CHAPTER DYNAMIC BEHAVIOR OF TYPICAL PROCESS SYSTEMS 5.1 INTRODUCTION Examples in the previous two chapters have demonstrated that physical systems, which involve very different physical principles, can have similar dynamic behavior. ‘The concept that a single model type can apply to a wide range of entities, process plants, biological units, economic communities, and so forth provides the basis for “systems” analysis. Thus, it is possible to acquire understanding of a large number of systems from a thorough study of a much smaller number of basic models. Ia this chapter we study some fundamental model structures that occur frequently in process plants, along with their effects on dynamic behavior. This experience Wil tenable us to recognize the effects of process designs on dynamic behavior: First, the behavior of some simple, basic systems, such as first- and second- order and dead-time systems, is summarized using the results from previous chap- ters, with some extensions. Second, the behavior of these simple systems in series structures is determined. Third, the behavior of parallel structures of simple 35 tems is derived, Fourth, the effects of recycle structures on dynamic responses are determined. The chapter concludes with an investigation of more complex physical systems of special importance in the process industries: staged systems and multiple input-multiple output systems. 148. DYNAMIC BEHAVIOR OF TYPICAL PROCESS sysTEMS 149 In these sections, the manner in which the behavior of simple systems is altered by common process structures is derived for simple, idealized models but is demon- ‘strated for important process examples involving levels, heat exchangers, chemical reactors, and distillation towers. This coverage demonstrates that the engineer must master both the physical principles of specific processes and systems analysis tech- 4, niques to determine the dynamics of complex processes quantitatively. a 5.2 BASIC SYSTEM ELEMENTS ‘The coverage of process dynamics begins with the simplest elements, which are ‘often combined to model more complex systems. Since examples of most of these {elements were included in previous chapters, the coverage here is concise. The basic structure for each element is first defined, and several physical examples are ith the system input designated by X and the output by Y. The chemical pro- principles should be apparent to the reader, while the electrical and mechanical Is are based on Kirchhoff's and Newton’s laws, and the reader is referred to (1992) and Weber (1973) for derivations. The graphical and analytical results, common inputs for several basic systems is summarized in Figure 5,1; the presen- n of results in such a figure seems to have originated with Buckley (1964). Only amplitude ratio is presented here, because more extensive frequency response lysis is presented in Chapter 10, where the importance of the phase behavior on lity is demonstrated and applied in control system analysis. Order System {order systems occur as the result of a material or energy balance on a lumped well-mixed) system, as demonstrated in Examples 3.1 and 3,6, Some further les are given in Figure 5.2. The differential equation and transfer function for border system are YO) Ke Xs) 547 ‘The step response is monotonic, with its maximum slope at the time of the , and the time to reach 63.2% of its final change is one time constant. The final ly-state change is equal to K (AX). Y'@ = KpAXV - e-*) 62) ‘An impulse input occurs over a negligible time and transfers a finite amount the system. For example, rapidly introducing a small amount of tracer into ied tank emulates a perfect impulse. The impulse response shows an im- te increase at the time of the impulse, which for the idealized stirred-tank le would mean that the concentration would change instantly by (Mass of (Volume). After the impulse (C), the system follows an exponential path in toiits final condition, KyXQ Gis) = 6) vo = fem 6.3); bance eee 7 i 7 - oa 24k ee componcat £ He | || Fem eeye od \afe ie eto ] TT Eneey % r 10 ¥ 5 4 FA. ove roo. A, é a TL erat oan z Z | | go BF w re rq . i ( ; = postion, k ~ spring constant, and f= fictions For the first-order system, the amplitude ratio is never greater than the process in K,, and it decreases monotonically as the frequency increases _ [PGa) K, IxGo)] ~ JT + ote? Mixing AR = |6Uio) 64) \d-Order System second-order system occurs when two first-order or one second-order ordinary ntial equation is required to model the dynamic behavior. Some examples are =n in Figure 5.3. The transfer function for the second-order system with again in ‘numerator (and no zeros) can be written as ' i b z 4 é 5 a (ora day 2 aio + age f¥O + YW = KX) co 3 g : : e oe 2 3 ay = YO 2 Ke & i a E XQ) Pt 42ers +1 Je aa Parameter £ is termed the damping coefficient, and ary are the two roots of characteristic polynomial, which determine the exponents ofthe time-domain 150152 process ovwamucs bance tt Out Ky ae Cort ’ Component Cyn vi tae tatty aoe satel rem ye Srey he tee quesion 32 (is, ra a 1 Fa sues [asiaz*s Be meat Lk AT, om ew a rage Second-order processes(E = voltage, ~ postion, k= spring constant, f= fiction coeficient. k = force, m = mass, 74 = VAF + Vi), and m = V/F) ‘output function. When the damping coefficients less than 1.0, the system is termed tunderdamped, the roots of the characteristic polynomial are complex, and the system will have periodic behavior for a nonperiodic input. For example, the nonisothermal reactor system in Example 3.10, which exhibits oscillations for a step input, has & damping coefficient of 0.15. When the damping coefficient is greater than 1.0, the system is termed overdamped, the roots ofthe characteristic polynomial ae real and the system will have nonperiodic responses to nonperiodic inputs. Finally, the series reactor system in Example 3.3 has a damping coeflicient of 1.0, which indicates real repeated roots; this type of system is termed critically damped. The possibilty of periodic behavior is not affected by & numerator that could be a function of s; thus, the simplest numerator is considered here, Kp. Two entries are given in Figure 5.1 for second-order systems; one is for an overdamped system, and the other is for an underdamped system. The step response for the overdamped system initially at steady state is monotonic with an initial slope of zero and an inflec- tion point. Note that the underdamped system experiences periodie behavior even for this simple input. OVERDAMPED STEP RESPONSE (€ > 1). 66 mana CRITICALLY DAMPED STEP RESPONSE = 1. ye Keax(t -( a! Je“) «67 :RDAMPED STEP RESPONSE (£ <1). ~ Ky et sin PAB E19) =m AMPED IMPULSE RESPONSE (& > 1). reel |CALLY DAMPED IMPULSE RESPONSE (€ = 1). Ct is Sew 6.10) ERDAMPED IMPULSE RESPONSE (& < 1), c ¥ oF sin| wee ou) Both the step and impulse responses for a second-order system have initial ses that are more gradual than for a first-order system. The overdamped sys- n approaches its final value smoothly, while the underdamped system experiences ions. : ‘The amplitude ratio of the frequency respor tonical mse is monotonically decreasing for overdamped system and begins to deviate substantially from K,, around the fre- equal to I/r. The amplitude ratio for second-order mance effect results from the inherent oscillatory tendency of the system rein. ig the input sine oscillations. " aie [¥Ge)| _ Gey] Y AR = |G(e)] = Time e dead time or cee delay was introduced in Example 4.3 for plug flow sand can als oct for tanspration of solid along a onveyor bel aa to have the following transfer function: 7 vor belli YO. om Xo) YW) =Xa-0) Gis) (5.13)154 maocess ywames ‘The step response, impulse response, and amplitude ratio can all be easily deter mined, because the output is the input translated in time by 8. For example, this leads to the conclusion that the amplitude ratio is equal to 1.0 for all frequencies, which can be demonstrated mathematically by AR (0) — jsin(w0)}| = Jeos?