Lecture 4 Response and Dynamics of Chemical Reactors and Instrumentation

N.T.U.A.
UGent Francqui Chair 2013 / 6-7th Lectures
Response and dynamics of chemical reactors and

instrumentation
Nikos Papayannakos, Professor

National Technical University of Athens
School of Chemical Engineering
Unit of Hydrocarbons and Biofuels Processing
10 April 2013
UGent/FCh13/6-7L Contents
Basics on Dynamics
First Order Systems
Thermocouple
Mixer
CSTReactor
Non-linear systems
Responses to standard stimuli
Second Order Systems
Non-interacting 1st-Order systems
Interacting 1st-Order systems
Generalized response of 2nd-Order systems
Response of Plug Flow Rectors
Time delay / Cin Changes – Q changes
TR/PR changes
Operation of combined CSTR – PFR networks
Conclusions
10 April 2013
UGent/FCh13/6-7L Introduction
Dynamics is related to changes with time
Dynamic equations describe the evolution of systems with time
Although the fashion in the field of dynamics is the study of complicated systems,
the study of simple systems, which in many cases compose the bigger systems,
provides useful knowledge to the chemical engineers’ background
The basis for building the dynamic equations in Chemical Engineering systems
Is the Mass and Energy balances
The dynamic equations and their solutions must be well understood

“physical meaning”
Not be considered simply as a way to ‘produce’ numbers describing
the evolution of the system
10 April 2013
UGent/FCh13/6-7L
First Order Dynamic systems
dy
τ + y = f(t) or A , Initial Conditions
dt
First order ordinary differential equations are often exactly

solvable by separation of variables
For I. C. : t = 0, y = y0
y(t) – A = (y0 - A)x( e-t/τ )
y(t) – y0 = (A - y0)x(1 – e-t/τ)
y(t) = A + (y0 - A)x(1 – e-t/τ)
10 April 2013
UGent/FCh13/6-7L 1st Order Systems
Thermocouple
Small and easy to use device for temperature measurements in process equipments
Principle of Operation Contemporary techniques

Transducer
emf
Analogue
Cold Or Digital
Junction Signal
Tc
T
T
emf = f(T-Tc)
Display T
• Electronic compensation
for the Joints Cold Point Tc
• Fast signal transmission from

measuring junction to display
10 April 2013
Simulation of the Dynamic operation of a Thermocouple
What will be the response

to a step Stimulus?
TB2 T TB
T
TB1
TB
t=0 T
l
Assumption to set up the Enegry balance
- The energy conveyed through the cables negligible
0 R
- Isothermal thermocouple bulb (Conduction>> Convection)
Final Equation
Energy balance
Heat flow into bulb = Accumulation mthxcpth dT
+ T = TB
Asphxhf dt
Asph x hf x (TB – T) x dt = mth x cpth x dT
τ = time constant
10 April 2013
Typical Response
Step Stimulus
A = TB2 t < 0, T = TB1 = 60 0C
t ≥ 0, TB2 = 80 0C
ys = TB1
y(t) = 80 - 20*(1 – e-t/4)

TB2 = 80 0C
t = 4 sec y(t) : T, 0C
t : sec
TB1=60 0C
10 April 2013
Q, CA0 Continuous Stirred Mixer

Mass Balance
Mass In-flow - Mass Out-flow = Accumulation
VR dCA
Q x CA0 - Q x CA = VR
Q, CA dt
VR dCA
Final Form : + CA = CA0
Q dt
τ = time constant = space time = τ*

