Professional Documents
Culture Documents
Warfvinge P. Process Calculations and Reactor Calculations For Environmental Engineering 2009 PDF
Warfvinge P. Process Calculations and Reactor Calculations For Environmental Engineering 2009 PDF
Reactor Calculations
For Environmental Engineering
Per Warfvinge
c
!Per Warfvinge
Department of Chemical Engineering
P. O. Box 124
221 00 Lund
Edition for academic year 2009/10
Contents
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3
Contents 4
5 Reactor calculations 52
5.1 Kinetic rate equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.1 First order kinetics . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.2 Second order kinetics . . . . . . . . . . . . . . . . . . . . . . . . 55
5.1.3 Overview of kinetic rate equation . . . . . . . . . . . . . . . . . 55
5.1.4 Reversible reactions . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1.5 Consecutive reactions . . . . . . . . . . . . . . . . . . . . . . . 56
5.1.6 Parallel reactions . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Mean residence time and reaction time . . . . . . . . . . . . . . . . . . 57
5.3 Reactor models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.3.1 The differential mass balance . . . . . . . . . . . . . . . . . . . 60
5.3.2 Methodology for reactor calculations . . . . . . . . . . . . . . . 62
5.4 The ideal completely stirred tank reactor, CSTR . . . . . . . . . . . . 62
5.5 The ideal batch reactor . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.6 The ideal plug-flow reactor, PFR . . . . . . . . . . . . . . . . . . . . . 71
5.6.1 PFR reactor modeling in terms of conversion . . . . . . . . . . 72
6 Non-ideal reactors 76
6.1 Examples of non-ideal mixing . . . . . . . . . . . . . . . . . . . . . . . 76
6.2 Residence-time distributions . . . . . . . . . . . . . . . . . . . . . . . . 77
6.2.1 The normalized RTD – E(t) . . . . . . . . . . . . . . . . . . . . 78
6.2.2 c(t) for an ideal CSTR . . . . . . . . . . . . . . . . . . . . . . . 79
6.2.3 E(t) for an ideal CSTR . . . . . . . . . . . . . . . . . . . . . . 81
6.2.4 Mean residence time from E(t) . . . . . . . . . . . . . . . . . . 81
6.2.5 The F(t)-distribution . . . . . . . . . . . . . . . . . . . . . . . . 82
6.3 Non-ideal reactor models . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3.1 The CSTR in series model . . . . . . . . . . . . . . . . . . . . . 85
6.4 The segregation model . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.5 Other reactor models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7 Instationära CSTR 91
7.1 Svar på ändring i ingångskoncentration . . . . . . . . . . . . . . . . . . 92
7.2 Arbetsmetodik . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Examples
1–1: Balance calculation for a bank account . . . . . . . . . . . . . . . . . . 14
1–2: Solving linear equation systems . . . . . . . . . . . . . . . . . . . . . . 16
1–3: Solving linear equations with fsolve . . . . . . . . . . . . . . . . . . . 17
1–4: Solving non-linear equations with fsolve . . . . . . . . . . . . . . . . 17
1–5: Numerical integration with Euler’s method . . . . . . . . . . . . . . . 19
1–6: Numerical integration with ode45 . . . . . . . . . . . . . . . . . . . . 21
2–1: Separation of ethanol and water . . . . . . . . . . . . . . . . . . . . . 26
2–2: Separation of ethanol and water - 6 unknowns . . . . . . . . . . . . . 28
2–3: Ethanol and water – concentration constraints . . . . . . . . . . . . . 29
2–4: Separation of ethanol and water – moved constraint . . . . . . . . . . 30
3–1: Limitation of the number of splitter constraints . . . . . . . . . . . . . 35
3–2: Process calculation based on total system analysis . . . . . . . . . . . 37
3–3: Degree of freedom analysis based on sub-systems . . . . . . . . . . . . 40
4–1: Catalytic dehydrogenation . . . . . . . . . . . . . . . . . . . . . . . . . 44
4–2: Atomic matrix with linearly dependent rows . . . . . . . . . . . . . . 45
4–3: The rank of a reaction matrix . . . . . . . . . . . . . . . . . . . . . . . 48
4–4: Application of the reaction parameter method . . . . . . . . . . . . . 50
5–1: Average residence time of a lake . . . . . . . . . . . . . . . . . . . . . 59
5–2: CSTR with a 1st order irreversible reaction . . . . . . . . . . . . . . . 63
5–3: Example 5–2 with matrix notation . . . . . . . . . . . . . . . . . . . . 64
5–4: Numerical solution for non-unity reaction order . . . . . . . . . . . . . 65
5–5: Batch reactor with 1st order irreversible reaction . . . . . . . . . . . . 67
5–6: Differential, numerical solution of Example 5–5 . . . . . . . . . . . . . 68
5–7: Integral, numerical solution of Example 5–5 . . . . . . . . . . . . . . . 69
5–8: Simulation of consecutive and parallel reactions . . . . . . . . . . . . . 70
5–9: Emission of a pollutant into a stream . . . . . . . . . . . . . . . . . . 73
6–1: Residence-time distribution from tracer experiments . . . . . . . . . . 83
6–2: Series with an infinite number of CSTR . . . . . . . . . . . . . . . . . 87
6–3: The segregation model applied to example 6–1 . . . . . . . . . . . . . 89
7–1: Exempel på simulering av instationär tankreaktor . . . . . . . . . . . 95
8–1: Estimation of flow velocity . . . . . . . . . . . . . . . . . . . . . . . . 102
8–2: Diffusive and dispersive spread . . . . . . . . . . . . . . . . . . . . . . 110
9–1: Infiltration of a pollutant into a soil . . . . . . . . . . . . . . . . . . . 116
9–2: Complex pollutant transport in a soil . . . . . . . . . . . . . . . . . . 122
10–1: Simulation of a biochemical reaction in a batch reactor . . . . . . . . 130
10–2: Filmtjocklek vid exakt penetrerad film . . . . . . . . . . . . . . . . . 139
10–3: Design of CSTR reactor volume . . . . . . . . . . . . . . . . . . . . . 140
10–4: Boundary layer thickness in a packed bed . . . . . . . . . . . . . . . 144
Examples 7
Notations
Note that the choice of mass unit (g, kg, mol) often is arbitrary
α Dispersivity m
δ Boundary layer thickness m
η Porosity
ηe Effective porosity
µ Substrate turnover rate g s−1
ρ Bulk density kg m−3
Θ Volumetric water content
CHAPTER 1
There are several situations when an engineer needs to gain an understanding of how
different substances flow and react in a system. The system in question may exist in
an industry where a substance should be produced, destroyed or redistributed. The
analyses of such processes is based on the principle of conservation of matter, and
the methods are built upon mass balance. In the chemical industry, and in chemical
engineering, process and reactor calculations based on mass balances are the single
most important tool for the analysis and design of chemical processes.
For natural systems such as lakes, streams, groundwater aquifers as well as entire
ecosystems, mass balances are used to describe and predict how matter is transported
and transformed. In nature, transformations of matter are often controlled by pro-
cesses that man cannot influence, making the design aspect less important for natural
systems.