(w0) + sinwe) = 1 (5.14) Jem “The dead time can be approximated by a transfer function composed of poly- nomial in the numerator and denominator; the results of this approach are referred to as Padé approximations (Coughanowr, 1991). This assisted design methods used inthe past, Which were limited to transfer functions that are ratios of polynomials in s. The Padé approximations are not widely used now in the analysis of process control systems. The importance of dead time to feedback control can be understood by consi ering an example such as steering an automobile. With dead time, the automobile ‘ould not respond immediately after the change in steering wheel postion, Cleary, such an automobile would be difficult to drive and would require a skilled and patient driver who could wait for the effect ofa steering wheel change o occur Integrator “The integrator isa special type of first-order system; a process example ofan int raor is level system, shih fo modeled based on an overall material balance to five pase = pha phi 615) In many cases the inet and outlet flows do not depend on the level (unlike the tank draining Example 36), When no causal relationship exists fom the level tthe los, the model has the following general form il a x2 40°) my = hldupsine 6 wm (s) THs r : G(s) = 6.17) Go) = F “The important difference between the integrator and the first-order system in equation (51) is the lack of dependence othe derivative onthe output variable (”) that is, dat is independent of ¥’, This results in a pole ats = 0 in the transfer function, The analytical expression forthe output ofthe integrator is yo =| xorar (5.18) A system like this simply accumulates the net input: thus, the name integrator If the deviation in the input remains nonzero and of the same sign, the magnitude of DYNAMIC BEHAVIOR OF TYPICAL PROCESS sySTENS 185 {he idealized model output increases without limit a time increases toward infinity ‘ora step input, For _ 3x, y’ (5.19) ‘The impulse response also demonstrate thatthe system integrates the impulse under the impulse function), and thn the output remains constant ait altered when X"(t) returns to zero. The value of the impulse response is Y' = C/ry. ‘The amplitude ratio canbe determined 1 be oj} 1 , Tue? | ~ Tyo 62) the frequency decreases, the amount accumulated by the integrator each half od (which is related tothe output amplitude increases [-Regulation unique behavior of the integrator demonstrates that not all (idealized) processes, the inherent characteristic of tending to a steady state after an input changes its initial value to another constant value. To make the distinction, the term regulation is introduced here. An integrator system is often referred to as non— gularing, because the rate of change of the output is independent ofthe output. ing that the definition of feedback isthe application of an output to influence ut, its cleat nat there is no inherent feedback in this process. Any non-self- ig process should be controlled, because even a small nonzero input over a time will result in the output process variable deviating far from its proper value. ‘modelling and control of non-self-regulating liquid level systems is covered in 18. By comparison, a flow in the level draining system in Example 3.6 depends the level. Also, the accumulation terms of all first- and second-order systems sidered to this point depend on the value of the system output. ‘These systems self-regulatory and tend to approach a steady state after a step input (if of the characteristic polynomial have negative real parts) Self-regulating are generally easier to operate, because they tend toward a steady state, ugh the output variable can deviate from its acceptable range of values due to input disturbances. A stable self-regulatory process could be said to have inherent negative feed- For example, the heat exchanger in Example 3.7 has inherent negative feed- because an increase in the output (outlet temperature) causes a decrease in a 1 input term —(F/V + UA/V pC,)T, which stabilizes the system by decreasing vate Fp, UA, \_[F, UA) Tot - (FY); F (pr eb tae)- (B= 20 External inputs Inherent negative feedbacl156 process ovNamics Some processes have inherent positive and negative feedback, for example, the nonisothermal chemical reactor with exothermic chemical reaction in Example 3.10. ar _ [t_, UA iF UA ¢- (Fm + Test) ~ (+ va) 622) Extemmal inputs Inherent negative feedback f ant +f AA) koe” Ca Vey Inherent positive Feadtock ‘The reactor has a negative feedback term in its energy balance, the same as for the heat exchanger. However, the exothermic chemical reaction contributes positive feedback, because the input term (~AH koe */*" Ca/V pC) increases when the output temperature increases, For the parameter values in Example 3.10, the inhes- ent negative feedback in the process dominates, and the process achieves a steady State after a step input. The positive feedback is substantial, however, which leads to the periodic behavior and complex poles. Additional comments on the behavior and stability of processes are given in Appendix C. In summary, many different systems obeying the models of these basic ele- iments behave ina similar manner, After the parameters have been determined, their ‘bchavior for specified inputs is well understood. Thus, the experience learned from 2 few examples can be extended, with care, to many other systems. 5.3 SERIES STRUCTURES OF SIMPLE SYSTEMS ‘A structure involving a series of systems occurs often in process control. As dis- cussed in Chapter 2, this structure can occur because of a processing sequence — for example, feed heat exchange, chemical reactor, product cooling, and product sep- aration, Also, a control loop involves final element (valve), process, and sensor in a series, as will be more fully discussed in Part Il. Therefore, the understanding of how series structures behave is important in the design of chemical plants and process contol systems. Noninteracting Series ‘There are two major categories of series systems, and the noninteracting system is covered first. tis worthwhile considering the mixing system, which conforms to the block diagram at the bottom of Figure 5.4a, in which each intermediate variable has physical meaning, vEEet = FC - FCA 6.23) DYNAMIC BEHAVIOR OF TYPICAL PROCESS systems 157 ® a0 [oY ricure sa {ie Ol Co Series of processes: (a) noninteracting Cimencing @ Mie _ pen ECE VOB = FC FC 6.24) [Note that the mode! equations have the general form fori =1,....0 with ¥j (8.25) Any system modelled with equations of this structure constitutes a noninteracting series system. Important features of the system follow from this model. 1. Only ¥-1 and ¥, (not Yq41) appear in the equation for d¥,/d. 2, Following from (1), the downstream properties do not affect upstream properties: in the example, the concentration in tank 2 does not affect the concentration in } tank 1 but does affect tank 3 3. The model for the general noninteracting series of first-order systems can be de- veloped by taking the Laplace transform of each equation (5.25) and combining them into one input-output expression For a series of systems shown in Figure 4a, each represented by a transfer function G(s), the overall transfer funct Yn XG) For n first-order systems in series, this gives » Yat) T]kr-s X@ Cth Ky and +; for the individual systems [ens +0 io Pal T]G.