For a Step Stimulus from CA01 ( t < 0) to CA02 (t ≥ 0) the analytical solution :
CA02
CA = CA02 + (CA01 - CA02 )( e-t/τ ) CA
CA01 τ
Time
10 April 2013
Q, CA0 CSTReactor
Reaction A  P
(-r) = [ k f(CA) ] = k CA
VR
Q, CA
Mass Balance
Mass In-flow - Mass Out-flow - Mass Reacting = Accumulation
dCA
Q x CA0 - Q x CA - (-r) x VR = VR
dt
VR dCA Q 1
Final Form : + CA = CA0 = CA0
Q + kVR dt Q + kVR 1 + kτ*
τ*
τ = time constant = 1 + kτ*
< τ*
Steady State Gain
10 April 2013
CIN,2
Steady
CIN,1 State 2
COUT,2
COUT
COUT,1
Time
Steady
State 1
COUT,1 COUT,2 COUT 1
At Steady State : = = =
CIN,1 CIN,2 CIN, 1 + kτ*
What is the reason of existence of
F(t)
Transition Period in Perfect Mixing?
The different Residence Times
Cumulative Distribution Function F(t) t/τ*
10 April 2013
Non-linear Systems
Pressure vessel with relief valve
K1 K2
PF POUT ( = 1 atm )
p, V
Input Valve Output Valve
Pressure
Vessel
Mass Balance
Mass In-flow - Mass Out-flow = Accumulation
dmGas V x MWGas dP
K1 x (PF – P)1/2 - K2 x (P – P0UT )1/2 = =
dt RT dt
C ≠τ
Non-linear Terms
At Steady State :
10 April 2013
Taylor Series approximation
C dP α
Final Form : + P = (P – P ) + P0
α + β dt α + β F F0
τ
10 April 2013
Responses to Standard Stimuli

1. Step Input - Stimulus
Dynamic Equation
r1
A = r1 – r0 dy
τ
r0 dt + y = A
I. C. : t = 0, y = 0
Solution of the Equation
• Method of separation of variables

y = yS + y t • Laplace method
• Method of Undetermined Coefficients
yS = Steady State or Particular Solution
yt = Transient or Complimentary Solution
10 April 2013
dyt
τ yt = B est
d( yS + yt ) dt + yt = 0
τ + yS + y t = A
dt
yS = A
y = yS + yt = B e-t/τ + Α
y – y0
y=A(1– e-t/τ) or = 1 – e-t/τ
y1 – y0
A
0.63(0.37A) - The system has no memory
- For any time period of τ, 63 %
0.63A of the remaining distance to the
new steady state is covered
0 - After t=3τ, 95% of the distance
0 τ 2τ 3τ
to new steady state (A) is covered
10 April 2013
2. Ramp/liner forcing
Dynamic Equation
r1 > r0
dy
r1 = r0 τ dt + y = Kt + r1
r1 < r0
I. C. : t = 0, y = y0 = r0
r0
r = r1 + Kt
Solution :
t=0 y = yS + y t
dyt
τ yt = B est
d( yS + yt ) dt + yt = 0
τ + yS + yt = Kt + r1
dt
yS = αt + β
y = (Kt + r1 – Kτ) + [(r0 – r1) + Kτ]e-t/τ
10 April 2013
General Response to a Ramp/liner forcing
Stimulus
(r1 < r0) y1

(r1 = r0) y2
(r1 > r0) y3
0 τ 2τ 3τ 4τ 5τ
- Dynamic Error Velocity Error

at Steady State : Κτ
- The Time lag at Steady State : τ
10 April 2013
3. Impulse Forcing and Response
r1
Response for 0 ≤ t ≤ Δt
yA y1 - r0
y – y0
= 1 – e-t/τ
y1 – y0
y0 = r0
or
Δt yA - y0 = (y1 - y0 )(1 – e-Δt/τ )
t=0
Response for t > Δt ( 1 – e-Δt/τ)Δt->0 = Δt/τ

y – yA
= 1 – e-t/τ yA = y0 + (y1 - y0 )Δt/τ
y0 – y A
I -t/τ
y = y0 + (y1 - y0 )(Δt/τ)e-t/τ = y0 + e Where I = (y1 - y0 )Δt
τ
- The response depends on the ‘total’ Input and not on the real form of the Impulse
10 April 2013
4. Sine Input
One of the most important stimuli that represents all the periodic
Inputs and its study gives useful deductions for system behavior
r = r0 + Asin(ωt) For y0 = r0 = 0 and

A No signal amplification
r0 = y0
dy
τ dt + y = Asin(ωt)
Cωτ cos(ωt + Ψ) + C sin( ωt + Ψ) = Α sin(ωt)

y = yS + y t
yS = Csin(ωt + Ψ) Ψ = arc tan (-ωτ) < 0 : Phase delay
yt = Βe-t/τ A
C = Β = C sin (arc tan (ωτ))
𝟏 + 𝝎𝟐𝝉𝟐
10 April 2013
A
y = sin (arc tan (ωτ)) e-t/τ + sin (ωt + arc tan (-ωτ)
𝟏 + 𝝎𝟐𝝉𝟐
A
For y0 ≠ 0 : y = y0 + sin (arc tan (ωτ)) e-t/τ + sin (ωt + arc tan (-ωτ)
𝟏 + 𝝎𝟐𝝉𝟐
C
Stimulus, Asin(ωt)
Response, y
yt
- The higher the stimulus frequency
y0 C
and
C The higher the System’s inertia
yS τ=1
The smaller the response amplitude
10 April 2013
Application of Stimulus/Response techniques