Every environmental engineer needs to know how to perform mass balance cal-
culations for different systems. Just as importantly, they also have great use of a
profound ability to think in terms of mass conservation and mass balances. Indeed,
the ability to think abstractly in terms of the principles of conservation is a hallmark
of engineers.
Mass balances play an important role in environmental studies and environmental
science. For example, atmospheric transport models, groundwater models and models
to predict the water quality of lakes and streams are all based on mass balances for
chemical components. The integrated models used to predict climate change also
include mass balance equations for different compartments and pools relevant for the
global carbon cycle.
This compendium treats various techniques to carry apply mass balance calcula-
tions to natural and engineered systems. The notations used throughout the text are
given on the previous pages. The text is structured as follows:
This Introductory chapter provides a brief introduction to mass balance calcula-
tions. The key concepts introduced are systems, system boundaries, the mass balance
principle, process and reactors, as well as the entities mass fraction and molar fraction,
and examples of numerical methods.
8
Why do we perform process and reactor calculations? 9
CO2 Fe3+
N2
org-C
NO3-
P
sludge sludge
solids O2 sludge
solids
denitrification nitrification P-precipitation
separation
CO2 Fe3+
N2
org-C
NO3-
P
sludge sludge
solids O2 sludge
solids
denitrification nitrification P-precipitation
separation
Figur 1.2: Example of how system boundaries can be drawn around the different
units in a process.
1.8
1.6
stationär fas
här inträffar en störning
1.4
någonting vad-som-helst
1.2
1
icke-statioär, dynamisk fas
0.8
0.6
ny stationär fas
0.4
0 1 2 3 4 5 6 7 8 9 10
tiden
The term “Input” refers to the flow of the substance into the system. The flow
can have the unit kg s−1 or similar. The term “Output” concerns, in the same way,
the flow of the substance out of the system.
The term “Prod” refers to the amounts of the substance that is produced inside
the system. It must however be noted that Prod can be either positive or negative.
If Prod is negative, it means that the substance is consumed inside the system. In
the chemical context, it is obvious that reactants are consumed (Prod < 0), while the
desired products are produced (Prod > 0).
The last term, “Acc” is conceptually the most difficult. Acc stands for the amount
of a substance accumulated per unit time inside the system. That means that in a
steady-state system Acc = 0 by definition. In a non-steady state system, Acc > 0 if
the amount of the substance increases in the system. If it decreases, Acc < 0. As we
shall see later, the term “Acc” introduces a time derivative in system models. That is
why the mass balance model for non-steady-state systems always is composed of one
or more difference or differential equations.
Solution 1
We apply the equation
Input + Prod = Output + Acc
Since the question dealt with the change during a time-interval, we are seeking the
Acc-term. The equation i re-written as:
Acc = Input − Output + Prod
With the information given the solution becomes:
SEK SEK SEK SEK
Acc = 50000 − 45000 + (2000 − 500) = 6500
year year year year
Problem 2
Given the information about Input and Output, how large must the annual interest
(i.e., the Prod term) be if the account balance should increase by 5 600 SEK a year?
Solution 2
We seek the Prod term and rewrite the equation as:
Prod = Output − Input + Acc
and calculate the interest as:
SEK SEK SEK SEK
Prod = 45000 − 50000 + 5600 = 600
year year year year
1.3 Process calculations and reactor calculations 15
AX = Y
by calculating
X = A−1 Y
However, the inverse of a matrix can only be calculated if all the rows of the (square)
matrix are linearly independent. Linear independence means that no rows can be
created by additions and/or subtractions involving the other rows. For example, of
the two matrixes:
! " ! "
1 1 1 1
A1 = , A2 =
2 3 2 2
1.4 Numerical methods 16
only A1 can be inverted, while A2 cannot. Furthermore, linear algebra has taught
us that the dimension of the largest (square) sub-matrix that is possible to invert is
called the rank of the original matrix. Hence, the rank of A1 =2, while the rank of
A2 =1.
Problem
Solve the following system of equations with Matlab:
3x1 + 4x2 = 10
x1 − x2 =2
Solution
The system of equations can be written as a matrix in the form:
AX = Y
where:
! " ! " ! "
3 4 x1 10
A= , X= , Y =
1 −1 x2 2
With Matlab the system of equations is solved as:
F = AX − Y, so that F < *
where * is so small that it, in practice, equals 0. The only thing one has remember
as a user is to write the equations in a correct form and then submit a guess of X is
that is not completely out of line.
1.4 Numerical methods 17
0 = AX − Y
where
! " ! "
3 4 10
, Y =
1 −1 2
Write an m-file: where f0 is a guess at the solution needed to initialize fsolve. The
Matlab function fsolve is run from workspace with: Of course, the solution is the
same as in the above example; x1 = 2.5714, x2 = 0.5714.
Problem
Solve the following system of equations with the routine fsolve in Matlab:
3x1 + 4x22 = 10
x31 − x2 =2
Solution
The system of equations can be written as a matrix in the form:
0 = 3x1 + 4x22 − 10
0 = x31 − x2 − 2
The two functions are the defined in an m-file exfsolve1.m: which is executed
from workspace with: Solution: x1 = 1.4708, x2 = 1.1819. The vector [1 ; 1], above
called f0, is an initial guess of what the solution vector X is.
In making the mass balance calculations for non-steady state systems, the starting
point is normally one or more ordinary non-linear differential equations. In this
compendium we will almost exclusively solve differential equations numerically. The
exceptions are the simplest cases.
How do we solve a differential equation numerically? First we will take a quick
look at the simplest (and worst) method, Euler’s solution method.
An ordinary differential equation can be written in the form:
dX
= f (t, X)
dt
This differential equation states that if X = Xt at time t, then X = Xt+dt at time
t + dt:
We can interpret this graphically if we follow the derivative of the function X(t)
a rectilinear piece dt, starting at the point (t, Xt ). The approximately estimated
subsequent value of the function is thus a small distance away, in the direction of the
derivative. This value is defined as (t + dt, Xt+dt ). After this value of the function has
been calculated, another value is calculated in the same way. The whole procedure is
repeated to create a vector of estimates function value.
As we see in Figure 1.4 the approximation may differ significantly from the real
function X. With Euler’s method, the calculation error is relatively large, but if one
reduces the dt, the calculation error is also reduced. However, the smaller the dt,
the longer the calculations take. For all practical purposes it is better to use more
sophisticated solution methods such as the ode45 routine in Matlab.
1.4 Numerical methods 19
5 X
X
t+dt
4
approximering av Xt+dt
3
dX/dt i (t,Xt)
2
1 Xt
t t+dt
-1
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Problem
Integrate (simulate) the ordinary non-linear differential equation:
#
dc t=0
= −0.5·c Initial value:
dt c=1
using Euler’s method. The simulations shall be done over the range t = 0 to t = 4 in
steps (dt) of 0.1 units of time.
1.4 Numerical methods 20
Solution
Write an m-file exEuler.m: which is executed from workspace with exEuler. In
the m-file, one can see how the time interval (from 0 to 4) has been divided into 40
pieces, each representing 0.1 units of time. The results are given below, together with
the exact solution. The conclusion is: As long as the solution is monotonous, this
simple integration method works quite well!
