Analysis of the Dynamic Behavior of Chemical Processes

III

In process control. the processes are analytically complex but relatively simple to control in practice ... These processes act quantitatively like nothing we know. but qualitatively (andfor certain quantitative purposes). they can be modeled in terms of simple gains. deadtimes, lags or combinations thereof ...

E. H. Bristol"

In Part III we study the dynamic and static behavior of several, simple processing systems. Understanding the dynamics of such simple systems allows us to analyze the behavior of more complex systems such as the chemical processes.

The analysis is limited to linear dynamic systems. This may seem incompatible with the fact that most of the chemical engineering processes are modeled by nonlinear equations. However, linear techniques are very valuable and of great practical importance for the following reasons: (1) There is no general theory for the analytic solution of nonlinear differential equations, and consequently no comprehensive analysis of nonlinear dynamic systems. (2) A nonlinear system can be adequately approximated by a linear system near some operating conditions. (3) Significant advances in the linear control

·Industrial Needs and Requirements for Multivariable Control," in Chemical Process Control, A. S. Foss and M. M. Denn (eds.), AIChE Symp. Ser. 72, No. 159 (1976).

111

112

Analysis of the Dynamic Behavior of Chemical Processes

Part III

theory permit the synthesis and design of very effective controllers even for nonlinear processes.

Fundamental, therefore, is the concept of linearization and the procedure for approximating nonlinear systems with linear ones which will be studied in Chapter 6.

The Laplace transforms, which we will discuss in Chapter 7, offer a very simple and elegant procedure for the solution of linear differential equations (Chapter 8) and consequently for the analysis of the behavior of linear systems.

In the remaining chapters of Part III we will discuss:

1. The development of simple input-output models for chemical processes, using the Laplace transforms (Chapter 9)

2. The dynamic analysis of various typical processes (Chapters 10 through 12)

Computer Simulation and the Linearization of Nonlinear Systems

6

In order to find the dynamic behavior of a chemical process, we have to integrate the state equations used to model the process. But most of the processing systems that we will be interested in are modeled by nonlinear differential equations, and it is well known that there is no general mathematical theory for the analytical solution of nonlinear equations. Only for linear differential equations are closed-form, analytic solutions available.

When confronted with the dynamic analysis of nonlinear systems, there are several things we can do, such as:

1. Simulate the nonlinear system on an analog or digital computer and compute its solution numerically, or

2. Transform the nonlinear system into a linear one by an appropriate transformation of its variables, or

3. Develop a linear model that approximates the dynamic behavior of a nonlinear system in the neighborhood of specified operating conditions.

Alternative 2 can be used in very few cases, whereas alternatives 1 and 3 are, in principle, always feasible. In this chapter we discuss the computer simulation of nonlinear processes very briefly because it is a subject to be covered primarily in a course on numerical analysis. More emphasis will be given on the approximation of nonlinear models by linear ones. It should be noted that all the theory for the design of control systems, available from past work, is based on linear systems,

113

114

Analysis of the Dynamic Behavior of Chemical Processes

Part III

and that very small advances have been made toward the development of a control theory for nonlinear systems.

6.1 Computer Simulation of Process Dynamics

Nonlinear differential and/or algebraic equations cannot, in general, be solved analytically, and computer-aided numerical solutions are required. Numerical solutions are also preferred for the equations which can be solved analytically, when the analytic solutions are very complex and provide little insight in the behavior of the system.

Let us consider two processes we have already modeled: the continuous stirred tank reactor and the ideal, binary distillation column.

The model for the CSTR (see Example 4.lO) is given by eqs. (4.8a), (4.9a), and (4. lOb). These constitute a set of nonlinear equations for which there is no analytic solution available. Therefore, in order to study the dynamic behavior of the CSTR, we must solve the modeling equations numerically using a computer.

The model for the ideal, binary distillation column (see Example 4.13) is composed of

2N + 4 nonlinear differential equations

and

2N + 1 nonlinear algebraic equations

It is not only the nonlinearity of the equations but also the size of model (24 differential and 21 algebraic equations for a modest lO-tray column) that necessitates a numerical solution in order to study the dynamic behavior of the column.

Today, computer simulation is used extensively to analyze the dynamics of chemical processes or aid in the design of controllers and study their effectiveness in controlling a given process. Analog and digital computers have been used for this purpose, with the emphasis having shifted almost entirely in favor of the digital computers.