-«s) 5.26) = GulS)Gy-1(8)- Gals) 6.27) The gains and time constants appearing in equation (5.27) are the same asthe Yalues forthe individual systems asin equation (5.25). Ths, the model of inter ing systems can be determined directly from the individual models.158 raocess pywamucs A. If cach system is stable (i.e., 7 > 0 for all i), the series system is stable. This follows from the important observation that the poles (roots of the characteristic polynomial) of the series system are the poles of the individual systems. Now the dynamic response of a series of noninteracting first-order systems can be considered. Since 50 many possibilities exist, the simplest case of m ident ‘al systems, all with unity gain, is considered. The response to a step in the input, X'(3) = Vs, is plotted in Figure 5.5, Note that the time is divided by the order of the system (ie. the number of systems in series), which time-scales the re- sponses for easy comparison. We note that the shape of the response changes from the now-familiar exponential curve for n = 1. Asi increases, the response begins to have an apparent dead time, which is the result of several first-order systems in series. For very large n, the output response has a very steep change at time equal to nz. Thus, we conclude that the series of identical noninteracting first-order systems approaches the behavior of a dead time with @ ~ nz for large n. Again looking ahead to feedback control, a system with several first-order systems in se- ries would seem to be difficult to control, for the same reasons discussed for dead times. "s second observation is thatthe curves all reach 63% of their output change at approximately the same value of /n7; this will be exploited later inthe section Finally, we note that the system is always overdamped, because the transfer function ih. thas 7 real poles, all at Output of mforder system ° 0s t 15 2 25 Sede time, tne FIGURE SS Responses of nienteal noninteracting fist-order systems with K = 1 in series to aunitsep at = 0. 159 m= 1,2,5,10,20,50 10? 103 10? 107 w 101 Sealed frequency (tie) 18 FIGURES. Brgy eynss on dena oineracting fst sytems wih K = 1 The amplitude ratio ofthe frequency resp be determined dit sponse can be determined directly from the transfer function in equation (5.27) to be fs Ide - foun) = (fx) PxGa] Vit ote? } The amplitude ratio is always less than or equal to the overall gain, and it decreases ly a the frequency becomes large. Amplitude ratios for several series of iden- ical first-order systems are sho i wn in Figure 5.6; again, the frequency is se der of the system to provide time-sealing, reaeeney is seledte ting Series Second major category of series systems is interacts , : ms is interacting systems. Again, itis while considering a physical example, this being the level-ow process in e S4b, Assuming thatthe low through each pipe i a function of the pres ene the mode can be derived based on overall material falance fr ech dt; dt -F = Kalli ~ by) ~ Kil ~ Lies) (Ps ~ Ps) for the linearized system, and the pressures are propor- "othe liquid eves, These model equations ha Thee model equation have the following gene 8 seis of two interacting fist onder systems: eel (5.29)160 process pywamucs — KY, — ¥3) (5.30) Ky ~ 3) ~ Kas ~ Ya) 631) ‘Many important physical systems, including that in Figure 5.4b, have struc- tures described by equations (5.30) and (5.31); thus, these equations are considered representative of interacting systems for subsequent analysis. Some important fea: tures of these systems follow from their model structure: 1. The variables ¥q-1, Yq, and Yps1 appear in the equation for d Yq/dt. 2. Following from (1), the downstream properties affect upstream properties; for ‘example, the exhaust pressure (P5) influences both levels in Figure 5.4b. 3 ‘The model for the general interacting series system of first-order systems can be developed by taking the Laplace transform of equations (5.30) and (5.31) and combining them into one input-output expression, which results in poles of the interacting system that are different from the poles of the individual systems. “The procedure for deriving the overall transfer function is shown in some de- tail, because the result is somewhat more complex than for a noninteracting system and because the procedure can be applied to systems of differing structures. First, the Laplace transform of equation (5.30) can be rearranged to give (with the primes, deleted) VK, 1 with my = Ht 30) STkOt eTHO) with = 32) ny ‘The parameter 7y; is the time constant for the first system when considered individ- ually, The Laplace transform of the second equation is Ki He 2s¥ (8) = FLY As) — Ya(s)] — [¥o(s) ~ Yabs with 7y2 = 22 (5.33) mrasVals) = FM) ~ Ya] — (2G) — YO] witha = FG ‘Again, the parameter 73 is the time constant for the second system when considered individually. The behavior of the combined system can be determined by substituting equation (5.32) into (5.33) to give, after some rearrangement, s+) 7) ¥a(s) = K, mums +(m tmemg el 634) Uke - soit fron Several important conclusions on the effect of the series structure on the dy- namic behavior can be determined from an analysis of the denominator of the trans- fet fur ation, The time constants of the interacting system (| and 72), which are the DYNAMIC BEHAVIOR OF TYPICAL PROCESS SysTEMS 161 af 7 jnverses of the poles, can be determined by solving the quadr: + goots of the characteristic polynomial to give equation for the % oe m2 = = dria — (8.35) aa mime Four characteristics ofthe dynamics of this type of series system are now established. Fins, the possibility of complex poles is determined to establish whether periodic ‘behavior is possible. The expression within the square root in equation (5.35) can be rearranged to give ® KY 2K (m+n +g) Anna = vim) tong em +2, aigt)>o (536) Since both terms in the right-hand expression are greater than zero, the entire ex- pression is greater than zero, and complex poles are not posible for this system Therefore, periodic behavioe cannot occur for nonpeiodic inputs, such asa stp. Second, the stability of the process can be determined from equation (5.35) Note thatthe numerator has the form ~a # (a? ~ BY, with a and b both positive Therefore, the poles fr both signs ofthe root are negative, andthe system fs table ‘Third the “speed” of response othe interacting series system can be compared ‘with the individual system responses. Since the poles are real, the characteristic paly- pomial in equation (5.34) can be written in an equivalent form as (as + Dns +1) = ans? + (1 + ms +1 637) ing the coefficients of like powers of s in equations (5.34) and (5.37) gives K ane mm and atr=mtmt me 6.38) fore, the sum of the time constants for the overall interacting 1 acting system is greater the sum of the individual systems. In other words, the interacting system is lower,” due to the interaction, than it would have been if the systems were nonin- (5.38) show that the product of the time constants is un- but the sum is greater. Therefore, the difference between the interacting time constants (7; ~ 72) is greater than the difference between the individ- time constants (ry; — 1y3); that is, one time constant begins to dominate. This lusion can be demonstrated by rearranging equations (5.38) to give ew ~ me) ty! 2 Ki (m= (on may +g (Ona + rtm) — 639) Since the noninteracting serie acting series systom hasbeen shown to