2
1.8
1.6
1.4
1.2
0.8
0.6 exakt
euler
0.4
0.2
0 0.5 1 1.5 2 2.5 3 3.5 4
1.4 Numerical methods 21
Solution
Write an m-file exODE45.m: and run it from workspace with: where [0 10] denotes
the time interval that should be covered, and the vector [1 1] represents the initial
values for r X1 and X2 respectively. The solution is presented as a diagram:
1
0.9
X2
0.8
0.7
Värdet på tidsvariabelt tillstånd X
0.6
X1
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
tiden t
CHAPTER 2
In this chapter we shall go through the methods for resolving integral mass balances
for the very simplest type of system. As mentioned in the previous chapter, process
calculations allow us to describe systems with flows of various chemical substances.
The process calculations my be used to quantify fluxes, but may also be extended to
process design and, partly, to explain what is happening within a process system.
Process calculations have played an enormously important role in industry, but
also in environmental research. Without mass balances, we would not have had
any quantitative understanding of flows of mass and matter in any one of earth’s
ecosystems.
Mass balances are known under different names. Often one uses the terms budget
calculations or material balances.
Here we will use a somewhat formal approach to process calculations. They will
involve models with several equations. Solving these equations involves the collection
of as many equations as that there are unknown variables. By working with degree of
freedom analyses as a tool we will systematically examine wether it is possible or not
to solve a certain system of equations. We will also see how they can become solved
by doing certain “tricks”.
In a series of very simple examples, different types of equations and information
will be slowly introduced and will illustrate key concepts in the mass balance calcu-
lations.
The methodology will probably seem pretty tedious, and give the impression of
being overly formal. If so, the impression is correct: the idea is to show a formal
and systematic approach that always leads to a solution if a solution exists, and that
reveals why a solutions sometimes cannot be provided.
The mass balance for element/component A will be denoted as MB A etc.
22
2.1 Component mass balances 23
systemgräns
W1,A W1,B W3,B
1 3
Process
2 4
W2,A W4,A W4,B
Figur 2.1: Illustration of the concepts of process, system and stream variable.
Input = Output
Figure 2.1 shows a process where two physical streams, W1 and W2 are input
fluxes, while W3 and W4 are output fluxes. W1 and W4 both contain the chemical
components A and B, while W2 only contains A and W3 only contains B.
The units for mass flux W may be mass (kg) or mass flux (kg/time). Under these
conditions, the integral component mass balances for A and B become:
MB A : W1,A + W2,A = W4,A
MB B : W1,B = W3,B + W4,B
We can also see that these component mass balances show linear independence (i.e.,
they are not identical). If they were independent, one of the balances would be
“worthless”. The solution would be trivial, of the type 0=0, or 1=1.
The variables W1,A , etc., are called stream variables. A stream variable refers to
the flux of a specific substance in a specific physical stream.
The maximum number of stream variables in a system always equals the number
of streams multiplied by the number of components. If there are 4 streams with 2
components in each, there is a maximum of 8 stream variables. This means that we
need 8 equations in order to allow the system to be unambiguously defined. This is
also true in the example above, although only 6 stream variables are explicitly written
out. The reason is that implicitly W2,B = W3,A = 0.
2.2 Degrees of freedom analysis 24
4. basis of calculations
With known fluxes it is meant that the mass flux of a component in a stream, Wi,j or
the molar flux, Fi,j is given. As mentioned above, the flux may either be given as an
amount (kg, mole) or a flux which is amount per unit of time (kg/time, mol/time).
In the real world, it is quite difficult to make direct measurements of component
fluxes. Instead, one measures the concentration of a component in a stream and
multiplies it by the total flux in the same stream. This will give Wi,j and/or Fi,j .
A known concentrations might mean the actual concentration ci,j (in kg/volume
or mole/volume) but just as often the mass fraction φi,j , or the molar fraction xi,j .
The total constraint is often very useful. This is in spite of the fact that the total
constraint merely states that the whole must equal the sum of the parts. Mathe-
matically, the total constraint implies that the sum of the mass fractions and/or the
molar fractions in a stream equals 1. For example, for a stream i that contains n
components j we can state that:
n
$ n
$
φi,j = 1 , xi,j = 1
j=1 j=1
One can also define corresponding total constraints for mass fluxes and molar fluxes.
For a given physical stream Wi , that contains n components j, we can state that the
2.2 Degrees of freedom analysis 25
systemgräns
1,A •W1 3,A •W3
1 3
1,B •W1 Process
4,A •W4
2 4
2,A •W2 4,B •W4
Figur 2.2: Steam variables expressed in terms of mass fractions φi,j and the total
flux Wi .
component fluxes (quantified by the stream variables) must add up to the total flux.
Hence:
)
n
j=1 Wi,j
Wi =
W )n φ
i j=1 i,j
The corresponding relationships also hold true when the stream variables are ex-
pressed in molar units:
)
n
j=1 Fi,j
Fi =
F )n x
i j=1 i,j
The equations are illustrated in Figure 2.2 which shows how the stream variables
in Figure 2.1 may be expressed in terms of the mass fraction φi,j and the total flux
Wi .
A basis of calculation consists of an arbitrary definition of a flux. This number is
used as an equation in the calculations. A basis of calculation may be (and has to
be!) introduced if no other values of fluxes (i.e. no values of stream variables) are
known or given. This occurs when only concentrations or mass/molar fractions are
specified. For example, a basis of calculation can be set to 1 kg, 100 kg/hour, etc.,
but may only be introduced of it does not contradict other information.
2.3 Process calculations for a separation process 26
The first examples dealing with separation of the two components is schematically
illustrated below. In the process, C2 H5 OH and H2 O flow into a process with one
input stream and leave the system in the other end via two output streams. The
mass flux of each component is denoted as Wi,j , where i denotes stream i, while j = 1
denotes C2 H5 OH and j = 2 denotes H2 O:
W2,1 kg C2H5OH
2
W2,2 kg H2O
W1,1 kg C2H5OH
separations-
1 process
W1,2 kg H2O
W3,1 kg C2H5OH
3
W3,2 kg H2O
460 kg C2H5OH
2
60 kg H2O
500 kg C2H5OH
1
500 kg H2O
W3,1 kg C2H5OH
3
W3,2 kg H2O
The next step is to carry out a degree of freedom analysis. The way the process chart
is drawn, there are only two unknown stream variables, W3,1 and W3,2 . Now, for each
component, it is possible to make a separate mass balance, called MB A and MB B.
Thus, the degree of freedom analysis becomes:
Unknown stream variables 2
Independent component mass balances -2
Degrees of freedom 0
Since the degrees of freedom = 0, the system is defined in an unambiguous manner!
It is thus adequate to define the equstions of the system forming the mass balance
equations in the form Input=Output:
MB C2 H5 OH : 500 = 460 + W3,1
MB H2 O : 500 = 60 + W3,2
or
W1,1 kg C2H5OH
separations-
1 process
W1,2 kg H2O
W3,1 kg C2H5OH
3
W3,2 kg H2O
There are six unknown stream variable, and the independent information are W1,1 =
500, W1,2 = 500, W2,1 = 460 and W2,2 = 60.
