end point contro!” configuration oF ratio control with the hy example, ID of 8. Explain “relation control? with the help of an example, Numerical Problems 1.1, Drow theme for jacketed reactor are is the primary. var le control tor tempers secondary variable 10.2. The temperature of @ furnace is to be controlled. The rate of flow furnace is the manipulated variable, Pressure of the fuel is the able, Dranv a easeade control scheme for this system 10.3, The block diagram of a process is shown below. which hot oil is 5 iable and jacket tem, Phlied, Peat OF fuel ty dary yan y nt Gn Fig. E10.7 Design the control scheme for this process using: (a) Feedforward control alone; assume that onl (b) Feedforward plus feedback control. Sey ly disturbance xy is measurable