ChE 441 Process Control Fall, 2008 Problem 3-27, Smith and Corripio

Washington State University Department of Chemical Engineering Richard L. Zollars

Consider the chemical reactor shown in Fig. P3-16. In this reactor, an endothermic reaction of the type A + 2B C takes place. The rate of appearance of A, is given by

rA (t ) = k 0 e E / RT ( t ) c A (t ) c B (t )

where
rA(t) k0 E R T(t) cA (t) cB (t) Hr

= = = = = = = =

rate of disappearance of A, kmole of A/(m3-s) frequency factor (constant) m3/(kmole-s) energy of activation (constant) cal/gmole gas law constant, 1.987 cal/(gmole-K) temperature in reactor, K concentration of A in reactor, kmole/m3 concentration of B in reactor, kmole/m3 heat of reaction, J/kmole

The heat input to the reactor is related to the signal to the heater by the following expression
q (t ) = r m1 (t )

where
q(t) r

= =

heat input to reactor, J/s constant

The flow of pure B through the valve is given by

f 2 (t ) = C v 2 vp 2 (t ) P2 G2

where
Cv2 P2 G2 vp2 (t)

= = = =

valve coefficient (constant) m3/(s-psi1/2) pressure drop across valve (constant) psi specific gravity of B (constant) dimensionless valve position, a fraction

You may assume that the reactor is well insulated, and that the physical properties of the reactants and products are similar. The flow rate f1 can be assumed to be constant. The valve position vp2 (t) is linearly related to the signal m2 (t). Develop the mathematical model that describes the interactions between the input variables m1 (t), m2 (t), cAi (t), and the outlet temperature T (t); determine the transfer functions; and draw the block diagram. Show the units of all gains and time constants. SOLUTION We will need three balances to describe this process; a balance on A, a balance on B and an energy balance. These would be (in order)
d cA = f1 c A1 (t ) ( f1 + f 2 (t )) c A (t ) k o e E / RT ( t ) c A (t )c B (t ) V dt d cB = f 2 (t ) ( f 1 + f 2 (t )) c B (t ) 2 k o e E / RT ( t ) c A (t )c B (t ) V V dt dT V cp = f1 c p T1 (t ) + f 2 (t ) c p T2 ( f1 + f 2 (t )) cv T (t ) + q (t ) H r k o e E / RT ( t ) c A (t )c B (t ) V dt V

Rearranging these equations gives

d cA = f1 c A1 (t ) f1c A (t ) f 2 (t ) c A (t ) k o e E / RT ( t ) c A (t )c B (t ) V dt d cB V = f 2 (t ) f 1c B (t ) f 2 (t ) c B (t ) 2 k o e E / RT ( t ) c A (t )c B (t ) V dt dT V cp = f 1 c p T1 (t ) + f 2 (t ) c p T2 f 1 cv T (t ) f 2 (t ) cv T (t ) + q (t ) dt H r k o e E / RT ( t ) c A (t )c B (t ) V V

There are a number of nonlinear terms in these equations. Doing a Taylors series expansion to get linear approximations gives

f 2 (t ) c A (t ) = f 2 c A + ( f 2 f 2 ) c A + (c A c A ) f 2 f 2 (t ) c B (t ) = f 2 c B + ( f 2 f 2 ) c B + (c B c B ) f 2 f 2 (t ) T (t ) = f 2 T + ( f 2 f 2 ) T + (T T ) f 2 e E / RT ( t ) c A (t ) c B (t ) = e E / RT c A c B + (T T ) + (c B c B ) e
E / RT

E RT cA
2

e E / R T c A c B + (c A c A ) e E / R T c B

We will also need to introduce deviation variables for all of the time dependent functions. These are
T1* = T1 T1 c* A1 = c A1 c A1 f 2* = f 2 f 2 T* =T T q* = q q c* A = cA cA
* = cB cB cB

Now substitute the linear approximations and the deviation variables into the material and energy balances from above to get
d c* E / RT * * * A c A cB = f1 c * A1 + c A1 f 1 c A + c A f 2 c A f 2 c A c A f 2 k 0Ve dt E E / RT E / RT * e c A cB c* cB cB k 0Ve E / RT c A T * k 0V A k 0Ve 2 RT * d cB * * V f 2 2 k 0Ve E / RT c A c B = f 2* + f 2 f 1 c B + c B f 2 c B f 2* c B c B dt E E / RT E / RT * e c A cB 2 c* cB 2 cB k 0Ve E / RT c A 2 T * k 0V A k 0Ve 2 RT * dT V cp = f1 c p T1* + T1 + f 2* + f 2 c p T2 f1 c v T * + T cv f 2 T + f 2* T + T * f 2 dt E / RT E E / RT E / RT * E / RT c A cB + T * e c A cB + c* cB + cB e cA + q * + q H r k oV A e 2 e RT V