Unknown stream variables 6
Independent component mass balances -2
Independent information -4
Degrees of freedom 0
The equations are:
MB C2 H5 OH : 500 = 460 + W3,1
MB H2 O : 500 = 60 + W3,2
Info 1 : 500 = W1,1
Info 2 : 500 = W1,2
Info 3 : 460 = W2,1
Info 4 : 60 = W2,2
500 kg C2H5OH
1
500 kg H2O
W3,1 kg C2H5OH
3
W3,2 kg H2O
The next step is to carry out a degree of freedom analysis. They way the process chart
is drawn, there are four unknowns, W2,1 and W2,2 as well as W3,1 and W3,2 . However,
we have two information: the concentration constraint 0.96 = W2,1 /(W2,1 + W2,2 )
and the total constraint 400 = W2,1 + W2,2 .
Unknown stream variables 4
Independent component mass balances -2
Independent information -2
Degrees of freedom 0
The four equations are:
MB C2 H5 OH : 500 = W2,1 + W3,1
MB H2 O : 500 = W2,2 + W3,2
Info 1 : 400 = W2,1 + W2,2
Info 2 : 0.96 = W2,1 /(W2,1 + W2,2 )
0 = −0.04W2,1 + 0.96W2,2
In matrix notation, AX = Y we have:
1 0 1 0 W2,1 500
0 1 0 1 500
A= , X = W2,2 , Y =
1 1 0 0 W3,1 400
−0.04 0.96 0 0 W3,2 0
2.3 Process calculations for a separation process 30
and the solution becomes: The mass fractions in stream 3 are thus φ3,1 = 116/(116+
484) = 0.1933, and φ3,2 = 484/(116 + 484) = 0.8067.
Problem:
A separation process is fed with an input stream consisting of 500 kg C2 H5 OH and
500 kg H2 O. The output consists of two streams. The first of these output streams
contains 96% C2 H5 OH, while the second amounts to a total of 600 kg. Calculate the
mass fractions of C2 H5 OH and H2 O respectively in the second stream.
Solution:
First, the process chart is drawn with the streams numbered 1-3 and the components
1-2.
500 kg C2H5OH
1
500 kg H2O
W3,1 kg C2H5OH
W3=600kg
W3,2 kg H2O
Just as in the previous example, there are four unknown stream variables. In addition,
the concentration constraint 0.96 = W2,1 /(W2,1 + W2,2 ) is the same as above, while
the total constraint is moved so that 600 = W3,1 + W3,2 .
Unknown stream variables 4
Independent component mass balances -2
Independent information -2
Degrees of freedom 0
The four equations are:
MB C2 H5 OH : 500 = W2,1 + W3,1
MB H2 O : 500 = W2,2 + W3,2
Info 1 : 600 = W3,1 + W3,2
Info 2 : 0.96 = W2,1 /(W2,1 + W2,2 )
0 = −0.04W2,1 + 0.96W2,2
In matrix notation, AX = Y we have:
1 0 1 0 W2,1 500
0 1 0 1 500
A= , X = W2,2 , Y =
0 0 1 1 W3,1 600
−0.04 0.96 0 0 W3,2 0
2.3 Process calculations for a separation process 31
and the solution is: The mass fractions of all constituents are, of course, the same as in
the previous example: φ3,1 = 116/(116+484) = 0.1933, and φ3,2 = 484/(116+484) =
0.8067.
CHAPTER 3
32
3.2 Common types of units 33
3. Degree of freedom analysis and equation system relating to all units and the
streams that connect them, from now on referred to as the “Total”.
What one often finds is that the degrees of freedom are > 0 for the process and
for any of the individual units. For other entities the number of degrees of freedom
= 0. If the problem is solvable, the number of degrees of freedom of the Total is 0.
In case any of teh individual units have degrees of freedom < 0, the system is defined
in a unique manner, and all stream variables can not be calculated.
If you solve the equation system by hand, you have to begin with a unit for
which the number of degrees of freedom is 0. The information thereby obtained is
used stepwise as independent information for other, non-solvable units for which the
number of degrees of freedom > 0. However, if a numerical equation solver is used,
the easiest way to handle to problem is by solving all equations simultaneously.
Mixer
For a mixer, the total output flux must equal the sum of the input fluxes. This holds
regardless of whether the fluxes are given in mass units or in molar units.
3.2.2 Splitter
A splitter divides one input stream into several output streams.
Splitter
For a splitter, it is evident that the sum of all output component fluxes must
be equal to the input flux. But in addition, the composition, as given by mass
fractions or molar fractions, identical in all output streams. This may be utilized in
the calculations by means of splitter constraints.
A splitter constraint states that the concentration of a component in any of the
output streams is equal to the concentration in the input stream. Thus, if an input
stream W1 with two components A and B is split into 3 output streams, W2 − W4
3.2 Common types of units 34
If one composition in W1 is known, one can use this information to re-write the above
expression as:
W2,A
φ1,A =
W2,A + W2,B
Solution:
We try to formulate 4 equations:
MB A W1,A = W2,A + W3,A
MB B W1,B = W2,B + W3,B
W1,A W2,A
Splitter constraint 1 W1,A +W1,B = W2,A +W2,B
W3,A W2,A
Splitter constraint 2 W3,A +W3,B = W2,A +W2,B
3.2.3 Recirculation
A recirculation stream returns material from a downstream part of a process to a
mixer upstream point. Through recirculation, a portion of the material that has
passed through a specific process step will pass through it once more:
Recirkulationsström
One quantifies the recirculation by a quantity called the recirculation ratio, defined
as the ratio between the recirculated stream and the feed. If the feed is larger than
the recirculation stream the recirculation ratio <1.
Recirculation is very common in complex processes. For example, the water treat-
ment plant illustrated in Figure 1.1 includes a recirculation stream from the nitrifica-
tion to denitrification stage in the process. Another example is municipal incineration
plants, where particles are recycled from the gas cleaning units to the furnace.
3.2.4 Bypass
Bypass is simply a unit that allows a stream to bypass a process unit:
Förbiledning
However, a bypass stream is very easy to ignore by mistake. In fact, bypass always
introduces two additional units: a splitter and a mixer. This must be accounted for
when making a complete degree of freedom analysis. In industry, bypass is used as a
tool to manage limited capacity in separation process units.
3.2.5 Purge
Purge means that a fraction of a recycle stream is led off:
Recirkulationsström Avtappning
I fact, the problem only states that the ratio between stream 2 and 3 needs to be
calculated. But to do this, all other stream variables have to be calculated.
Degree of freedom analysis:
First, a degree of freedom analysis is carried out. For an analysis based on the total
system, including all sub-systems, the system boundaries are drawn as follows:
3.2 Common types of units 38
We can now make the degree of freedom analysis to investigate if the problem is
possible to solve. Calling the basis of calculation BoC we get:
Totalt
Stream variables 11
MB (A and B, Splitter) -2
MB (A and B, Separation) -2
MB (A and B, Mixer) -2
Info conc. of A in W1 -1
Info conc. of A in W5 -1
Info separation of A -1
Splitter constraint -1
Basis of calc. -1
Degrees of freedom 0
Equations:
Here, we will utilize the fact that the system can be solved by solving for all the 11
equations at once.
