Model Non-Ideal Flow Reactors

Mass Transport: Non-Ideal Flow Reactors
Advanced Transport Phenomena

Module 6 - Lecture 28
Dr. R. Nagarajan
Professor
Dept of Chemical Engineering
IIT Madras
1
MODELING OF NONIDEAL-FLOW REACTORS
 Simplest approach: apply overall material/ energy/

momentum balances to the reactor
 “black box’ approach, insufficient
 Most rigorous: Divide into small subregions, approximate

each region with PDEs
 Impractical
 Intermediate solution: model as discrete network of small

number of interconnected ideal reactor types (SS PFR &
WSR)
2
 RTDF  residence time distribution function (exit-age

DF), E(t)
 E(t) dt  fraction of material at vessel outlet stream that
has been in vessel for times between t and t ± dt
 PFR: E(t) is a Dirac function, centered at residence time
V / m /  
3
 V  vessel volume
 m  feed mass flow rate
 e.g., straight tube through which incompressible fluid
flows with a uniform plug-flow velocity profile
 Partial recycle can alter RTDF
4
Tracer residence-time distribution functions for ideal and real vessels (for e.g.,
reactors) (adapted from Levenspiel (1972))
5
Ideal plug-flow reactor (PFR) with partial “recycle” (recycle introduces a

distribution of residence times, and reduces the residence time per pass within the
PFR)
6
 WSR:
E WSR   dF / dt   t flow  exp t / t flow 

1
 Most likely residence time in a WSR is zero!
 Mean residence time = V / m /  

 Not all fluid parcels have same residence time, unlike
PFR
7
 WSR:
 Dimensionless variance s2 about mean residence time
 indicator of spread of residence times
 Mean residence time related to first moment of E(t), i.e.:


t flow   t.E (t ) dt
0
s2 is related to 2nd moment of E(t):

 

 t  t 
1 1
.E (t )dt  2   t E  t  dt  t flow 
2
 
2
2 flow
2 2
t flow 0 t flow  0 
 = 1 for a WSR, 0 for a PFR
 PFR with infinite recycle behaves like WSR 8
 RTDF for Composite Systems:
 If RTDF for vessel 1 is E1(t) and for vessel 2 is E2(t),
RTDF for a series combination of the two is:

t
E1 2 (t )   E1  t  .E2  t  t  dt
' ' '
(convolution formula)
9
 If vessel 1 is characterized by tflow,1, and s12, and vessel

2 by tflow,2, and s22, then for the series combination,
mean residence times and variances are simply
additive:
t flow,12  t flow,1  t flow,2
 2
1 2   
2
1
2
2
10
 RTDF for Composite Systems:
 For a network of n-WSRs of equal volume, for which:
1 n 1
 t 
t flow  t 
E (n  WSRs )  .  .exp   

 n  1!  t flow   t flow 
(tflow  V / ( m /  ) ) for each vessel in series)
11
For vessels 1, 2, 3,…., n in parallel, receiving fractions f1,

f2, f3, …., fn of total flow:
E  f1E1 (t )  f 2 E2 (t )  ...  f n En (t )
Where  i
fi  1 , and for each vessel:
 E  t  dt  1
0
i (i  1, 2,..., n)
12
 Real reactors as a network of ideal reactors: Modular
modeling
 Network of ideal reactors can be constructed to

approximate any experimental reactor RTDF:
 
 
 tracer (t ) 
Eexp (t )  
 
  tracer  t  dt 
0  reactor exit
(where tracer is input as a Dirac impulse function)
13
Real reactors as a network of ideal reactors: Modular

modeling
GT combustor; proposed interconnection of reactors comprising “modular”

model (adapted from Swithenbank, et al.(1973))
14

modeling
 Info obtained from tracer diagnostics & from

combustor geometry, cold-flow data, etc.
 Important since RTD-data alone cannot discriminate

between alternative networks with identical RTD-
moments
 
t flow   tE  t  dt ,  t E  t  dt , ...,
2
etc.)
0 0
15
 Equivalent vessel network is nonunique
 Each alternative may capture one aspect (e.g.,
combustor efficiency) but not another (e.g., domain