have al el poles, dynamic responses ofan intacting system of fstonder systems have many of same characteristics as those ofa noninteracting system; that is, they ate stable ser 2a they are sab162. aocess pywamics “The previous results for interacting systems are applicable to (only) those systems that conform to the model; in addition to having variables Y,-1. Ym and Yrs appear in the equation ford ¥q/dt, the coefficients of each linearized term must conform to the structure and range of values in equations (5.30) and (5.31) Many systems have the same model structures but different ranges for the val- ues ofthe parameters. For example, the model of the nonisothermal CSTR in Exam- ple 3.10 has both output variables in both equations, but the coefficients, the aij, do not conform to the ranges of values of systems analyzed in this section. Therefore, the chemical reactor would not be considered an interacting series of first-order s tems, and it can experience periodic behavior. Ifthe type of system is not obvious from the structure ofthe equations and the values of the model parameters, the model can be analyzed using the procedure just applied to the equations (5.30) and (5.31) to determine important characteristics ofits dynamic behavior. Finally, some alternative terminology used in automatic control is introdu: ‘What has been referred to here asa series system is sometimes termed a cascade sys- tem; however, cascade is used in process control for another topic covered in Chapter 14, Also, the two types of series systems are sometimes referred t0 as nonloading and loading, for noninteracting and interacting respectively. Noninteracting Series with Dead Time 'As will become more apparent in the next chapter, we often use first-order-with- dead-time models to approximate more complex systems with monotonic step in- put responses, Therefore, noninteracting series of first-order-with-dead-time systems are considered to conclude this section. The direct application of equation (5.26) results in Fin)oo( Se) Yo) Fy ( 4 A Kie™ #8 = Thaw = Sf with Gis) = SE (5.40) re TIe@s+0 ‘This overall transfer function provides the basis forthe following equations, which give values for key parameters of a noninteracting series of first-order-with-dead- time systems, Exactrelationships K=[]K; @= 3.6 Galo it ron tore ~ 01+ 7 (5.416) a The results for the overall gain and dead time follow directly from equation (5.40) ‘The approximation for the time for the output response to a step input to reach 63% ofits final value, ray is based on fitting an approximate model to the response Approximate relationship 163 r of the series system, using the method of moments. The derivation of this expres- i sions provided in Appendix D. The relationships in equations (5.41) are useful for J guickly characterizing the approximate behavior of a noninteracting series system from the individual systems; comparison to solutions of noninteracting systems (e.g, Figure 5.5), shows that the expression for fesq is a reasonable approximation but not exact. Example 5.1. The four first-order-with-dead time systems, with parameters in the fol- lowing table, are placed in a noninteracting series. Describe the output response of this system toa step change inthe input tothe series atime System 10203 4 Dead time, @ 04009012 «1.70 Timeconstant,r 1S 33 5.2095 Gain, K Lo 025 3.01.33 “The results in this section on noninteracting systems indicate tha the ouput response willbe an overdamped sigmoid. Equations (4) ean be used to estumate hey values tf the response. Note thatthe input wccured a =, s thatthe points nated on Figure 57 are based on the following ess as measured fom = 2 Kp2 10 0-2=42 S@+n= 151 tion) ‘The overall sponses compared wit he approximation in Figure 5.7, which demon- sts the wells of pproimaton fe aes Dees ves an approxinac “ime sal’ forte response. However, many sigmoidal euvescouldbe dan Uwough the two points in the figure. The entire curve canbe determined through analytical or sumercl solution ofthe defining equations 2= 15.1 1+ 06 (a= Approximate ‘Approximate ead time os wis mas so Time 87 respons of series processes in Example 5.1 for a unit step att ~ 2164 reocess oywames Example 5.2 Input-output response. Two series systems involve only transporta- tion delays and mixing tanks, A step change is introduced into the input feed composi tion of each system with the flow rates constant, Determine and compare the dynamic responses of the output for each system. Since there is no chemical reaction, the sys- tems have a gain of 1.0 and dynamic parameters given in the following table, 4 4 on on a 4 Case 1 nea 2 0 0 2 2 0 Case2 0 2 12 2 Oe “The solution can be developed in several ways. The most general sto derive the overall, input-output transfer functions for these systems, Vals) = Gls)¥a(9) GANGMNGANGUIKS) As) 1Oe 128506) 989 Xi) ~ Gast Dins+ Dias + Dees + Loe © G4 D+) Since the overall transfer functions are the same for the two systems, their dynamic input-output behaviors are identical. This is verified by the transient responses of the two eases fora step input at time= 2 in Figure 5.8, with each variable ¥,(0) on a sep- arate sea. gy & Te Case 2 responses FIGURE 58 Dynamic responses for series system in Example 5.2 oa unit step at» = ‘The responses in Figure 5.8 show that two systems can have the same input-output behavior with diferent values for intermediate variables In conclusion, the analysis inthis section has demonstrated that both noninter- aeting and interacting series of n first-order systems can be modelled by a transfer function with a characteristic polynomial of order n. Much about the dynamic re- sponses of the series systems can be determined from the models ofthe individual 3s. The results are summarized in Table 5. The series systems in this section provided additional reinforcement for the {importance of transfer function poles. The strongest general conclusions were based ‘on the manner in which the poles of the overall system were or were not affected by series structure. These conclusions concerned stability and the related property ‘of periodic behavior. Since these generalizations dealt with properties completely determined by the poles, they are independent ofthe numerators inthe transfer func- ‘tons. In fact, the gencralizations on stability and periodicity can be extended to any cries transfer functions with denominators expressed as a polynomial ins. However, the values of the poles do not provide general conclusions for the f¥me-domain responses to step and sine inputs. Since both the numerator and de- inator ofthe transfer function influence the dynamic behavior, the more specific ‘on dynamic responses are valid only for systems consistent with the assump- in the derivations — that is, with a constant for the numerator of each series fer function element, In particular, Figures 5.4 and 5.5 and all conclusions on step response and amplitude ratio are specific to systems whose component el- hhave constant numerators. Finally, such strong conclusions for an overall tem, based on the individual elements, are not always possible, as demonstrated the structures considered in the remainder of this chapter, LE 8.1 ies of series systems with first-order elements (responses between t, X, and output, ¥,) ‘Noninteracting Interacting series system, series system ‘equations (5.30) and ($31) frstorder systems ih-oder system border system ‘Stable, not periodic constants, 2 Time cons ats are 1, ‘Time constants are nots. They ceed ‘must be determined by solving the characterise polynomial. is sable Stabe, not