MB A Splitter: W1,A = W2,A + W3,A
MB B Splitter: W1,B = W2,B + W3,B
MB A Separation: W2,A = W4,A + W6,A
MB B Separation: W2,B = W4,B
MB A Mixer: W3,A + W4,A = W5,A
MB B Mixer: W3,B + W4,B = W5,B
Conc. of A in stream 1: W1,A = 0.1(W1,A + W1,B )
Conc. of A in stream 5: W5,A = 0.03(W5,A + W5,B )
Separation of A: W4,A = 0.1W2,A
Splitter constraint (trick): W3,A = 0.1(W3,A + W3,B )
Basis of Calc.: W1,A + W1,B = 1000
3.2 Common types of units 39
Then, all equations are re-written so that all the stream variables end up on the
left side:
In many natural and engineered systems chemical reactions take place in which reac-
tants are converted into products. Products and reactants are the usual nomenclature
in chemistry and chemical engineering. Regarding biological systems one says that
a substrate is metabolized to a metabolite. The treatment plant in Figure 1.1 is an
example of a technological system that is dominated by metabolic chemical processes.
It is not as easy to describe a system with chemical reactions as it is to describe
non-reaction systems. One reason is that “new” components are created in the system;
another reason is that we may not know exactly which components are actually in
the system.
This chapter describes different techniques to make process calculations in the
system with one or more chemical reactions. Some important new concepts are defined
to help when independent information is to be interpreted.
N2 + 3H2 −→ 2NH3
41
4.2 Important concepts 42
form compared to if mass balance equations are set up for the elements N and H:
Element mass balance or N, H : Input = Output (4.1)
Component mass balance for N2 , H2 , NH3 : Input + Prod = Output (4.2)
The difference, obviously, is caused by the fact that elements are indivisible while
components may react and change their nature in chemical reactions. In terms of
constraints related to stream variables, this can be expressed more formally as:
$ $ $ $
For components: Wi,j $= Wi,j and Fi,j $= Fi,j
in out in out
$ $ $ $
For elements: Wi,j = Wi,j and Fi,j = Fi,j
in out in out
When one solves mass balance problems for reaction systems one should therefore:
1. Never make mass balances in terms of mass units, but always use molar units.
2. Only use component mass balances if there are good reasons, otherwise element
mass balances should be used (still in molar units).
Unfortunately, in most natural systems one can never know exactly which compo-
nents are present, how they react and what they form. However, by use of chemical
analytical methods it is quite possible to track the elements that are present in vari-
ous phases, and how they are transported within the system. In conclusion, systems
where biological transformations are important should normally be handled be means
of element mass balances.
Another exciting possibility, not further discussed here, is the use of isotopes.
For example 18 Oand 16 O both have element characteristics, but do in fact behave
somewhat differently in nature.
4.2.2 Conversion
The conversion is a number that represents the extent to which a reactant takes part
in a chemical process. The conversion X for a certain reactant is defined as:
Converted amount
Input flux
Input flux - Output amount
Input flux
X=
$
F
1− $
output
F
input
) )
Hence, if the output flux ( output F ) is equal to the input flux ( input F ) the conver-
sion is obviously X = 0 (i.e., the reactant
) has not reacted at all within the system).
On the contrary, if the output flux ( output F ) is equal to 0, all of the reactant has
been converted (i.e., X = 1, or as percentage in the range 0-100%).
4.2.3 Yield
The yield quantifies the amount of a substance formed, in relation to what could have
been formed if the limiting reactant were converted to a desired product. With this
definition, the yield is 100% if:
1. All of the reactant converted is converted to the desired product (i.e. no-side
reactions take place).
2. The conversion of the reactant in question is 100%.
For example, let us define the yield if the limiting reactant R is converted to the
main product P. The stoichiometry of the reaction is:
νR R −→ νP P
and the yield is thus defined as:
$
Fi,P /νP
output
Yield = $
Fi,R /νR
input
4.2.4 Selectivity
The selectivity tells the amount of the main product that is formed in relation to the
amount of (undesired) bi-products that are formed. If the molar fluxes of all products
have been determined, the selectivity is calculated as:
$
Fmain product
output
Selectivity = $
Fbi-products
output
4.3 The element mass balance method 44
FH 2 O = 1
then
FH = 2 and FO = 1
This section shows how to set up element mass balances for reaction systems. The
degree of freedom analysis includes the analysis of the atomic matrix that determines
how many independent element mass balances can be set up.
The degree of freedom analysis for element mass balances is:
Number of (unknown) stream variables
- Number of independent element mass balances
- Number of independent information
= Number of degrees of freedom
In a first, trivial example, the methodology for process calculations for reaction
system and use of mass balances will be introduced.
Problem
In a process, the following reaction takes place:
catalysis
C2 H6 −−−−−−−→C2 H4 + H2
systemgräns
C2H6 C2H6
1 Process 2
C2H4 H2
Determine all fluxes (i.e., stream variables) in the system if the feed contains 100 mol
C2 H6 and the conversion is 60%.
Solution
If we assume that the element mass balances for C and H are independent, the degree
of freedom analysis becomes:
Stream variables 4
- Independent element MB -2
- Independent information -2
Degrees of freedom 0
4.3 The element mass balance method 45
Hence the system is uniquely defined. Using F to denote molar flux, we can set
up mass balances that describe the number of moles of C and H that are associated
with the input and output streams. For example, a flux of 1 mole of C2 H6 obviously
“carries” 2 moles of C and 6 moles of H. The system of equations then becomes:
MB C : 2F1,C2 H6 = 2F2,C2 H6 + 2F2,C2 H4
MB H : 6F1,C2 H6 = 2F2,H2 + 6F2,C2 H6 + 4F2,C2 H4
Info 1 : F1,C2 H6 = 100
Info 2 : F2,C2 H6 = (1 − 0.6)F1,C2 H6
In matrix notation AX = Y we get:
2 0 −2 −2 F1,C2 H6 0
6 −2 −6 −4 F2,H2 0
A=
, X =
F2,C2 H6 , Y = 100
1 0 0 0
(1 − 0.6) 0 −1 0 F2,C2 H4 0
and the Matlab solution becomes:
Problem
In a process, the conversion:
catalysis
2C2 H4 −−−−−−−→C4 H8
4.3 The element mass balance method 46
systemgräns
C2H4 C4H8
1 Process 2
C2H4
Make a degree of freedom analysis provided that the feed contains 100 mol C2 H6 and
the conversion is 60%.
Solution
The atomic matrix is:
C 2 H4 C 4 H8
0 1
C 2 4
H 4 8
Obviously, the rank for the atomic matrix is 1, which means that only one of the
element mass balances for C or H is dependent on the other. Hence, the degree of
freedom analysis is:
Stream variables 3
- Independent element MB -1
- Independent information -2
Degrees of freedom 0
We can thus conclude that the reaction may actually be viewed as if:
catalysis
2A−−−−−−−→A2
where A = C2 H4 . Obviously, the elements C and H will always appear in the same
proportions. They are thus not independent, but dependent on each other.
4.4 The component mass balance method 47
As we see, ξ will always be a positive number since Fin,A > Fout,A and Fin,B < Fout,B .