of stable operation)
16

modeling
 Tracer methods can:
 Guide development of “modular” models
 Diagnose operating problems with existing chemical

reactors or physical contactors
 RTD data can show up dead-volumes, flow-
channeling, bypassing (all cause inefficient
operation)
 Geometric or fluid-dynamic changes in design can
correct these flaws
 Perturbation in feed can be used as “tracer”
17

modeling
 RTD function, E(t), does not capture role of
concentration fluctuations due to turbulence,
incomplete mixing (at molecular level– “micromixing”)
 When tracer concentration fluctuates at reactor exit,
we only collect data on <E(t)>  arithmetic average of
N tracer shots, each yielding RTD Ej(t) (j = 1, 2, …., N)
18
 Two networks with identical <E(t)> but with different
shot-to-shot variations, as measured by variance:

N 
1
   E  t  
2
lim j E (t )  dt
N  N j 1 0
will perform differently as chemical reactors
19
 Statistical micro flow (Random Eddy Surface-Renewal)
models of interfacial mass transport in turbulent flow
systems
 Mass/ energy transport visualized to occur during
intervals of contact between turbulent eddies & surface
 “stale” eddies replaced by fresh ones
 Effective transport coefficient calculated by time-

averaging RTDF-weighted instantaneous St(t)
20
 Statistical micro flow (Random Eddy Surface-Renewal)

systems
 If E(t) is defined such that:
Relative portion of each unit interfacial area


E (t )dt  covered by fluid eddies having "ages" between
 t and t+dt,

21
then: 
St   St (t ).E (t )dt
0
St(t)  calculated from transient micro fluid-dynamical

analysis of individual eddy flow
St  time-averaged transfer coefficient
 Interfacial region being viewed as a thin vessel w.r.t
eddy residence time
22
 Statistical microflow (Random Eddy Surface-Renewal)
systems
 Earliest & simplest model: each eddy considered to
behave like a translating solid body
 Large compared to transient diffusion BL
(penetration) thickness
 Dimensional time-averaged mass-transfer coefficient
given by:
23
 4 D 1/2
 A  for E ( PFR) ( Higbie 1935
 j A'' , w   tm 


  A ,b   A , w  
 DA 
1/2
for E (WSR) ( Danckwerts 1951)

 
 tm 
tm  mean eddy contact time (1/(average renewal
frequency))
 Related to prevailing geometry & bulk-flow velocity
 Versatile alternative to Prandtl-Taylor eddy diffusivity
approach
24
 Extinction, ignition, parametric sensitivity of chemical
reactors:
 Simplest modular model for steady-flow behavior of
combustors: WSR + PFR
25
 mmaxupper limit to total mass flow rate, m, at each

upstream condition (Tu, pu, mixture ratio F) above
which extinction of exoergic reaction (flame-out)
abruptly occurs
 For m  mmax , two possible SS conditions exist: one
corresponding to high fuel consumption & high
temperature in WSR, the other to negligible fuel
consumption & rise in T
26
reactors:
Simple, two-ideal reactor “modular” model of gas turbine, ramjet, or rocket

engine combustor
27

reactors:
 Parametric sensitivity: change in reactor performance
for a small change in input or operating parameter

(e.g., Tu)
28
 Example: WSR module with following overall
stoichiometric combustion reaction:
1 gm O + f gm F  1  f  gram P+fQcal(heat)
29

reactors:
 Allow a 2nd reactant (oxidant) & associated heat
generation
 Governs WSR operating temperature, T2
 WSR species mass balance:
m. i 2  i 1   ri '''