periodic ‘Overdamped, sigmoidal __Overdamped, sigmoidal AR= K, forallo ARE K, forall o166 reocess pynamres 54 PARALLEL STRUCTURES OF SIMPLE SYSTEMS Parallel structures involve systems with two or more paths between the input and output variables and thus are fundamentally different from series systems, which have only one path. Two process examples of parallel structures are considered in this section, from which general conclusions will be drawn about the unique features of these systems. The key difference between this and the previous sections is the ‘concentration on the numerator, as well as the denominator, of the transfer function. As discussed in Chapter 4, the numerator contributes important characteristics to the ‘dynamic behavior that cannot be determined from the transfer function denominator alone. The process examples in this section demonstrate the dramatic effects the ‘numerator can have on the dynamic response. Example $.3 Heat exchanger with bypass. Often a process stream must be heated or cooled a variable amount using a heat exchanger. A common method for variable heating isa heat exchanger with bypass, as shown in Figure 5.9 for a cooler; the bypass provides the parallel structure inthis example. The flows through the exchanger and through the bypass are adjusted while the total process flow is maintained constant, ‘(How to achieve such flow behavior via the feedback control will be covered late.) "The behavior of an industrial shel-and-tube heat exchanger would be difficult to model, because itis a distributed-parameter system with complex flow pattems; therefore, the system is approximated as a stirred-tank heat exchanger, which retains the key properties of the system dynamics, in particular the response of the measured ‘temperature signal toa step change in the flow to the exchanger. Assumptions. 1. The same assumptions apply as in Example 3.7, 2. ‘There is no transportation delay in short pipes. {3 The total flow (exchanger and bypass) is constant: Fr ‘ch + Fey = constant Ty Frc pL Py Tee Fe Heat exchanger with bypass and sensor. DYNAMIC BENAVIOR OF TyPICAL PROCESS SYSTEMS 167 ‘Data, Note that these parameters are not realistic for a heat exchanger, although the (s) through two paths. Note that fora parallel path to exist. a split must occur in the block diagram. The overall168 process pywamics 1 16) no Fox) Gls) ® G49 Graf) FIGURE 5.10 Block diagram of exchanger wit bypass and sensor. ‘transfer function, relating the flow tothe exchanger to the measured temperature, can bbe derived from block diagram algebra DYNAMIC BEHAVIOR OF TYPICAL PROCESS SYSTEMS 169 TM) = GAsNGeuls) + GralNGe(9) Fes) 6.49) Ts) _ Kew KexaKras - 550) nal) 3841” Gans + Dis + 1) oo |(_ Krwrexon +t] _ mt keaton (iaieama)*!| -ouaesn a Gaas + Dtes* 1) Gras From equation (5.50) we conclude thatthe poles ofthe overall system are the poles of the individual systems, In this example the system is second-order with rel, distinct poles at —1/enand ~I/r: thus its table and is nt periodic Due wo the parallel srutue, the transfer function has a zero in the numerator, which for this exampleis ats ~ ~(Kru+KeuaKranWKrarTexs and is real and negative fortis example, This ro can significantly affect the ymamie behavior af thes); therefore, the response of the system ta step input canna be determined using Figure 55, which assumed a constant numerator. The dynamic response can be determined by inverting the Laplace transform of 74s) for step in Fa). Solution. By substituting the data in the problem statement into equation (5.50). in cluding the step input, Fexa(3) = ~ 10/5, and determining the inverse using entry 10in Table 4.1, the following analytical solution forthe linear approximation can be found: Ty) = 1.0~2.333e7*S + 1.3336" sn) Results analysis. Dynamic responses are given in Figure 5.11 for the nonlinear and ‘approximate linearized models. They both show that the system output, Ts, overshoots and then approaches its final value smoothly. Unlike the result in Example 4.6, which ‘was a second-order system with distinct factors anda constant numerator. this response experiences maximum. (The occurrence of the overshoot depends on the relative ‘magnitudes of the numerator and denominator time constants.) The time at which the ‘maximum occuts can be determined by seting the derivative of equation (5.51) to zer0 and solving for time, giving r = 1.3 minutes after the step. Thus, the parallet structure hhas fundamentally altered the dynamic behavior ofthis second-order system. ‘The reason for this behavior can be understood by considering the two parallel paths in the physical system. When the exchanger flow is decreased, temperature 7 is initially unaffected, and the modified flow ratio to the mixing point results in ‘an immediate increase in temperature 7. However, the exchanger outlet temperature T; decreases witha first-order response because of the lower flow othe exchanger. As 2 LK. Nonlinear Nontinear Sioa incre *” ° ~ ° 0 20 Time ‘Time or —y 06 Fy os a ° 10 0 ‘Time {URE S.11 a and linearized dynamic responses for Example 5.3. result, the mixture temperature decreases from its initial peak tots final value with a first-order response, The measured temperature follows the mixture temperature afer the sensor frt-onder lag. Note thatthe overshoot is not due toa complex pole and that the behavior is nor periodic. Rather, the behavior is due to a paallel process path with significant differences in dynamics for the two paths, "Naturally, such behavior should be considered in designing and operating the Process. Again, imagine driving an automobile that tends to overshoot the change in ‘direction indicated by the steering wheel; a careful and skilled driver (or contro algo- ‘thm would be required, Example 54 Series reactors. This example demonstrates that the parallel paths do ‘ot have to be extemal bypass streams but can be separate mechanisms within a single rocess. The process considered is a series of two CSTR, shown in Figure 5.12, with the same vessel size, flow rate, and chemical reaction asin Example 3.3; thus, the re- ‘ctor models are identical to those derived in Example 3.3, equations (3.24) and (3.25) (Page 69). In this example the response of the reactant concentration at the outlet ofthe second feactor toa step change inthe solvent flow isto be determined, Formulation. Whereas in Example 3.3 the wal flow was constant and the inlet con ‘centration was changed, in this example the flows of the reactant and solvent can be ‘changed independently. Also, the solvent flow is s0 much larger than the component A flow (F) that we assume that the total flow is the solvent flow that is, F =F, and.170 process pyvawes Series chemical reactors for Example 5.4 Can = (CaPaVF. (Note that F, = 0,085 and Cay = 0.925, so that CF = 0.0786 ‘mole/min.) With these assumptions, the following transfer functions can be derived, Cul). Gt = = -anomelen’ = Fay = Gal) Kx iain , Cw) Grey Fs) oa Gn) 633) Cus) +Vk Kn 0609 ef a2 Eun ~ OM = aes Eee / sh ‘The linearized mode! is represented in the block diagram in Figure 5.13, which shows the parallel paths, In this example, the parallel paths result fom the differ: cent effects of the solvent flow, through changes in the feed concentration and flow rate (residence time), on the outlet concentration of the second reactor, The overall Fo) | Cal xe) ee FIGURES.3 Block diagram of reactors in Example 5.4 DYNAMIC BEHAVIOR OF TYPICAL process sysrEMS 171 transfer function can be derived using the block diagram to give the overall input-output relationship. G20 ~ Gr) + Cul NGeH(9) + CrAIGAIGonl8) 