In order to generalize, let us consider more complex situation where one chemical
reaction takes place:
N2 + 3H2 −→ 2NH3
For the two reactants N2 and H2 , ξ has a negative sign in front of it since Prod<0.
In addition, there is a stoichiometric constraint since the number of moles of H2 that
are consumed is always 3 times greater than the number of moles of N2 produced.
If we let ξ denote the reaction parameter for the reaction in question and while
the stoichiometric coefficients are denoted νi we get:
ξ ξN
N
2 = 2
ν −1
N2
ξH2 ξH2
ξ= = (4.3)
νH2 −3
ξ ξ
N2 = N H3
νN H3 2
Consequently, it is also true that:
ξN2 = νN2 ξ , ξH2 = νH2 ξ , ξN H3 = νN H3 ξ
4.4 The component mass balance method 48
In many (most) systems, more than one chemical reaction takes place. In that case
it is necessary to define one reaction parameter for each reaction. This is indeed very
useful since the ratio between calculated values of the different reaction parameters
is the same as the ratio between the extent to which the chemical reactions occur.
If there are many chemical reactions i in the system, the general component mass
balance for component j will thus become:
all
recations
$
Finput,j + νi,j ξi = Foutput,j
i
Problem
Investigate the number of independent chemical reactions in a system where the fol-
lowing chemical reactions occur:
Reaction 1 A −→ 2B
Reaction 2 B −→ C
Reaction 3 A −→ 2C
Solution
Form the reaction matrix by setting up the chemical reactions as rows and the chemical
components involved as columns:
νA νB νC
Reaction 1 −1 2 0
Reaction 2 0 −1 1
Reaction 3 −1 0 2
4.4 The component mass balance method 49
We now use Matlab to calculate the rank of the reaction matrix: Obviously, the rank
of the matrix is 2, not 3. Hence only two of the rows and reactions are linearly inde-
pendent. These two can, in this case, be chosen freely among the three reactions.
2. Output terms are quantified by stream variables representing fluxes out of the
system.
3. Prod terms are quantified by means of reaction parameters, one for each reac-
tion.
This means that the total number of unknowns in the equation system is increased
by the number of reaction parameters. Consequently, the degree of freedom analysis
becomes:
Number of (unknown) stream variables
+ Number of reaction parameters
- Number of independent component mass balances
- Number of independent information
= Number of degrees of freedom
4.4 The component mass balance method 50
Reaction 1 C 2 H6 −→ C2 H4 + H2
Reaction 2 C 2 H6 + H 2 −→ 2CH4
The feed (i.e., the input flow of reactant gas) consist of 85% C2 H6 while the rest is
inert gas (N2 ). Calculate the magnitude of all output fluxes, if the total conversion
of C2 H6 is 50.1%, while the yield with respect to formed ethene C2 H4 is 47.1%. In
addition, calculate the selectivity with respect to ethene relative to formed CH4 .
Solution
The process chart becomes::
The next step is to form a reaction matrix. Although it is obvious that the reactions
are linearly independent (CH4 only appears in one of the reactions), this can be
analyzed formally by calculating the rank of:
νC2 H6 νC2 H4 νCH4 νH2
Reaction 1 −1 1 0 1
Reaction 2 −1 0 2 −1
Obviously, the rank of the reaction matrix is 2, there are 2 independent reactions
and we need to introduce 2 reaction parameters in our calculations. The degree of
freedom analysis, combined with the information given results in:
Stream variables 7
+ Reaction parameters 2
- Independent MB -5
- Independent info -3
Degrees of freedom 1
We now introduce the basis of calculation F1 = 100 mole in order to eliminate the
last degree of freedom. This is allowed since no other flux is given. The equations
4.4 The component mass balance method 51
become:
MB C2 H6 : F1,C2 H6 − ξ1 − ξ2 = F2,C2 H6
MB C2 H4 : ξ1 = F2,C2 H4
MB H2 : ξ1 − ξ2 = F2,H2
MB CH4 : 2ξ2 = F2,CH4
MB Inert : F1,N2 = F2,N2
Conc. inF1 : F1,C2 H6 = 0.85(F1,N2 + F1,C2 H6 )
Conversion : F2,C2 H6 = (1 − 0.501)F1,C2 H6
Yield : F2,C2 H4 = 0.471F1,C2 H6
Basis of calc. : F1,N2 + F1,C2 H6 = 100
and the Matlab solution becomes: Thus, the selectivity is F2,C2 H4 /F2,CH4 = 40.03/5.1 =
7.849.
CHAPTER 5
Reactor calculations
In the previous chapters, we have mainly dealt with mass balance problems to calcu-
late certain material flows over system boundaries, utilizing other flows, information
and balances. For reaction systems, we have also defined valuable quantities such as
conversion, yield and selectivity.
However, we have not given the chemical reactions much attention. The integral
mass balances can, at the most, provide a possibility to calculate the reaction param-
eter ξ which is a measure of the total production/consumption of a substance in a
system. The integral mass balances cannot, however, explain why a reaction occurs
to a certain extent, and certainly cannot be used to describe the dynamics of the
reaction system.
In this chapter, we will study the chemical reactor by which we mean simply a
volume in which occur one or several chemical reactions. Figure 5.1 shows some
chemical reactors: a lake, a denitrification basin in a sewage treatment plant and
dough which is in the process of becoming bread in a bakery. Despite the differences
between reactors, they have two important features in common: they have a clear
system boundary, and they contain reactants that participate in chemical reactions.
In several aspects, the three reactors shown are very different. The lake is largely
mixed, although it has periodically stable stratification between the epilimnion and
the hypolimnion. On the contrary, the denitrification basin is horizontally but not
vertically mixed. If it were, oxygen would be mixed into the water and the denitrifica-
tion process, which occurs under anaerobic, reducing conditions would slow down or
even stop. In terms of system boundaries, both the lake and the basin have inflow and
outflow streams. Flow rates vary with time, as well as the concentrations of dissolved
substances in the inputs and outputs as well as within the system itself.
The bread dough undergoing fermentation, however, has no inflows or outflows.
The chemical process occurring in the “dough systems” is primarily a transformation
of carbohydrates (sugar) to CO2 and H2 O. However, these products do not leave
the dough; they stay within the system. We see this as the dough rises. When the
fermentation has progressed to a certain level, the dough is placed on a moving belt
that will convey it into a continuous baking oven.
In this chapter, we limit ourselves to treating chemical reactions in liquid systems.
These are called homogeneous liquid phase systems. We will not treat gas phase
52
Reactor calculations 53
Figur 5.1: Three examples of reactors; a lake, a basin in a treatment plant and
a piece of dough at Lockarps in Malmö.
5.1 Kinetic rate equations 54
reaction systems.1
We will focus on three fundamental concepts that are very important for chemical
reactor calculations:
• Kinetic models used to calculate the rate of chemical reactions, and rate con-
stants.
• The concept of residence time which describes how much that that flows in and
out of the system.
• The mixing models or reactor models that are used to describe how the reactants
come in contact with each other within the reactor.
number of moles change due to the reaction, the pressure and the flow rate will also change.
5.1 Kinetic rate equations 55
Tabell 5.1: Examples of kinetic rate equations. The abbreviations “rev.” and
“irrev.” denote “reversible” and “irreversible” reactions respectively. The units
are based on the use of mole, liters (L) and hours (hr) but any measure of volume
and time may be used.