to study Ideal – Non-Ideal flow
CSTR
– Assumption : Total mixing CiF Q, CiF
- Step Tracer Input 0  CiF Tracer Input
Concentration
0
Mass balance :
VR
dCiE
Q CiF - Q CiE = VR dt Q, CiE
I. C. : t=0, CiE = 0
Plug Flow
From solution : F(t)
𝐶𝑖𝐸
= 1 − 𝑒-t/τ = 𝐹(𝑡)
𝐶𝑖𝐹
Cumulative residence time distribution =

% flow with Residence Time < t t=τ
VR/Q 10 April 2013
Dynamics of non-homogenuous CSTReactors
Berty Stationary Basket Catalyst Reactor Non-catalytic solid decomposition

Carberry Spinning Basket Catalyst Reactor
Free flowing Catalyst Reactor
Q, CiF Q, CiF
Vcat (Mcat) VSolid (MSolid)
VR VR
Q, CiE Q, CiE
VR τ : time constant VR
Q + k*VCat Measure of System’s Inertia Q + k** VSolid
K* = first or linearized-first order K** = first or linearized-first order

catalytic reaction rate constant decomposition rate constant
10 April 2013
UGent/FCh13/6-7L
Second Order Dynamic systems

System with two non-interacting 1st Order stages in series
Q, C0
𝑑𝐶1 𝑄
𝜏1
𝑑𝑡
+ C1 = 𝑦
𝑄+𝑘1 𝑉𝑅1 0
= K1y0
(A)
τ1 = VR1/(Q + k1VR1)
VR
Q, C1
-r = k1C
𝑑𝐶 𝑄
𝜏2 𝑑𝑡2 + C2 = 𝑦
𝑄+𝑘2 𝑉𝑅2 1
= K2y1
VR (B)
τ2 = VR2/(Q + k1VR2)
Q, C2
Ly2 K2
Laplace Transform : (A)  Ly1
= Ly2 K1 K2
1 + τ2s =
Ly1 K1 Ly0 1 + τ1s 1 + τ2s
(B)  =
Ly0 1 + τ1s
10 April 2013
UGent/FCh13/6-7L 2nd Order Systems
System with two interacting 1st Order stages in series
Sheath Junction
TL
TS
Sheath
TL0 = TS0 = TT0 TT
Steady State
TL
TT Assumptions :
TS
1. Negligible heat loss through cables and sheath top
2. Isothermal sheath walls
3. Isothermal bulb of the measuring junction
Resistances to Heat Flow : Thermal Accumulations :
1. From the fluid to the thermowell 1. In the sheath

o o dTS
qL = hL (TL – TS) qS = mScS
dt
2. From the sheath to the thermometer junction 2. In the thermometer head
o o
qV = hV (TS – TT) qT = mTcT dTT
dt 10 April 2013
Energy Balances :
o o o
1. Sheath : qL = qS + qV
o o
2. Thermometer Bulb : qV = qT
Final Equation :
𝑑2𝑇𝑇 𝑑𝑇
𝜏SτΤ 𝑑𝑡2 + (𝜏𝑆 + 𝜏Τ + τST) 𝑑𝑡𝑇 + 𝑇𝑇 = 𝑇𝐿
mScS mΤcΤ mTcT

τS = τΤ = τST =
hL hV hL
𝑑𝑇𝑇
With Initial Conditions : at t = 0 : TT = 0, =0
𝑑𝑡
and Laplace transform
𝐿𝑇𝑇 1
G(s) = =
𝐿𝑇𝐿 1 + (τ + τ + τ )s + τ τ s2
S Τ ST S Τ
10 April 2013
𝐿𝑇𝑇 1
=
𝐿𝑇𝐿 (1 + τ1s) (1 + τ2s)
Where :
τ1 τ2 = τS τT
τ1 + τ2 = τS + τT + τST
τ1 and τ2 are the ‘apparent’ time constants of the system
• Both systems composed of two first order systems in series