55) ‘Ths expresson leary shows that thee separate effects ofthe input inuence the out patconcenaton. The st eet, xs), nthe flow oresdene tine inthe second tect this elect begins istananesl and inrestes the concentration. The second etet ia()Gr s)he reidence time inthe fst ect, which incase th fed concentration te aes the second eat The hides valves the decease the feed concentration, Cys hae lower ut limasly decreases the second acto outlet concentration, Tie over effct canbe determined by sabtitang the indi tal waster functions into equaton (5) an reaanging to ive +14 Krak FA) Gs +1? =1.66(-808 + 1) 82554 1P (5.56) ‘Again, the system is second-order and has the same poles as the individual el- ‘ements in the system, but because ofthe numerator dynamics the response cannot be determined from a simple series system (ie., Figure $5) Also, the result in this exam- ple is different from the previous example, because the transfer function in equation (6.56) has a positive numerator zero (s = 1/8.0). This is due to the last term in the ‘numerator being large and negative, since Kis is less than 2er0. This result indicates ‘a mechanism for inverse response ofthe outpot variable, in which the initial response of the output can have the sign apposite to is final, steady-state change. Solution, Again, the response can be determined by solving forthe inverse Laplace transform using Table 4.1, entry 8. Substituting the data in the problem) statement, including the input step of Fs(s) = AF,/s = 0.00855, gives, Cralt) = 0.0141 + 0.0141 + 0.00337 25 657) Results Analysis. The response to a step change in the solvent feed flow, with reactant flow unchanged, is shown in Figure 5.14 for the nonlinear and approximate linearized ‘models. Note that the outlet reactant concentration initially increases, because of the ‘decrease in residence time, which affects both reactors, including the last, immediatly. However, the decreased feed concentration decreases the reactant concentration, ini- Lally in the first reactor and ultimately inthe final reactor. Thus, the outlet concentration in the second reactor experiences an initial inverse response, because the fast effect of| the residence time influences the output before the larger, slower feed concentration effect. Behavior similar to this example is observed in other physical systems, espe- Gially tubular reactors, The series of CSTRS is selected in this example because the ‘mathematical analysis is simpler, but lumped systems in series can serve as an approx- {mation forthe distributed system (Himmelblau and Bischoff, 1968). Modelling and experimental results for inverse responses in tubular reactors are presented by Silver Sein and Shinnar (1982) and Ramaswamy etal. (1971),im “Tank 2 concentra on 009 Solvent low 008 fy 0 30 30 w 30 o Time FIGURE 5.14 Response for series chemical reactor to step in solvent flow in Example 5.4 “The dynamic characteristics demonstrated in this example would be expected to hhave great influence on feedback control Imagine driving an automobile that responded initially inthe inverse direction toa change in steering! The most appropriate response ‘would be 1 eliminate the inverse response by redesigning the process, if possible In summary, parallel paths exist in many processes either because of complex interconnecting flow structures of individual systems or because of parallel effects within a single process. Since the poles are unaffected by a parallel structure, sta bility and damping of the overall system are not affected. This can be seen from equations (5.50) or (5.56), in which the denominator of the overall transfer func- tion has the poles of the individual transfer functions. However, the parallel paths can have a significant effect on the dynamic behavior of the system, and the most complex behavior—overshoot or inverse response—occurs when parallel paths have significantly different speeds of response, so that parallel responses from an input af- fect the output at different times. Also, the approximate time to reach 63 percent of the output change for a step input is affected by the numerator, and itis not simply the sum of the individual time constants. ‘The behavior of parallel systems of first-order individual systems is summa- rized in Table 5.2. Also, the model structures that yield numerator zeros and the ranges of parameter values giving various dynamic behaviors are addressed in ques- tions 5.11 and 5.21. ‘The behavior presented in this section can cause some difficulty in terminology. since a stable overdamped system (€ = 1) is usually thought to have a monotonic 173, TABLE S2 E Properties of parallel systems with first-order elements eaividual first-order systems ach is first onder Parallel system ‘Order of the highest order in a parallel path Stable, not periodic Poles are W/m, Lp fos 0a Step response CCan be mopotonic or experience overshoot or inverse response Frequency response Amplitude ratio can exceed steady-state process gain (or some frequency range) response to a step input. This is true when the transfer function numerator is a con- stant, but itis not necessarily true when the numerator isa function of s. The potential jamic behavior is summarized in the following table. Response to nonperiodic input Periodic Nonperiodic Monotonic response to step Not possible Possible, depends on numerator plex ‘The emphasis on complex dynamic responses in the examples in this section does not indicate that all systems with numerator er0s give unfavorable dynamics such as large overshoot or inverse response, ‘The engineer can analyze the physical process for possible parallel paths with liferent dynamics to identify potentially complex dynamics and then use quanti- ive methods to determine whether the behavior may cause difficulty for control ch input must be considered separately, because the characteristics of the output ic response differ for different inputs. RECYCLE STRUCTURES yele structures are used often in process plants, o return valuable material for re- ssing and to recover energy from effluent streams through heat exchange. Such onneetions, termed process integration, are often cited as potential causes of| culty in plant operations inspite of their advantages in the steady state; therefore, is important to understand the effects of recycle on process dynamics. Again, this ture will be introduced through a process example and then will be generalized. Example 5.5 Reactor with feed-effluent heat exchanger. The process design shown in Figure 5.15 has a feed-effluent heat exchanger aa can be used for a chem- ‘cal reactor with a high feed temperature and a need for cooling the product effluent stream,174 Process vynamics 1p 19) Gi) Ga) h [en ricunesis ricunesis Eos Formulation. The analysis begins with the transfer functions of the following ind vidual input-output relationships, represented inthe block diagram in Figure 5.16 Ti) _ _ Kn Ta) _ P Trey 7 Gm) = =P FA = Gini) 638) Tas) THs) = THs) + TH) Gris ee Tus) ~ Os ‘The block diagram shows the output ofthe reactor returning to an inpot to the reactor. This is feedback that has heen introduced into the penoese by a recycle of energy. To determine the behavior of the integrated system, the overall input-output ‘wansfer function must be determined using block diagram algebra TU) = Gu(SITS) = GylSHLTV(8) + TH15)} 639) GulSNGna(NTAS) + Gus s)TOSN Tas) __GalS)Guls)_ 1) ~ T= GeOGn) IL is immediately apparent from the overall transfer function that recycle has fundamentally changed the behavior ofthe system, because the characteristic polyno ‘ial in equation (5.59) has been influenced and the poles of the averall system are not ‘the poles of the individual units. Thus, the stability of the overall system cannot be guaranteed, even if each individual system is stable! ‘The chemical reaction could be either exothermic or endothermic. neither case, the result of an increase in reactor feed temperature 75 i an increase inthe reactor outlet temperature Ts, The difference is thatthe steady-state gain for ATW/AT; is between (0.0 and 1.0 for an endothermic reaction and greater than 1.0 for an exothermic reactor. In this example, an exothermic reaction is considered. The parameters in the linearized transfer functions are as follows. 