Reaction
Type Example rA Units
order
0 irrev. A −→B rA = k k: mol
L·hr
1 irrev. A −→B rA = −kcA k : hr
1
1 rev. A !B rA = −kf cA + kb cB k: 1
hr
catalysis K + cA L
A + B −→ C
The theoretical basis is that, in order for A and B to react they must “meet”, and
the probability that a certain molecule of A meets B within the system is directly
proportional to the product of the concentrations of A and B. Thus, the general rate
equation becomes:
r = kcA cB
Since the stoichiometric coefficients are -1 or 1, we get the following rate equations
for A,B and C:
rA = −kcA cB
rB = −kcA cB
rC = kcA cB
The reaction is called second order because the sum of exponents on cA and cB in the
rate equations is 1+1=2.
The table is based on the reaction order of the reaction in question. One should
note that if the reaction rate is independent of the concentrations of reactants, the
reaction is said to be a zero order reaction
One should also be aware that the the unit of the rate coefficient will differ with
reaction order. The principle is that the units of the coefficient must match the
reaction order (expressed as cA , c2A , cA cB , etc.) so that the unit of r is correct.
A special case is the biological reactions described by Monod kinetics or Michaelis-
Menten kinetics. Many enzyme catalyzed reactions follow this type of kinetics, which
can be inferred theoretically. Such reactions are of 1st order at low concentrations of
reactants, but at 0 order at high concentrations of reactants. The cut-off is gradual,
but affected by the the value of the half-rate coefficient, K. More on this in Chapter
10.
Here , B is the intermediate in the reaction A−→C. If we break this reaction down
into its’ two first order reactions, given index 1 and index 2 respectively, we can
deduce the rate equation for each component. First we recognize that:
r1 = k1 cA
r2 = k2 cB
5.2 Mean residence time and reaction time 57
As we see, the stoichiometric coefficients are either -1 (for reactants) or 1 (for prod-
ucts). Hence the rate equations for A, B and C become:
rA = r1 = −k1 cA
rB = r 1 − r2 = k1 cA − k2 cB
rC = r2 = k2 cB
It is thus possible to express the equilibrium constant in terms of a ratio between two
rate coefficients, or cA and cB .
Then, if:
r1 = k1 cA
r2 = k2 cA
we get:
rA = −r1 − r2 = −(k1 + k2 )cA
rB = r1 = k1 cA
rC = r2 = k2 cA
One example of parallel reactions in nature, is the reaction of nitrous oxide, NO2 •, in
the formation of nitric acid, HNO3 , from reactions either with OH• or NO3 •:
V
• The average residence time for the reactor: θ = (sec, hours, etc.)
Q
Technical systems, the flow rate is often constant. In natural systems however, the
inflow to lakes or the flow in watercourses can vary greatly. Then the flow Q is not
a constant value, but may vary 1-2 orders of magnitude between high flow periods,
such as when snow melts in as compared to dry periods in the summer.
If the average volumetric flow rate is to be measured, we can distinguish three
different cases. First, the flow can be constant. Second, the flow may be variable
and the measurements may be evenly or unevenly spaced in time. Finally, the flow
may be variable but the measurements continuous. For these three cases, the average
volumetric flow rate is determined as:
Q constant flow
$ n
Q̄ = n
1
Qi n evenly spaced measurements (5.1)
i=1
2 t
1 Q(t)dt continuous measurements
t
0
Based on the calculation of the average volumetric flow Q̄, the residence time θ is
calculated from the volume as V /Q̄. An example of a calculation with variable flow
and monthly measurements (case 2) is given in Example 5–1 on the following page.
For an open system there is no “time=0”, since the flow is continuous. For a
closed system however, which has no volumetric inflow or outflow, one can say that
“time=0” is when the “doors are closed”. In terms of chemical reactions, t = 0 is when
the chemical reaction starts in the closed system. In later sections in this chapter, we
will look into this situation in conjunction with the batch chemical reactor.
5.2 Mean residence time and reaction time 59
2500
Uppmätt flöde (m3/dygn)
2000
1500
Medelflöde
1000
500
0
1 2 3 4 5 6 7 8 9 10 11 12
Månad
5.3 Reactor models 60
As the drawings illustrate, these types of chemical reactor all differ since:
1. The ideal tank reactor is open and perfectly mixed.
2. The ideal batch reactor is closed and perfectly mixed.
3. The ideal plug-flow reactor is open and not mixed.
Although based on Input + Prod = Output + Acc, there are differences relative to
how the mass balance principle was used in Chapters 2-4: the Prod-term is expressed
in terms of kinetic rate equations, and the Acc-term deals with possible changes with
time.
The Input and Output-terms: The terms representing the fluxes across the re-
actor system boundaries, Fin and Fout , are given by the product of the concentration
c and the volumetric flow rates Qin and Qout respectively. If the flow rate is constant,
Qin = Qout = Q.
The Prod-term: The Prod-term quantifies how much of a substance that is pro-
duced in the reactor per unit of time. The Prod-term is formed by multiplying the
kinetic rate equation (mole/unit time and unit volume) by the volume. This is logical
since scaling up will lead to a higher “Prod” while scaling down will decrease the
amount converted in the chemical reaction.
If, for example, only the reaction A −→ B occurs with rA = −kcA , we get:
Prod = −kcA · V
In more general terms, we can express the Prod-term for component j reacting in
several chemical reactions i, as the sum of the contributions for all reactions:
all i
$
Prodj = νi,j ri V
i=1
Thus, in terms of calculations there are three actions related to the Prod-term:
1. Identify all reactions.
2. Define the corresponding kinetic rate equations.
3. Include the kinetic rate equations in the Prod-term with the right stoichiometric
coefficients ν and signs (+ or -).
The Acc-term: The Acc-term quantifies the change in the number of moles per
unit of time (i.e., the time–derivative of the molar amount). The molar amount in the
reactor is given by the product of the concentration c and the volume V (i.e., c · V ).
By taking the differential of cV , we get the Acc-term as:
dc dV
V +c if V varies with time
d dt dt
Acc = cV =
dt
dc
V if V is constant
dt
In summary: For a component A which reacts in one single reaction in a reactor
system where the flow rate and the volume is constant, the differential mass balance
becomes:
d(cA V ) mole
Qin cin,A + rA V = Qin cout,A + (5.2)
dt unit time
5.4 The ideal completely stirred tank reactor, CSTR 62
1. Define the flow and mixing conditions in terms of an ideal reactor model.
2. Determine the reactor-specific parameters.
3. Define the kinetic equation(s) and rate constant(s).
4. Combine the reactor model and the kinetic equations(s) in the differential mass
balance.
5. Do the calculations, analytically or numerically.
cin
Qin crV
cut
Qut
V
or, if we introduce the residence time θ = :
Q
cin,A
cin,A = cA (1 + kθ) , cA =
1 + kθ (E5–2.5)
c
in,B = −kcA θ + cB , cB = cin,B + kcA θ
Equation E5–2.5 allows us to solve directly for cA . Before inserting numbers, we
5
recognize that θ = = 4 minutes, so that:
1.25
2
cA = = 0.667 mole m−3
1 + 0.5·4
and thus, cB = 0.2 + 0.5 · 4 · 0.667 = 1.533 mole m−3 .