Historically, analog computers were the first to be used to simulate the dynamics of chemical processes with or without control. They permitted a rapid solution of the modeling equations, thus providing useful insight as to how a process would react to external disturbances or how effective was the control of the process using various measurements, manipulated variables, and control configurations. The analog computers have several serious drawbacks: (1) Significant time is required to set up the problem and get it running; (2) the need of one hardware element per mathematical operation prohibits the simulation

Chap. 6 Computer Simulation and the Linearization of Nonlinear Systems

115

of large, complex systems; (3) the nonlinear terms are simulated by rather expensive hardware elements (function generators) with limited flexibility; and (4) they do not possess memory like the digital computers. The subsequent revolution, brought about by the digital computers, made the analog computers obsolete. Today, they are still used on a small scale and primarily to train operators on the dynamic operation of chemical plants.

The computational power introduced with the digital computers, together with the resulting low cost of computations, has expanded tremendously the scope and the practical significance of computer simulation for process dynamics and control. The availability of sophisticated equation-solving routines for almost every digital computer system available has simplified the required groundwork for process simulation and has relieved the engineer of the need to be an expert in numerical analysis.

Digital computer simulation of process dynamics involves the solution of a set of differential and algebraic equations, which describe the process. There are several categories of numerical methods which can be used to integrate differential equations and solve algebraic ones. Let us examine briefly the simplest and most popular among them.

Numerical solution of algebraic equations

At steady state, the state equations turn to simple algebraic equations, since the rate of accumulation becomes zero. Therefore, in order to determine the steady-state behavior of a process under given conditions, we should be able to solve sets of algebraic equations. All available methods use an iterative trial-and-error procedure, which ', ..

I

approaches (hopefully) closer and closer to the solution with each itera-.

tion. The key question is to select the appropriate method, which for the given set of equations converges rapidly to the correct solution. Unfortunately, this is a very difficult task and in all but a few instances is impossible to know a priori how successful a method will be in finding the solution to a particular set of equations. Quite often, a method will not converge to the solution, or in other instances it approaches the solution very slowly. Among the most commonly used techniques are the following: (1) interval halving, (2) successive substitution, and (3) Newton-Raphson.

Numerical integration of differential equations

Here again we have a very large number of available techniques.

Numerical integration implies an approximation of the continuous differential equations with discrete finite-difference equations. The vari-

116

Analysis of the Dynamic Behavior of Chemical Processes

Part III

ous integration methods differ in the way they implement this approximation. Thus we have explicit methods which march on in time yielding the solution in one pass, or we have implicit methods with predictor-corrector capabilities. The key questions for an integration technique are the stability of the procedure and the speed with which it reaches the solution. But again, these are questions which, in general, cannot be answered to our satisfaction ahead of time. Among the most popular integration methods is the explicit fourth-order Runge-Kutta, which provides satisfactory accuracy and stability of computations as well as low cost.

Digital computer simulation of the chemical process dynamics is used extensively at the present. It allows the engineer to anticipate the behavior of a process not only qualitatively but also quantitatively. It has helped to design more complex and sophisticated control systems. The major drawback of computer simulation is that "it only gives you numbers" and not a general analytic solution in terms of arbitrary, unspecified parameters, which in turn you taylor to your particular problem. Therefore, the results of computer simulation are of ad hoc nature, and you will have to make several runs with different values for the input variables and parameters before you can establish a good understanding of the process dynamics for a range of operating conditions.

6.2 Linearization of Systems with One Variable

Linearization is the process by which we approximate nonlinear systems with linear ones. It is widely used in the study of process dynamics and design of control systems for the following reasons:

1. We can have closed-form, analytic solutions for linear systems.

Thus we can have a complete and general picture of a process's behavior independently of the particular values of the parameters and input variables. This is not possible for nonlinear systems, and computer simulation provides us only with the behavior of the system at specified values of inputs and parameters.

2. All the significant developments toward the design of effective control systems have been limited to linear processes.

First, we will study the linearization of a nonlinear equation with one variable and then we will extend it to multi variable systems.

Consider the following nonlinear differential equation, modeling a given process:

Chap. 6 Computer Simulation and the Linearization of Nonlinear Systems

117

dx -=f(x) dt

(6.1)

Expand the nonlinear function f(x) into a Taylor series around the point Xo and take

f(x) =f(xo) + (df) x - Xo + (d2f) (x - XO)2

dx xo I! dx? xo 2!