3 Ts +1 ‘The transfer functions were selected to be representative but simple enough to yield a low-order overall model that ean be understood easily. These transfer functions caf be substituted into equation (5.59) to give the effect ofthe inlet temperature Ty on the reactor temperature 7, forthe system with and without recycle Gas) = Guns) = 040 Gya(s) = 030 DYNAMIC BEHAVIOR OF TYPICAL PROCESS SYSTEMS 175 peo (an)! Tad Gu(s)Gals) ed (5.61) Tas) 3| Results analysis. The foregoing expressions and the dynamic responses for a step input of 2°C in Ty in Figure 5.17 show the dramatic effect of reeyele onthe steady- state gain and time constant; both increase by a factor of 10 due to recycle. Ths can be understood by analyzing the interaction hetween the exchanger and reactor during a transient; an increase in Ty causes an increase in Ts and then Ts, which causes an increas in T;, which causes an increase in, and Soon; in shot, the output change is reinforced though the recycle (feedback) exchanger. The system is still tale and self-regulatory, because ofthe dominant inherent negative feedback forthe parameter values in this example, but the recycle has created an inherent postive feedback in the process, which has significantly affected the dynamic response. The potentially unfa- 30 100-150 20) 250 300 350400 «450500 sa7 Fesponses for a 2°C step in To at time = 0 Example 55 with and without176 eocess pyNamies vorable dynamic effects of recycle can be reduced through automatic control strate gies, which retain most of the process performance benelits, as demonstrated for this ‘chemical reactor design in Figure 24.11 ‘The simple example in this section demonstrates the potential effects of recycle ‘on dynamic behavior: 1. Recycle can alter the stability and possibility for periodic behavior of the over il system, because it affects the poles of the overall system. 2. The time constants and steady-state gain of the overall system with recycle can be changed substantially from their values without recycle. Again, understanding the effect of recycle on dynamic responses is an important aspect of process dynamics, and the material in this section is enhanced by additional analysis in Appendix C and by reference to the studies of recycle in the Additional Resources at the end of this chapter. Finally, the process recycle considered here has the characteristic of having & process output return to affect a process input, which is the definition of feedback introduced in Chapter |. In later parts of the book, the negative-feedback principle is applied, to reduce the effects of disturbances, by means of control calculations that adjust a process input. Thus, recycle and feedback involve the same principle: the term recycle is usually used for process structures and feedback for external control calculations and the resulting input manipulations. 5.6 STAGED PROCESSES Staged provesses are used widely inthe process industries for multiple contacting of streams and can be considered as a special interconnection of elements, in which an clement exchanges material and energy with only the adjoining stages. Some com- mon examples are vapor-liquid equilibrium (Treybal, 1955), multieffect evaporation (Nisenfeld, 1985), and flotation (Narraway et al., 1991). Staged systems can experi- ‘ence a wide variety of dynamic behavior depending on the physical processes (8 sass transfer, heat transfer, and chemical reaction) that occur at each stage. The fundamental model for a staged system must include all significant bal- ances on every stage. However, the variables at every stage are not always of great {importance for the overall performance of the process, because only the properties of the streams leaving the process are usually of interest. In some cases, a few interme- diate variables could be important; an example isthe flows on stages of a stripping tower, which might approach or exceed the hydraulic limits for proper contacting efficiency. We will assume in this section thatthe only output properties of interest are inthe product streams. Tn this section the dynamics ofa distillation tower, shown in Figure 5.18, are considered as an example of staged systems to introduce the modelling approach and describe typical dynamic behavior. An accurate model of a multicomponent DYNAMIC BEHAVIOR OF TYPICAL PROCESS SystEMS 177 PM, Distlte = FMyBowns IRE S18 Xe Disillation tower, stillation tower must consider complex thermodynamic relationships and employ al numerical algorithms for the simultaneous solution of equilibrium expres ons and material and energy balances. To simplify the presentation while main a realistic model, the tower considered will separate only two component the phase equilibrium is assumed to be well represented by a constant relative folatlity (Smith and Van Ness, 1987). Also, the energy balance at each stage can be fied by the assumption of equal molal overflow, which implies that the heats ‘vaporization of both components are equal and mixing and sensible heat effects negligible. ‘The assumptions are ‘The liquid level on every tray remains above the weir height. Equal molal overflow applies. Relative volatility a and heat of vaporization A are constant. Holdup in vapor phase is negligible ‘The following nomenclature is used: MM = molar holdup of liquid on tray FM = molar flow rate of liquid X = mole fraction of light component in liquid A = heat of vaporization YM = molar flow rate of vapor Y = mole fraction of light component in vapor ‘The schematic of a general tray in Figure 5.19 shows that every tray has the tal for feed and product flows and heat transfer. With the assumptions and the eral tray structure, the basic overall and component balances for each stage or @ = 1...) can be formulated as178 process pywamies Lut Vapor EM vM, i iM) —+ VM: Ae : 2 m+ vm, FIGURE S.9 4, vn Genera tray sein modelling distin Overall material (molar) balance on liquid phase: aMM 0 Mier — FM, FEM ys — FM 6.02) a f * Quasi-stead-stae overall material (molar) balance on vapor phase: YM = VM}_, — VM + 2 (5.63) VM = VM)-1 + VMj; oo) YM-1Yien + YM pi? ao Yat 665) Ya ae Light component balance on the tra: aonxi) - SOM eM Kiss + FMyXyi— Mp +EMOX 5.6) (VM; + VMpi)¥i + YM} .