either interacting or non-interacting
are described by the same Laplace equation
• For the non-interacting system the two time constants are those
corresponding to each first order system
• For the interacting system the time constants are combinations
of the characteristic parameters of the system
10 April 2013
Response of systems represented by two 1st Order systems is series
𝐿𝑦 K1K2
=
𝐿𝑦0 (1 + τ1s) (1 + τ2s)
1/τ2 1
Ly = MK1K2 (s +1/τ1
1/τ1) (s + 1/τ2) s
1st Case τ1 ≠ τ2
A B
Ly = MK1K2 + + C
(s + 1/τ1) (s + 1/τ2) s
τ1
y = H[1+ e-t/τ1 - τ2 e-t/τ2 ] H = MK1K2
τ2 - τ1 τ2 - τ1
10 April 2013
Properties of a 2nd Order response curve
1
Experimental Curve τ1 τ2 ln(τ1/τ2)
tx =
Ψ = y/H
τ2 - τ1
dΨ
Ψx d2Ψ Ψx = 1 + (τ1 + τ2)
Inflection point =0 dt tx
dt2 tx
0
tx t
Calculate τ1, τ2
2nd Case τ1 = τ2 = τ
1/τ 1
Ly = MK1K2 (s 1/τ
+ 1/τ) (s + 1/τ) s
Ψx = 0.265
y = H ( 1 – (1+t/τ)e-t/τ ) tx = τ
10 April 2013
General Second Order Dynamic System

The General Form of 2nd Order Systems is :
𝑑2𝑦 𝑑𝑦
A 𝑑𝑡2 + 𝐵 𝑑𝑡 + 𝑦 = f(t)
And for a step stimulus it can be transformed to :

1 𝑑2𝑦 𝑑𝑦 1 𝐵
+ 2𝑧𝑛𝑐 + 𝑐2𝑦 = 𝑐2𝐻 c= zn =
𝑐 𝑑𝑡
2 2 𝑑𝑡 𝐴 2 𝐴
With Laplace transform :
Hc2
Ly = With roots : s1,2 = -znc ± c 𝑧𝑛2 − 1
s(s2 + 2zncs + c2)
1. zn > 1 : s1,2 real numbers corresponding to two 1st order systems in series
2. zn = 1 : Two equal roots. As previously described
10 April 2013
3. zn < 1 : In this case the response to a step stimulus will be

1
y= H[1+ 𝑒 −𝑧𝑛𝑐𝑡 𝑠𝑖𝑛(𝑐𝑡 1 − 𝑧𝑛2 − ϕ)]
1−𝑧𝑛2
ϕ = 𝑠𝑖𝑛−1 1 − 𝑧𝑛2
3.1 Special case : zn = 0
y = H(1 – cos(znt))
zn = 0
zn > 1 Over-damped behavior
zn = 0.2 zn = 1 Critically damped behavior
zn = 1 0<zn <1 Under-damped behavior
zn = 2
zn = 0 Undamped behavior
Sustained Oscilation
10 April 2013
UGent/FCh13/6-7L Plug Flow Reactors
Transfer Delay in Plug Flow Reactors
By definition of the ideal situation of Plug Flow all the molecules/masses

will have the same Residence Time inside the reactor
Q Q
CIn Cout
T, P
The elementary masses entering the reactor can be considered as batch reactors
operating for a time period equal to Residence Time.
𝑉
Residence Time Empty Tubes – Homogeneous reactions t = τ = 𝑄𝑅
Tubes filed with solids (catalyst – inerts)
𝜀𝑉
One fluid phase t = τ = 𝑄𝑅
ℎ 𝜀 𝑉
Two fluid phases t=τ= 𝑄 𝑅
10 April 2013
Possible changes during Operation : Cin, Q, T/P
CIN Change
Response at the exit delays for t = τ

Cout(t) = K Cin(t – τ) Cin
τ
Cout
Κ : Attenuation/Amplification Constant
For 1st Order Reaction : K = 1-exp(-k1τ) τ
t=0
Q Change
Residence time will vary according to the Q change

Response at the exit has no-delay
τ
Q
f(t) is a function of reaction kinetics
and residence time change Cout C=f(t)
t=0 τ
10 April 2013
T/P Change
Residence time is constant
Response at the exit has no-delay

τ
f(t) is a function of reaction kinetics T/P
and residence time change Cout C=f(t)
t=0 τ
For a step change of T/P
During the transition period ( 0<t<τ ) the exit concentration will be a combination
of the initial period before the stimulus and the second period after the stimulus.
10 April 2013
Systems with CSTR and PFR