(6.2)

+ ... + (dnf) (x - xo)n + ...

dx" xo n!

If we neglect all terms of order two and higher, we take the following approximation for the value of f(x):

f(x) ~ f(xo) + (df) (x - xo) (6.3)

dx xo

It is well known that the error introduced in the approximation (6.3) is of the same order of magnitude as the term

1= (d2f2) (x - XO)2

(6.4)

dx -o 2!

Consequently, the linear approximation (6.3) is satisfactory only when x is very close to Xo, where the value of the term I is very small.

In Figure 6.1 we can see the nonlinear function f(x) and its linear approximation around Xo. From the same picture it is also clear that the linear approximation depends on the location of the point Xo around which we make the expansion into a Taylor series. Compare the linear approximation of f(x) at the points Xo and Xl (Figure 6.1). The approximation is exact only at the point of linearization.

n»

[(xl'" [(xol + (¥X )xo (x - xol

x

Figure 6.1 Linear approximation of a nonlinear function.

118

Analysis of the Dynamic Behavior of Chemical Processes

In eq. (6.1), replace !(x) by its linear approximation given by eq. (6.3) and take

dx = !(xo) + (d!) (x - xo)

dt dx =o

This equation is the linearized approximation of the initial dynamic system given by eq. (6.1). In later chapters the design of the process controller will be based on such approximate linearized models.

Example 6.1

Consider the tank system shown in Figure 6.2a. The total mass balance yields

dh A-=Fi-Fo dt

(6.6)

where A is the cross-sectional area of the tank and h the height of the liquid level. If the outlet flow rate F is a linear function of the liquid's level,

F; =ah

where a = constant

we take

dh

A -+ah =Fi

dt

which is a linear differential equation (modeling a linear dynamic system) and no approximation is needed.

If, on the other hand,

r, =p Jh

h

h
A

r Fo (a)

(b)

Figure 6.2 (a) Tank system of Example 6.1; (b) approximation of liquid level response.

Part III

(6.5)

Chap. 6

Computer Simulation and the linearization of Nonlinear Systems

119

the resulting total mass balance yields a nonlinear dynamic model, dh

A dt +p Jh = r,

(6.7)

Let us develop the linearized approximation for this nonlinear model. The only nonlinear term in eq. (6.7) is p Jh. Take the Taylor series expansion ofthis term around a point ho:

pJh = p.Jho + [!!_ (pJh)] (h - ho) + [!f_ (pJh)] (h - hO)2 + ...

dh h=h, dh 2 h=h, 2!

= p.Jho + -P_ (h - ho) - -P_ (h - ho)2 + ...

2.Jho 8 fh~

Neglecting terms of order two and higher, we take

p Jh = p.Jho +-P_ (h - ho) 2.Jho

which, if introduced in the nonlinear dynamic system (6.7), yields the following linearized approximate model:

dh P P

A-+-- h = r, --.Jho

dt 2.Jho 2

(6.8)

Let us compare the linearized, approximate model given by eq. (6.8) to the nonlinear one, given by eq. (6.7). Assume that the tank is at steady state with a liquid level ho. Then at time t = 0, we stop the supply of liquid to the tank, while we allow the liquid to flow out. Thus at t = 0 the liquid level is at the steady-state value [i.e., h(t = 0) = ho). Curve A in Figure 6.2b is the solution of eq. (6.8) and curve B in the same figure is the solution ofeq. (6.7). We notice that the two curves are very close to each other for a certain period of time. This indicates that the linearized model approximates at the beginning very well the nonlinear model.

As the time increases and the liquid level continues to fall, its value h deviates more and more from the initial value h» around which the linearized model was developed. Figure 6.2b indicates very clearly that as the difference ho - h increases, the linearized approximation becomes progressively less accurate, as was expected.

6.3 Deviation Variables

Let us now introduce the concept of the deviation variable, which we will find very helpful in later chapters for the control of processing systems.