) (

At the prior steady-state these equations would give

0 = f 1 c A1 f1 c A f 2 c A k 0Ve E / RT c A c B 0 = f 2 f 1 c B f 2 c B 2 k 0Ve E / RT c A c B 0 = f 1 c p T1 + f 2 c p T2 f1 cv T cv f 2 T + q H r k oVe E / RT c A c B

Subtracting this from the equations above gives

V d c* E E / RT * * * * A = f1 c * e c A cB A1 f 1 c A f 2 c A c A f 2 T k 0V 2 dt RT
* E / RT c* cB cB k 0Ve E / RT c A A k 0Ve * d cB E E / RT * * = f 2* f1c B f 2* c B c B e c A cB f 2 2 T * k 0V 2 dt RT * E / RT 2 c* cB 2 cB k 0Ve E / RT c A A k 0Ve

V cp

dT* dt

= f1 c p T1* + f 2* c p T2 f 1 cv T * cv f 2* T cv T * f 2 + q * H r k oV T * E RT
2 E / RT * E / RT e E / RT c A c B H r k oVc * c B H r k oVc B e cA A e

The equation relating the flow of B through the valve and the valve position can be put into deviation variable form
* f 2* + f 2 = vp 2 + vp 2 C v 2

P2 G2

At the prior steady-state this gives

f 2 = vp 2 C v 2 P2 G2

Subtracting this from the equation above it gives

* f 2* = vp 2 Cv2

P2 G2

The relationship between the valve position (vp2) and the signal m2 is also given as linear. Thus we have vp 2 = K v m2 in deviation variables as well as normal variables. Replace vp2 in the expression above with this relationship to get

* f 2* = m2 K v Cv2

P2 G2

Similarly the expression for the rate of heat addition can be put into deviation variables to get
* q * = r m1

Substitute these last two expressions into all of the balance equations to get
P2 d c* E E / RT * * * A c A c* e c A cB = f1 c * V A1 f 1 c A m 2 K v C v 2 A f 2 T k 0V 2 dt G2 RT
* E / RT cB cB k 0Ve E / RT c A c* A k 0Ve * d cB P2 P2 E E / RT * * * * V K v Cv 2 K v Cv2 cB cB f 2 2 T * k 0V e c A cB = m2 m2 f 1c B 2 dt G2 G2 RT * E / RT cB cB k 0Ve E / RT c A 2 c* A k 0Ve

V cp

dT* dt

* K v Cv 2 = f1 c p T1* + m2

P2 P2 * K v Cv 2 c p T2 f1 cv T * m2 cv T cvT * f 2 G2 G2 E
2 E / RT * E / RT e E / RT c A c B H r k oVc * c B H r k oVc B e cA A e

* + r m1 H r k oV T *

RT

Rearranging these equations gives V d c* P2 E E / RT * * A c A m2 e c A c BT * + f1 + f 2 + k 0Ve E / RT c B c * k 0V A = f 1 c A1 K v C v 2 2 dt G2 RT

* k 0Ve E / RT c A c B * d cB P2 E E / RT * * e c A cB T * c B m2 + f1 + f 2 + 2 k 0Ve E / RT c A c B = K v Cv 2 2 k 0V 2 dt G2 RT

( (

2 k 0Ve E / RT c B c * A V dT* * H r k oV E E / RT P2 * * e c c T2 T m2 + f1 + f 2 + A B T = f 1 T1 + K v C v 2 2 dt c G 2 p RT H r k oV E / RT H r k oV E / RT r * * m1 e cB c* e c A cB + A cp cp cp

These three equations give the mathematical relationship between T* and the variables requested (although cA and cB cannot be separated at this point). If we take the LaPlace transforms of each of these equations we get

[Vs + ( f [Vs + ( f

+ f 2 + k 0Ve E / RT c B C A = f1 C A1 K v C v 2

)]