O 2 , F 2 , T 2  .VWSR
(i = O, F, P)
30

reactors:
 Overall energy balance:
mc p . T 2  T 1   rF  O 2 , F
'''
2 , T  QV
. WSR
 Source terms for oxidizer & fuel related by:
r  r / f
'''
O
'''
F
 So, O 2 and F 2 can be expressed in terms of T2
31

reactors:
 Overall kinetics represented by Arrhenius-type mass-
action rate law:
  E 
n
1  pM  vO vF
r   A.exp    .O F
'''
F  . vO vF 1 . 
  RT  M O M F  RT 
 LHS  straight line intersecting RHS at 3 distinct T2
values, middle one unstable, upper  ignited WSR
SS, lower  extinguished WSR SS (no chemical
reaction)
32
reactors:
Influence of feed mass flow rate on WSR operating temperature and space (volumetric
heating rate(SHR);(straight line is the LHS of the energy balance equation)
33

reactors:
 Maximum volumetric rate of fuel consumption (hence,
maximum chemical heating rate) occurs at WSR
temperature:
Tb
Tr '''  max 
1  n( RTb / E )
 Only slightly > “extinction” temperature (previous
Figure)
Tb adiabatic, complete-combustion temperature
 Typical E, n values listed in following Table
34

reactors:
aSupplemented, rounded (and selected) values based on Table 4.4 of Kanury (1975)
bUnits are: 1014s-1 (g-moles/cm3)-(n-1), where n is the overall reaction order.
cunits are: 109 BTU/ft3/hr
dStoichiometric mixture, no diluent (“diluent” is N ) unless otherwise specified
2
35

reactors:
 Black-box modular-models capture many important
features of real reactors, useful for correlating
performance data on full-scale & small-scale models
 Predictive ability limited compared to more-detailed
pseudo-continuum mathematical models
 All have, as their basis, macroscopic conservation
principles outlined earlier in this course.
36
PROBLEM1
 The length requirement for a honeycomb-type automotive

exhaust catalytic converter is set by the need to reduce
the CO concentration in the exhaust to about 5% of the
inlet concentration (i.e., 95% conversion). Consider the
basic conditions:
Inlet gas temperature 700K
Inlet gas pressure 1 atm
Inlet gas composition y(N2)=0.93, y(CO)=0.02,
(mole fraction) y(O2)=0.05
37
PROBLEM1
Inlet gas velocity 103 cm/s

Channel cross-section dimensions 1.5mm by 1.5mm (each
channel)
Assumed channel wall temperature 500 K
Assume that the Pt-based catalyst used on the walls of
each channel is active enough to cause the surface-
catalyzed CO oxidation reaction to be diffusion-controlled,
that is, the steady-state value of the CO-mass fraction
established at (1 mean-free-path away from)
38
PROBLEM1
the wall, CO,w , is negligible compared to CO,b(z) within

each channel. Also assume that the gas-phase kinetics of
CO oxidation under these conditions preclude
appreciable (uncatalyzed) homogeneous CO-
assumption in the available residence times. Answer the
following questions:
a. By what mechanism is CO(g) mass transported to the

channel wall, where chemical consumption (to produce
CO2) occurs? What is the relevant transport coefficient
39
PROBLEM1
and to what energy-transfer process and transport
properly coefficient is this “analogous”?
b. Are the mass-heat transfer analogy conditions (MAC,

HAC) discussed in this module approximately met in this
application? What is the inlet mass fraction of CO gas?
c. Estimate the Schmidt number Sc  v / DCO mix mix for CO

Fick diffusion through the prevailing combustion gas
mixture, using the experimental observation that
40
PROBLEM1
1.73
0.216  T  cm 2
DCO  N2  . 
p  300  s
where p is the prevailing pressure (expressed in

atmospheres) and T the mixture temperature (expressed
in kelvins)
d. Under the flow rate, temperature, and pressure

conditions given above and using the mass-transfer
analog, estimate the catalytic duct length
41
PROBLEM1
required to consume 95% of the inlet CO concentration,
and the mixing cup (bulk) stream temperature at this
length.
e. List and defend the principal assumptions made in

arriving at the length estimate (of Part (d))
f. If the catalyst were “poisoned” (e.g., by lead

compounds), what could happen to the CO exit
concentration? Which of the assumptions used in
predicting the required converter length (Part (d)) would
42
be violated?
PROBLEM1
g. If the heat of combustion of CO(g) is about 67.8 kcal/g-