Suppose that x s is the steady-state value of x describing the initial

120

Analysis of the Dynamic Behavior of Chemical Processes

Part III

dynamic system (6.1). Then

dxs = 0 =/(xs) dt

(6.9)

Consider x, the point of linearization for eq. (6.1) (i.e., Xo = x.). Then eq. (6.1) yields the following linearized model:

dx (dl)

dt =/(xs) + dx x}x - x,)

(6.10)

Subtract eq. (6.9) from (6.10) and take

d(x - x,) = (dl) (x _ x,)

dt dx Xs

If we define the deviation variable x' as

(6.11)

x'=x-xs

then eq. (6.11) takes the form

dx' (dl) ,

dt = dx xsx

Equation (6.12) is the linearized approximation of the nonlinear dynamic system (6.1), expressed in terms of the deviation variable x'.

The notion of the deviation variable is very useful in process control. Usually we will be concerned with maintaining the value of a process variable (temperature, concentration, pressure, flow rate, volume, etc.) at some desired steady state. Consequently, the steady state becomes a natural candidate point around which to develop the approximate linearized model. In such cases the deviation variable describes directly the magnitude of the dislocation of a system from the desired level of operation. Furthermore, if the controller of the given process has been designed well, it will not allow the process variable to move far away from the desired steady-state value. Consequently, the approximate linearized model expressed in terms of deviation variables will be satisfactory to describe the dynamic behavior of the process near the steady state.

In subsequent chapters we will make extensive use of the linearized forms of differential equations, in terms of deviation variables.

(6.12)

Example 6.2

Consider the linearized model of the tank system given by eq. (6.8) of Example 6.1. Let h , be the steady-state value of the liquid level for a given value, Fi,s of the inlet flow rate. Then the linearized model around h ,

Chap. 6 Computer Simulation and the Linearization of Nonlinear Systems

121

gives

Adh +-P_ h = r, _I!. Jhs

dt 2 Jhs 2

At steady state from eq. (6.7) we also have

A ~+P Jhs= r.,

0= rt

Subtract eq. (6.14) from (6.13),

. d(h-hs) P

A + -_ (h - h ) = F - F

dt rz:- s II,S

2 yhs

Defining the deviation variables

(6.13)

(6.14)

h'=h-hs

and

F, = F, - Fi,s

we take the following linearized form in terms of deviation variables:

A dh' + -P_ h' = r: dt 2 Jhs

( 6.15)

6.4 Linearization of Systems with Many Variables

In previous sections we developed the linearized approximation of a nonlinear dynamic system that had only one variable. Let us now extend that approach to systems with more than one variable.

Consider the following dynamic system:

dXI

dt =/1(xI, X2) (6.16)

dX2

dt = 12(X I, X2)

(6.17)

Expand the nonlinear functions II(xI, X2) and 12(XI, X2) into Taylor series around the point (x 1,0, X2,O) and take

II(x I, X2) = II(x 1,0, X2,O) + [all] (x I - x 1,0) + [all] (X2 - X2,O)

aXI (XI,O,X2,O) aX2 (XI,O,X2,O)

+ [a2/1] (XI - X 1,0)2 + [a2/1] (X2 - X2,O)2

axr (XI,O,X2,O) 2! ax~ (XI,O,X2,O) 2!

122

Analysis of the Dynamic Behavior of Chemical Processes

Part III

and

[ a2f ]

+ 2 (XI - XI,0)(X2 - X2,0) + ' ..

aXlaX2 (XI,0,X2,ol

Neglect terms of order two and higher and take the following approximations:

II(xI, X2) "'=' II(XI,O, X2,O) + [all] (XI - X 1,0) + [a all] (X2 - X

ax I (x I,O,X2,ol X 2 (x l,o,x2,ol

and

Substitute the foregoing linear approximations of II(x I, X2) and 12(XI, X2) into eqs. (6.16) and (6.17) of the initial nonlinear dynamic system and take

dx I = II(x 1,0, X2,O) + [all] (x I - x 1,0) + [all] (X2 - X2,O) (

dt aXI (XI,o,x2,ol aX2 (XI,O,X2,ol

dX2 = 12(X 1,0, X2,O) + [aI2] (XI - X 1,0) + [a/2] (X2 - X2,O) (

dt aXI (XI,o,x2,ol aX2 (XI,o,x2,ol

These last two equations are linear differential equations and constitute the linearized, approximate model of the initial nonlinear system described by eqs. (6.16) and (6.17).