P2 E E / RT c A M 2 k 0V e c A c BT 2 G2 RT

k 0Ve E / RT c A C B
1

+ f 2 + 2 k 0Ve E / RT c A C B = K v C v 2

)]

P2 E E / RT c B M 2 2 k 0V e c A cB T 2 G2 RT

2 k 0Ve E / RT c B C A H r k oV E E / RT T = f 1 T1 + K v C v 2 P2 T2 T M 2 f1 + f 2 + e c c Vs + A B 2 c p RT G2

H r k oV E / RT H r k oV E / RT r M1 e cB C A e cA CB cp cp cp

where the variables in capital letters are the respective functions in the LaPlace domain. Putting these into the standard form for a first order system gives the set of equations shown below. Note that the time constants and gains are different in each equation.

 f1 V s C A1 M2 1 + CA = E / RT cB f1 + f 2 + k 0Ve E / RT c B f 1 + f 2 + k 0Ve E / RT c B f1 + f 2 + k 0Ve k 0Ve E / RT c A R T T CB f 1 + f 2 + k 0Ve E / RT c B f 1 + f 2 + k 0Ve E / RT c B

2

K v Cv 2

P2 cA G2

k 0V

e E / RT c A c B

E E / RT 2 k 0V e c A cB 2 V R T s + 1 C B = M2 T E / RT E / RT E / RT f f k Ve c f f k Ve c 2 2 2 f f k Ve c + + + + + + A A A 2 0 1 2 0 1 2 0 1 K v Cv 2 2 k 0Ve E / RT c B f 1 + f 2 + 2 k 0Ve E / RT c A CA

P2 cB G2

f1 V s + 1 T = T1 H r k oV E E / RT H r k oV E E / RT e c A cB f1 + f 2 + e c A cB f1 + f 2 + c 2 c p RT 2 p RT P2 T2 T G2 M2 + H r k oV E E / RT f1 + f 2 + e c A cB c p RT 2 K v Cv2 r cp M1 + H r k oV E E / RT f1 + f 2 + e c A cB c p RT 2 H r k oV E / RT e cB cp CA H r k oV E E / RT f1 + f 2 + e c A cB c p RT 2 H r k oV E / RT e cA cp CB H r k oV E E / RT f1 + f 2 + e c A cB c p RT 2

Using the subscripts 1, 2, and 3 to designate the first, second and third equations, respectively gives the set of equations shown on the subsequent page.

[ 1 s + 1]C A = K 1, A1 C A1 K 1.M 2 M 2 K 1,T T K 1,CB C B [ 2 s + 1]C B = K 2,M 2 M 2 K 2,T T K 2,CA C A [ 3 s + 1]T = K 3,T 1T1 + K 3,M 2 M 2 + K 3,M 1 M 1 K 3,CA C A
In these equations the time constants are given by

K 3,CB C B

1 = 2 = 3 =

V f1 + f 2 + k oVe E / RT c B V f1 + f 2 + 2 k oVe E / RT c A

V H r k oV E E / RT f1 + f 2 + e c A cB c p RT 2 The gains in the first equation are given by

K 1, A1 =

f1 f 1 + f 2 + k 0Ve E / RT c B K v Cv 2 P2 cA G2

K 1.M 2 =

f 1 + f 2 + k 0Ve E / RT c B k 0V E
2

K 1.T = K 1.CB =

RT f 1 + f 2 + k 0Ve E / RT c B k 0Ve E / RT c A f1 + f 2 + k 0Ve E / RT c B

e E / RT c A c B

The gains in the second equation are given by

K v Cv 2 K 1.M 2 = P2 cB G2

f1 + f 2 + 2 k 0Ve E / RT c A RT f1 + f 2 + 2 k 0Ve E / RT c A 2 k 0Ve E / RT c B f1 + f 2 + 2 k 0Ve E / RT c A 2 k 0V E

2

e E / RT c A c B

K 2,T = K 2,CA =

And the gains in the third equation are given by

K 3,T 1 =

f1 H r k oV E E / RT e c A cB f1 + f 2 + c p RT 2 K v Cv 2

K 3, M 2

P2 T2 T G2 = H r k oV E E / RT e c A cB f1 + f 2 + c p RT 2 r cp = H r k oV E E / RT e c A cB f1 + f 2 + c p RT 2 H r k oV E / RT e cB cp = H r k oV E E / RT e c A cB f1 + f 2 + c p RT 2 H r k oV E / RT e cA cp = H r k oV E E / RT e c A cB f1 + f 2 + c p RT 2