VR2
Q2 Cin
Q
Q
Cout Q1 Cout
VR1 τ1
τ2
Q2 t=0
Q
Q1 Q Cin
Cout
t=0
Q2
Q
Q Cin
Q1 Cout
VR1 τ1
t=0 10 April 2013
UGent/FCh13/6-7L
CONCLUSIONS
First Order Systems
1. The way of imposing an Impulse Stimulus does not affect the Response
2. A continuously Changing stimulus results in response Delay and Offset
3. For a Periodic Stimulus an Attenuation of the amplitude is observed with
input frequency. The response frequency is the same as the stimulus frequency
Second Order Systems

4. For Non-Interacting two first order systems in series the system’s time
constants are those of the two 1st Order systems
5. A sigmoid response curve can be simulated by a second order system with time
constants calculated from the time and the response of the inflection point
Plug flow reactors

6. The response of a plug flow reactor depends on the Stimulus.
Changes in input concentration result in response delays equal to R.T. (τ)
Changes in feed flow rate or operating conditions result in response without delay
10 April 2013

Lecture 4 Response and Dynamics of Chemical Reactors and Instrumentation

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Lecture 4 Response and Dynamics of Chemical Reactors and Instrumentation

Hochgeladen von

Copyright:

Verfügbare Formate

N.T.U.A.

UGent Francqui Chair 2013 / 6-7th Lectures

Response and dynamics of chemical reactors and

Nikos Papayannakos, Professor

Dynamics is related to changes with time

Dynamic equations describe the evolution of systems with time

The dynamic equations and their solutions must be well understood

First Order Dynamic systems

First order ordinary differential equations are often exactly

y(t) – A = (y0 - A)x( e-t/τ )

y(t) – y0 = (A - y0)x(1 – e-t/τ)

y(t) = A + (y0 - A)x(1 – e-t/τ)

Principle of Operation Contemporary techniques

• Fast signal transmission from

Simulation of the Dynamic operation of a Thermocouple

What will be the response

y(t) = 80 - 20*(1 – e-t/4)

Q, CA0 Continuous Stirred Mixer

τ = time constant = space time = τ*

What is the reason of existence of

Taylor Series approximation

Responses to Standard Stimuli

Solution of the Equation

• Method of separation of variables

y = (Kt + r1 – Kτ) + [(r0 – r1) + Kτ]e-t/τ

General Response to a Ramp/liner forcing

(r1 < r0) y1

(r1 > r0) y3

- Dynamic Error Velocity Error

- The Time lag at Steady State : τ

3. Impulse Forcing and Response

Response for t > Δt ( 1 – e-Δt/τ)Δt->0 = Δt/τ

r = r0 + Asin(ωt) For y0 = r0 = 0 and

Cωτ cos(ωt + Ψ) + C sin( ωt + Ψ) = Α sin(ωt)

yS = Csin(ωt + Ψ) Ψ = arc tan (-ωτ) < 0 : Phase delay

Application of Stimulus/Response techniques

From solution : F(t)

Cumulative residence time distribution =

Dynamics of non-homogenuous CSTReactors

Berty Stationary Basket Catalyst Reactor Non-catalytic solid decomposition

Vcat (Mcat) VSolid (MSolid)

K* = first or linearized-first order K** = first or linearized-first order

Second Order Dynamic systems

System with two interacting 1st Order stages in series

Resistances to Heat Flow : Thermal Accumulations :

1. From the fluid to the thermowell 1. In the sheath

mScS mΤcΤ mTcT

τ1 and τ2 are the ‘apparent’ time constants of the system

• Both systems composed of two first order systems in series

Response of systems represented by two 1st Order systems is series

Properties of a 2nd Order response curve

General Second Order Dynamic System

And for a step stimulus it can be transformed to :

With Laplace transform :

3. zn < 1 : In this case the response to a step stimulus will be

3.1 Special case : zn = 0

zn = 1 0<zn <1 Under-damped behavior

Transfer Delay in Plug Flow Reactors

By definition of the ideal situation of Plug Flow all the molecules/masses

Possible changes during Operation : Cin, Q, T/P

Response at the exit delays for t = τ

Residence time will vary according to the Q change

Residence time is constant

Response at the exit has no-delay