The comments made earlier for the one-dimensional case apply also here:

1. The approximation deteriorates as the point (x., X2) moves away from the point (x 1,0, X2,O) of linearization.

2. The linearized approximate model depends on the point (x 1,0, X2.0) around which we make the Taylor series expansion.

Let us now express the linearized system in terms of deviation variables. Select the steady state (x I,s, x 2,s) as the point around which you will make the linearization [i.e., in eqs. (6.18) and (6.19) put XI,O'" XI,s and X2,O ... X2,s]' At the steady state, eqs. (6.16) and (6.17) yield

(6.20)

Chap. 6 Computer Simulation and the Linearization of Nonlinear Systems

123

o =/2(XI,s, X2,s) Subtract eq. (6.20) from (6.18) and (6.21) from (6.19) and take

d(xi - XI,s) = [all] (XI - XI,S) + [all] (X2 - X2,s) (6.22)

dt aXI (XI,s,X2,sl aX2 (XI,s,X2,sl

(6.21)

and

d(X2 - X2,s) = [a/2] (XI _ XI,s) + [a/2] (X2 - X2,s) (6.23)

dt aXI (XI,s,X2,sl aX2 (XI,s,X2,sl

Defining the deviation variables by

xi = XI - XI,s

and

eqs. (6.22) and (6.23) take the following form in terms of deviation variables:

dx', , , - = a II X I + a I2X 2 dt

where

[all]

all = -

ax I (x l,s,X2,sl

A final comment is in order. In the previous and present sections we considered the presence of state variables only in the nonlinear functions. Thus for systems with one variable, we had only the state X, and for systems with two variables we had only states XI and X2. The formulation above should not be perceived as restrictive, but it is easily expanded to include the presence of input variables, like the manipulated variables and the disturbances. The following example demonstrates this point.

Example 6.3

Consider a dynamic system described by two state variables XI and X2 and the following state equations:

dXI f

-- = I(XI, X2, ml, m2, dl)

dt

dX2 f

- = 2(XI, X2, m I, m2, d2)

dt

124

Analysis of the Dynamic Behavior of Chemical Processes

where m I and m» are two manipulated variables and d, and d: are two disturbances affecting the system.

Linearization ofthe equations above around the nominal values (x 1,0,

X2,0, ml,O, m2,0, dl,o, d2,0) will yield

d;, =/I(XI,O, X2,0, ml,O, m2,0, dl,o) + (:~IJo (x - X 1,0) + (:~IJo (X2 - X2,0)

and

d;2 =/2(XI,O, X2,0, ml,O, m2,0, d2,0) + (:~2Jo (XI - X 1,0) + (:~2Jo (X2 - X2,0)

() (af) (6.25)

+ ( a/2) (m 1- m 1,0) + a/2 (m2 - m2,0) + -d2 (d2 - d2,0)

ami 0 am2 0 a 2 0

where all the derivatives have been computed at the point oflinearization (denoted by the subscript 0).

Assuming that the point of linearization corresponds to the steadystate operation of the system, we can define the deviation variables by

d'; = di> dl,o d2=d2-d2,0

X2 = X2 - X2,0

Introducing these deviation variables in eqs. (6.24) and (6.25), we take

(6.26)

and

dX2 , 'b' b '+ d'. -=a2lXI+a22X2+ 21ml+ 22m2 C22

dt

where the constants a ij, b ij, and C i are the appropriate derivatives in the eqs. (6.24) and (6.25):

all = (a/I)

aXI 0

(6.27)

bl2 = (a/I)

am2 0

and

a22 = (a/2)

aX2 0

Part III

Chap. 6

Computer Simulation and the Linearization of Nonlinear Systems

125

It should be noted that the modeling eqs. (6.26) and (6.27) are in the form that we would like to have for process control purposes (i.e., linearized approximation of the nonlinear state equations), in terms of deviation variables.

Example 6.4: Linearization of a Nonisothermal CSTR

The modeling equations for a CSTR were given in Example 4.10 by eqs. (4.8a), (4.9a), and (4.10b). Assume that the volume V of the reacting mixture remains constant. Then the dynamic model of the reactor is reduced to the following:

dCA I ( k -EIRT

-=- CA-CA)- oe CA

dt r I

(6.28)

and

dT 1 UA

-d =-(Ti- T)+Jkoe-EIRTcA- __ ' (T- Tc)

t r pc, V

(6.29)

where, 1/. = Fd V. This model is nonlinear due to the presence of the nonlinear term e-E1RT CA, while all the other terms are linear. Thus to linearize eqs. (6.28) and (6.29), we need only to linearize the nonlinear term around some point (c A,O, To).