K 3, M 1

K 3,CA

K 3,CB

To get T in terms of the functions that we can control we need to solve to get an expression for CA (s) from above. This gives
CA =

[ 1 s + 1]

K 1, A1

C A1

K 1,T K 1,CB K 1. M 2 M2 T C [ 1 s + 1] [ 1 s + 1] [ 1 s + 1] B

Substitute this into the equation for CB (s) to get

[ 2 s + 1]C B

K 1,CB K 1,T K 1, A1 K 1.M 2 = K 2, M 2 M 2 K 2,T T K 2,CA [ s + 1] C A1 [ s + 1] M 2 [ s + 1] T [ s + 1] C B 1 1 1 1

Solve this for CB to get

K 1, A1 K 2,CA K 1,T K 2,CA K 1.M 2 K 2,CA K 1,CB K 2,CA T [ s + 1] C A1 M2 K 2,T [ s + 1] C B = K 2, M 2 + [ s + 1] 2 s + 1 [ s + 1] 1 1 1 1

Solving this for CB (s) gives

K1.M 2 K 2,CA K1,T K 2,CA K1, A1 K 2,CA K 2,M 2 + [ s + 1] K 2,T [ s +1] [ 1 s +1] 1 1 M T CB = C A1 2 K1,CB K 2,CA K1,CB K 2,CA K1,CB K 2,CA 2 s +1 [ s + 1] 2 s +1 [ s +1] 2 s + 1 [ s + 1] 1 1 1

This result can be substituting back into the expression for CA (s) to get
CA =

[ 1 s +1]

K1, A1

C A1

K1,T K1.M 2 M2 T [ 1s +1] [ 1 s +1]

K K K K K1, A1 K 2,CA K 2,M 2 + 1.M 2 2,CA K 2,T 1,T 2,CA K1,CB [ 1s +1] [ 1s +1] [ 1 s +1] M2 T C A1 K1,CB K 2,CA K1,CB K 2,CA K1,CB K 2,CA [ 1 s +1] s s s + 1 1 + + 1 2 2 2 s s s + 1 1 1 + + [ ] [ ] [ ] 1 1 1 K1,CB K1.M 2 K 2,CA K1, A1 K1,CB K 2,CA + K 2 , 2 M 2 K + + s s [ ] [ ] 1 1 K [ ] s 1 + 1 , 1 A 1 1 M 1. M 2 1 + CA = C A1 + [ 1 s + 1] 2 [ 1 s + 1] K1,CB K 2,CA K1,CB K 2,CA + + s s 1 1 2 2 [ 1 s +1] [ 1 s +1] K1,CB K1,T K 2,CA K 2 , T K + + 1 1 s s [ ] [ ] 1,T 1 1 T [ 1 s +1] K K 1,CB 2 ,CA + 1 s 2 [ 1 s +1]

The equations for CA (s) and CB (s) then can be substituted into the equation for T (s) to get

[ 3 s + 1]T =

K 3,T 1T1 + K 3, M 2 M 2 + K 3, M 1 M 1

K 1, A1 K 1,CB K 2,CA K 1,CB K 1.M 2 K 2,CA K 2, M 2 + 2 [ 1 s + 1] [ 1 s + 1] K 1, A1 + C K 1.M 2 + [ 1 s + 1] M2 A1 [ s + 1] [ 1 s + 1] K 1,CB K 2,CA K 1,CB K 2,CA 1 s 1 1 s + + 2 2 s 1 1 [ ] [ ] s + + 1 1 K 3,CA K 1,CB K 1,T K 2,CA K 2,T K 1,T [ s + 1] [ 1 s + 1] T 1 [ 1 s + 1] K 1,CB K 2,CA 2 s + 1 [ s + 1] 1 K 1.M 2 K 2,CA K 1,T K 2,CA K 1, A1 K 2,CA K 2,T [ s + 1] K 2, M 2 + [ s + 1] [ 1 s + 1] 1 1 M2 T C A1 K 3,CB K K K K K K 2 s + 1 1,CB 2,CA 2 s + 1 1,CB 2,CA 2 s + 1 1,CB 2,CA [ 1 s + 1] [ 1 s + 1] [ 1 s + 1]