¥7-, ‘This formulation is adequate for every equilibrium tray in the tower. For most trays, feed flows, product flows, and heat transferred are zero, while at least one tray has a nonzero feed. The top tray has a liquid feed, which is reflux, and its vapor stream goes to the total condenser. The bottom tray has its liquid go to the kettle reboiler, which is also an equilibrium stage. Note that although the equations can be formulated as shown, the computer implementation in this form would involve extensive multiplications for the zero streams; thus, an efficient implementation fot a specific design would eliminate streams that are always zero. Since there are many more variables than equations in the conservation bal ances, the model is not completely specified by these balances alone. The model requires constitutive expressions to relate liquid and vapor compositions. The phase equilibrium equation for a binary system with constant relative volatility « is aX, a 6.67) T+@- Dx, ¥ DYNAMIC BENAUIOR OF TyPICAL PROCESS SysTEMS 179 ‘The model also requires constitutive expressions to relate liquid flows and inventories on the trays. The liquid flow from a tray is related to the level (L; MM/pwA) above the weir height, Ly, by (Foust et al., 1980) MM, _ pu with A being the cross-sectional area and py moles/m?. The modelling effort is not ‘complete until models are developed for the associated equipment, which for this, distillation tower includes the heat exchangers that vaporize part of the liquid ulated in the bottom drum and condense the overhead vapor. The behavior of se is not particularly complex but requires feedback control to model properly. ‘maintain simple model structures without the need for control at this point, the ler duty is assumed to be proportional to the heating medium flow, and the va- ‘overhead is assumed to be completely condensed without subcooling, so that the sure is maintained at a constant value by adjusting the condensing duty, thus Qeons = VM nA (5.69) Ques = KesFeon (5.70) Also, the volumes in the overhead and bottom accumulators can be modelled overall and component balances. In reality, the levels of these inventories would ‘controlled by adjusting the product flows; in this example, the levels are assumed tly constant, so that the models become FMp = VM, ~ FMe 6.7) FMg = FM ~ VMo (5.72) j,__ The composition in the overhead accumulator (X11 = Xp) can be determined component material balance: VM, Yn ~ Xp(FMp + FMg) = VM,(¥q — Xp) (5.73) Again, with the inventory constant, the kettle reboiler can be modelled with a t material balance (Xo ~ Xp), equilibrium relationship, and a calculation ‘vapor flow based on heat transferred wm, 245 = FM.X; — FMaXp — VMo¥o (6.74) aXe Yo= TF @-DXe 2 VMy = 2a 6.76) _ To specify the system completely, sufficient external input variables must be fied that the degrees of freedom are zero. The feed flow and composition must Specified along with two additional variables, here sclected to be the distillate flow Fp and the reboiler heating flow Fy. With these external variables ified, the degrees-of-freedom analysis summarized in Table 5.3 shows that180 process pyxames Distillation degrees of freedom for n trays Bxternal specified Equations Variables dependent) variables independent a (G02) G60 foreach MM, FM, VM, X, PX, YM Yy ™ tray ¥,¥", V" for each tray FM, VMp, @ for each - plus FMyo1, Xeoss Vow tray am % tne om Overhead (569), (5.71, and aoe PMc of FMp, MM 67 ° 0 ® Reboiler (5.70, (872), 57, Xp. Ma, and Ooo Faas MM . (S75), nd(8.36) 8) ° ® Toad Tat Tn 8 tne the system is exactly specified. The number of equations is equal to the num- ber of variables; thus, there are zero degrees of freedom. Note that the parameters (Qa, Ke. MM, Krety MM, and L,,) were excluded from the analysis, because they are always constant. Also, the feed variables are determined by upstream process conditions. Typically, external variables like the reboiler heating flow rate and the distillate product flow rate are adjusted to achieve the desired product compositions; here, they are asstimed known external variables. The model formulation included assumptions, like constant accumulator levels and pressure, that are not necessary but simplify the model and presentation Example 5.6. Determine the dynamic behavior ofa binary distillation tower with the parameters in Table 5.4. The model equations can be integrated numerically t0 determine the response ofthe system rom specified initial conditions for any valves TABLE 54 Base case design parameters for example binary distillation Relative volatility 2 Number of rays " Feed tay 9 Analyzer dead times 2 min Feed light key x, = 080 Distillate light key Xp = 0.98 faction Bovtoms light key Ny = 0.02 faction Feed flow FM, = 100 kmolemin Reflux flow FMy = 833 kmoe/nin Disillate how PM, = SO kmole/min ‘Vapor reboited YM = 1353 kmolefmin Tray holdup MM, = 10 kmole Holdup in drums MMy = MMp = 10.0kmole DYNAMIC BEHAVIOR OF TYRICAL pRocess systems 181 or functions ofthe external variables. The dynamic responses are obtained by estab- lishing a steady-state operating condition and introducing a single step change to one fof the external variables; each step is 1 percent ofthe base case input value, (This is ‘exactly how the experiment would be performed on the physical tower, as explained in ‘Chapter 6.) The results are shown in Figures 5.20a and b, The composition responses are smooth monotonic sigmoidal curves in spite of the complexity of the process. Note that changing a single input affects both product compositions—an important factor in subsequent control design as discussed in Chapters 20 and 21 Staged systems present a wide range of dynamic behavior and challenging ntrol problems, Because of their industrial importance, they have been studied Fpxtensively. A few of the important results are summarized in the following list. B). All modet equations could be linearized, and the linearized model and transfer functions could be analyzed. An analytical solution has been developed for the response to a step input of a linearized binary tower section, without reflux or reboil with constant liquid holdup on all trays (Amundson, 1966). The poles ofthe model described in this section have been demonstrated to be all real and distinct (Levy et al., 1969). More realistic models, which include energy } balances, have complex conjugate poles; however, the constants associated with p these poles are generally quite small (Levy et al., 1967; Heekle et a., 1975) ‘Therefore, one can expect the real poles to dominate the behavior of the response, and expect simple distillation towers without control to experience nonperiodic behavior The numerator zeros of the system, when they exist, tend to cancel some of the poles approximately, so that the system for some inputs behaves like a lower ‘order process (Kim and Friedly, 1974). This explains the response of the bottom ‘composition in Figure 5.20a, which appears close to a first-order response. ‘The dynamic response depends strongly on the operating conditions of the distl- lation tower (Fuentes and Luyben, 1983; Kapoor et al., 1986). Thus, an approxi- ‘mate linear model at one set of operating conditions can be a poor approximation for other operating conditions. Although the liquid holdup on each tray is small, and thus the time constant of each tray when considered individually is on the order of tens of seconds, the dynamics of a distillation tower can be long, on the order of hours for some tow- ers. Experience indicates that slow dynamics tend to occur for high-purity towers With low relative volatility. This large difference in dynamic response from the individual elements results from the tray structure with recycle in the design of istillation. This statement is supported by numerical results for towers with re- ‘istic numbers of trays in Fuentes and Luyben (1983) and Kapoor etal. (1986) This summary presents a small sample of the results available on distilla- Pa dynamics. They have been presented as general guidelines for the behavior of ‘Product distillation with simple thermodynamics (e.g., no azeotropes) and no al reaction. The reader is encouraged to refer to the citations and Additional for further details. This distillation example will be considered in later