mole CO consumed, calculate how much must be
removed to maintain the channel-wall temperature
constant at 500 K?
h. Automatic operating conditions are never strictly steady,

so that in practice the mass-flow rate, temperature, and
gas composition entering the catalytic afterburner will be
time-dependent. Under what circumstances (be
43
PROBLEM1
 quantitative) can the design equations you used be
defended if used to predict the conditions exiting the duct
at each instant?(Quasi-steady approximation)
 i. At the design condition, estimate the fractional pressure
drop, p / p0 , in the honeycomb-type catalytic afterburner.
If, instead of the honeycomb type converter, a packed
bed device were used to achieve the same reduction in
CO-concentration, would you expect p / p0 to be larger
or smaller than the honeycomb device of your preliminary
design?
44

Model Non-Ideal Flow Reactors

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Model Non-Ideal Flow Reactors

Uploaded by

Copyright:

Available Formats

Mass Transport: Non-Ideal Flow Reactors

Advanced Transport Phenomena

 Simplest approach: apply overall material/ energy/

 Most rigorous: Divide into small subregions, approximate

 Intermediate solution: model as discrete network of small

 RTDF  residence time distribution function (exit-age

 m  feed mass flow rate

 e.g., straight tube through which incompressible fluid

flows with a uniform plug-flow velocity profile

 Partial recycle can alter RTDF

Ideal plug-flow reactor (PFR) with partial “recycle” (recycle introduces a

E WSR   dF / dt   t flow  exp t / t flow 

 Most likely residence time in a WSR is zero!

 Mean residence time = V / m /  

 Dimensionless variance s2 about mean residence time

 indicator of spread of residence times

 Mean residence time related to first moment of E(t), i.e.:

 RTDF for Composite Systems:

 If RTDF for vessel 1 is E1(t) and for vessel 2 is E2(t),

RTDF for a series combination of the two is:

 If vessel 1 is characterized by tflow,1, and s12, and vessel

 RTDF for Composite Systems:

 For a network of n-WSRs of equal volume, for which:

(tflow  V / ( m /  ) ) for each vessel in series)

For vessels 1, 2, 3,…., n in parallel, receiving fractions f1,

 Network of ideal reactors can be constructed to

Real reactors as a network of ideal reactors: Modular

GT combustor; proposed interconnection of reactors comprising “modular”

 Real reactors as a network of ideal reactors: Modular

 Info obtained from tracer diagnostics & from

 Important since RTD-data alone cannot discriminate

 Equivalent vessel network is nonunique

 Each alternative may capture one aspect (e.g.,

combustor efficiency) but not another (e.g., domain

 Real reactors as a network of ideal reactors: Modular

 Diagnose operating problems with existing chemical

 Real reactors as a network of ideal reactors: Modular

 Two networks with identical <E(t)> but with different

shot-to-shot variations, as measured by variance:

will perform differently as chemical reactors

 Mass/ energy transport visualized to occur during

intervals of contact between turbulent eddies & surface

 “stale” eddies replaced by fresh ones

 Effective transport coefficient calculated by time-

 Statistical micro flow (Random Eddy Surface-Renewal)

 If E(t) is defined such that:

Relative portion of each unit interfacial area

St(t)  calculated from transient micro fluid-dynamical

for E (WSR) ( Danckwerts 1951)

 Simplest modular model for steady-flow behavior of

combustors: WSR + PFR

 mmaxupper limit to total mass flow rate, m, at each

Simple, two-ideal reactor “modular” model of gas turbine, ramjet, or rocket

 Extinction, ignition, parametric sensitivity of chemical

 Parametric sensitivity: change in reactor performance

for a small change in input or operating parameter

 Example: WSR module with following overall

stoichiometric combustion reaction:

 Extinction, ignition, parametric sensitivity of chemical

 Extinction, ignition, parametric sensitivity of chemical

 So, O 2 and F 2 can be expressed in terms of T2

 Extinction, ignition, parametric sensitivity of chemical