e-E1RT CA ~ e-E1RT'CA,0 + [a[e-E1RT CA1] (T _ To)

er (TO,CA,O)

[a[e-E1RT CA1]

+ (CA - CAO)

ac A (TO,cA 0) ,

= e-E1RT'CA,0 + (R~5 e-E1RT'CA,0}T - To) + e-E1RT'(CA - CA,O) Substituting the approximation above into eqs. (6.28) and (6.29), we take the following linearized model for a nonisothermal CSTR:

dCA 1

- = - (CA' - CA) (6.30)

dt • I

- k{ e-E1RT,c A,O + R~5 e-E1RT,c A,O( T - To) + e-E1RT,( C A - C A,O)]

dT I [ E

- = - (Ti - T) + Jko e-E1RT'CAO + -_ e-E1RT,c (T - T)

dt r ' RTij A,O 0

+ e-E1RT'(CA - CA,O)] - UA, (T - Tc)

pCpV

(6.31)

We can proceed a step further to develop a more convenient form for eqs. (6.30) and (6.31) using the deviation variables. Assume that To and CA,O are the steady-state conditions for the CSTR and for given input conditions CAi,O' Ti,o, and Tc,o. Then from eqs. (6.28) and (6.29), we take

126

Analysis of the Dynamic Behavior of Chemical Processes

O 1 ( ) k -E/RTo

= - CA;O - CA.O - oe CA.O

r .

(6.32)

and

1 E/RT VAt (T T)

O=-(T;o- To) + Jkse: °CA.O--- 0- c.O

r . pc , V

Subtract eqs. (6.32) and (6.33) from (6.30) and (6.31), respectively, and take

(6.33)

dCA 1 )] k E -E/RTo (T T)

- = - [(CA· - CA·O) - (CA - CA.O - 0 --2 e CA.O - 0

dt r 1 I. RTo

(6.34)

dT 1 )] Jk [E -E/RTo (T T)

-=-[(T;- T;.o)-(T- To + 0 --2e CA.O - 0

~ r R~

(6.35)

Define the following deviation variables:

T; = T; - T;.o T~ = T; - Tc.o

Then eqs. (6.34) and (6.35) take the following form, in terms of the deviation variables:

T' = T - 1"0

dCA 1 (' ') [koE e-E/RToc ] T' _ [k e-E/RTo]c' (6.36)

-=- CA·-CA - -- AO 0 A

dt r 1 RT5 .

dT' =! (T; _ T') + JkO[(_!!_ e-E/RTOCA.O)T' + e-E/RTOCA]

dt r RT5

_ VAt (T' _ T~) (6.37)

pCpV

THINGS TO THINK ABOUT

l. What is computer simulation, and what is it used for?

2 Discuss the methods of interval halving, successive substitution, and New. ton-Raphson for solving nonlinear algebraic equations. What are their relative advantages and disadvantages?

3. Do the same with Euler's and the fourth-order Runge-Kutta integration methods.

4. What is linearization?

Part III

Chap. 6

Computer Simulation and the Linearization of Nonlinear Systems

127

5. Why are the linearized approximate models useful for process control purposes?

6. When is the linearized model more accurate-near or far from the point of linearization-and why?

7. What is the most attractive point of linearization for control purposes, and why?

8. Which linearization is more useful, the one around point A or the one around point B, and why? (See Figure Q6.1.)

f(x)

B

x

FigureQ6.1

9. What are the deviation variables? What is the point of linearization in order to define the deviation variables that will be useful for process control purposes?

10. Consider the tank system discussed in Example 6.1 (Figure 6.2a), where the flow rate of the outlet stream is proportional to the square root of the height of the liquid level. Show that we should relinearize the balance equation every time we change the desired liquid level at steady state.

11. The following differential equations provide the mathematical models for several processes. Which of them are linear and which nonlinear?

x dx + x = lOt + 5 dt

dXI dX2 d

Process II: a 1- + a2 - = m l(t) + I(t)

dt dt

Process I:

b dXI b dX2 d

1-+ 2 - = m2(t) + 2(t)

dt dt

dXI dX2

Process III: al- + XI - = ml(t)

dt dt

dXI dX2

bl-+ b2-=coswt

dt dt