Collecting all of the similar terms gives

K 1,T K 2,CA K 1,CB K T 2 , K + + 1 1 [ ] [ ] s s T 1 , 1 1 K s + 1 K 3,CA 3,CB 3 K 1,CB K 2,CA [ 1 s + 1] + 1 s 2 [ 1 s + 1] [K 3,T 1 ]T1 + [K 3,M 1 ]M 1 K 1,T K 2,CA K , T 2 [ 1 s + 1] T = K 1,CB K 2,CA 1 + s 2 [ ] 1 + s 1 K 1.M 2 K 2,CA + K 2, M 2 [ 1 s + 1] M 2 K 1,CB K 2,CA 2 s + 1 [ s + 1] 1

K 1.M 2 K 2,CA K 1,CB K 2, M 2 + [ 1 s + 1] [ 1 s + 1] K 1.M 2 K + K 3, M 2 + K 3,CA + 3,CB K 1,CB K 2,CA 1 s + 1] [ 2 s + 1 [ s + 1] 1 K 1, A1 K 1,CB K 2,CA [ 1 s + 1]2 K 1, A1 K 3,CA + K 3,CB K K [ 1 s + 1] 2 s + 1 1,CB 2,CA [ 1 s + 1] K 1, A1 K 2,CA

[ 1 s + 1] C A1 K 1,CB K 2,CA 2 s + 1 [ s + 1] 1

This would give the following transfer functions for T depending upon the various inputs.

GT 1 =

GM 1

K 1,T K 2,CA K 1,T K 2,CA K 1,CB K K 2 , 2 , T T [ 1 s + 1] [ ] [ ] + 1 1 + s s K 1,T 1 1 K 3 s + 1 K 3,CA 3,CB K K K K [ ] + 1 s 2 ,CA 1,CB 2 ,CA 1,CB 1 + 1 1 + s s 2 2 + 1 1 [ ] [ + ] s s 1 1 K 3, M 1 T ( s) = = M 1 ( s) K 1,T K 2,CA K 1,T K 2,CA K 1,CB K 2,T K 2,T [ 1 s + 1] [ 1 s + 1] [ 1 s + 1] K 1,T 3 s + 1 K 3,CA K 3,CB K K K K [ 1 s + 1] 2 s + 1 1,CB 2,CA 2 s + 1 1,CB 2,CA [ 1 s + 1] [ 1 s + 1]

T ( s) = T1 ( s )

K 3,T 1

K 1.M 2 K 2,CA K 1.M 2 K 2,CA K 1,CB K 2, M 2 + K 2, M 2 + [ 1 s + 1] [ 1 s + 1] [ 1 s + 1] K 1.M 2 + K 3, M 2 + K 3,CA K 3,CB K 1,CB K 2,CA K 1,CB K 2,CA [ 1 s + 1] 1 1 + + s s 2 2 1 1 + + [ ] [ ] s s T (s) 1 1 = GM 2 = M 2 ( s) K 1,T K 2,CA K 1,T K 2,CA K 1,CB K 2,T K 2,T [ 1 s + 1] [ 1 s + 1] [ 1 s + 1] K 1,T 3 s + 1 K 3,CA K 3,CB K K K K [ 1 s + 1] 2 s + 1 1,CB 2,CA 2 s + 1 1,CB 2,CA [ 1 s + 1] [ 1 s + 1] K 1, A1 K 1,CB K 2,CA K 1, A1 K 2,CA 2 [ 1 s + 1] [ 1 s + 1] K 1, A1 K 3,CA + K 3,CB K 1,CB K 2,CA K 1,CB K 2,CA [ 1 s + 1] + + 1 1 s s 2 2 [ ] [ ] + + 1 1 s s T ( s) 1 1 = GcA1 = c A1 ( s ) K 1,T K 2,CA K 1,T K 2,CA K 1,CB K 2,T K 2,T [ 1 s + 1] [ 1 s + 1] [ 1 s + 1] K 1,T 3 s + 1 K 3,CA K 3,CB K K K K [ 1 s + 1] 2 s + 1 1,CB 2,CA 2 s + 1 1,CB 2,CA [ 1 s + 1] [ 1 s + 1] The block diagram for this process would look like the following.

m2

vp2

Cv2

P2 G2

f2

Gpf2

m1

GM1 +

+ cA1 GcA1 + +

T1

GT1

The gains will have the following units K T 1 dimensionless K psig K KM2 psig K K cA1 kmol / m 3 K M1 All time constants will have units of seconds.