5.4 The ideal completely stirred tank reactor, CSTR 64
Solution
Here, the equation E5–2.5 will contain a non-linear term:
cin,A = cA + kc1.2
A θ ≡ cin,A = cA (1 + kc0.2
A θ)
c
in,B = − 2 kcA θ + cB
1 1.2
crV
Prod = Acc
dc
rV = V
dt
or, if the volume V is eliminated:
dc
=r (5.5)
dt
At this point, one can see that there are two ways to use the above reactor model,
the differential view and the integral view. In both cases, we must be aware that the
boundary conditions are not related to input and output fluxes (i.e. Qcin or Qcout ).
Instead, the boundary conditions relate to time. The relevant boundary condition is
thus the initial condition, or inititial concentration co , which denotes c at time t=0.
With the differental view we solve equation 5.5 to get:
dc
= r " c(t) from t=0 to t=t
dt
In this case we get an expression that allows us to evaluate c for every t between 0
and t. It provides an answer to a question such as, “What is the concentration after
a certain time.”
With the integral view, we do the opposite:
dc
= r " t required to change c from c0 to c
dt
5.5 The ideal batch reactor 67
This method is more useful if one asks, “How long time does it take to reach concen-
tration c.”
In more mathematical terms, the two cases can be expressed as:
#
dc t = 0, c = c0
Differential : =r (5.6)
dt t = t, c = c
2 t 2 c
dc
Integral : dt = (5.7)
0 c0 r
rA = νA r = −kcA
Combining the mass balance equation and the kinetic equation results in the reactor
model:
dcA
= dt
−kcA
2 2 t
−1 cA dcA
= dt
k c0,A cA 0
2 cA 2 t
d ln cA = −k dt
c0,A 0
[ln cA ]ccA
0,A
= −k[t]t0
cA
ln = −kt
c0,A
cA = c0,A · e−k·t (E5–5.8)
With numbers inserted we get cA = 2e−0.5·4 = 0.2707 mole m−3 .
5.5 The ideal batch reactor 68
1.8
1.6
koncentrationen av A (mol/m3
1.4
1.2
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2 2.5 3 3.5 4
tid (minuter)
5.5 The ideal batch reactor 69
We can now form the function that should be integrated, and define it in the m–
file ex56: Thereafter the function in ex56 is integrated from cA = c0,A = 2 to
cA = 0.2707. In Matlab this is done as: The result is t ≈ 4, which of course
corresponds perfectly with the result obtained in Example 5–5.
5.5 The ideal batch reactor 70
0.9
0.8
0.7
koncentrationen av C
0.6
0.5
0.4
koncentrationen av B
0.3
0.2
koncentrationen av A
0.1
F + rdV = F + dF
dV dV dV dV dV dV dV dV dV dV dV
dV dV dV dV dV dV dV dV dV dV dV
dV dV dV dV dV dV dV dV dV dV dV
Figur 5.3: Illustration of how a PFR can be seen as a “train” of infinitely small
batch reactors with the volumes dV that together add up to the volume V.
The mass balance equations for the PFR are virtually identical to the correspond-
ing equations for the batch reactor, as shown in equation 5.6. This becomes even
clearer if we substitute V /Q with θ:
#
dc θ = 0, c = c0
Differential : =r (5.11)
dθ θ = θ, c = c
2 θ 2 c
dc
Integral : dθ = (5.12)
0 c0 r
If one, such in Example 5–5 solves the mass balance model analytically for the
first order reaction A−→B, one gets:
V
rA = −kcA " cA = c0,A e−k Q = c0,A e−kθ (5.13)
Obviously, the PFR and the batch reactor are modeled in exactly the same way,
the only difference being that the reaction time t for the batch reactor is substituted
by the residence time θ for a PFR. This “coincidence” has been illustrated in Figure
5.3. The figure shows that a PFR can be seen as a “train” moving through a “tunnel”
where each “wagon” is independent of all the others. The time that a “wagon” stays
in the “tunnel” corresponds to the residence time, while the time the reactants stay
in the “wagon” corresponds to the reaction time,
the conversion of A as XA , the mass balance over the element dV in a PFR becomes:
In + Prod = Out
Qcin,A (1 − XA ) + rA dV = Qcin,A (1 − XA − dXA )
Combined with the kinetic equation for a first order reaction rA = −kcin,A (1 − XA )
we get:
Qcin,A (1 − XA ) − kcin,A (1 − XA )dV = Qcin,A (1 − XA − dXA )
dV
−k(1 − XA ) = −dXA
Q
At this point one has to substitute variables, i.e. α = (1 − XA ) so that:
dα = −1·dXA
2 2
V
dV 1 α dα
= −
0 Q k 1 α
1 1 α
[V ]V = − [ln α]1
Q 0 k
Since ln(α = 1) = 0 one gets, with α = (1 − XA ) re-inserted:
ln(1 − XA )
θ=−
k
This equations can be seen as a normalized inverse of equation 5.8, which gave cA as
a function of the residence time θ.
The substitution is made by recognizing that the control volume dV can be written
as the product AdL. We can then re-write equation 5.8 as:
r
dV = dc
Q
rA
dL = dc
Q
Combined with the rate equation rA = −kc0.5
A we get:
A
−kc0.5
A dL = dcA
Q
In this case, the conversion of A, XA should be 90%. We can then chose to write the
concentrations cA in terms of cin,A and the conversion so that dcA = −cin,A dXA :
A
0.5
−kcin,A (1 − XA )0.5 dL = −cin,A dXA
Q
0.5 A
kc−0.5
in,A (1 − XA ) dL = dXA
Q
We will now solve the problem from a differential as well as an integral perspective:
#
dXA −0.5 0.5 A L = 0, XA = 0
Differential : = kcin,A (1 − XA )
dL Q L = L, XA = 0.9
2 2
L Qc0.5
0,A
0.9
dXA
Integral : dL =
0 kA 0 (1 − XA )0.5
5.6 The ideal plug-flow reactor, PFR 75
We let Matlab solve the integral form with quad8. The Matlab m-file is: and it is
executed with: The integration shows that the concentration of A will have decreased
by 90% (or to 10%) of the initial concentration ca 2900 meter downstream the source
point. The same result is provided by the differential form, as simulated with the
command:
1
0.9
0.8
0.7
omsättning av föroreningen
0.6
0.5
0.4
0.3
0.2
0.1
0
0 500 1000 1500 2000 2500 3000
»ode45(’ex58diff’,[0 3000],[0]): sträcka nedströms utsläppspunkten (m)
CHAPTER 6
Non-ideal reactors
Chapter 5 presented the basis for chemical reactor calculations. It included theory
and examples on how to calculate the extent to which a chemical reaction occurs as a
function of the reaction kinetics, the concentration of reactants and products, as well
as mixing conditions. The mixing conditions were incorporated by using the ideal
reactor models CSTR, batch and PFR. These reactor types can be seen as extremes,
where the tank reactor and batch reactor are based on complete mixing, while the
PFR is stirred in two dimensions but not along the direction of flow. In all cases,
except 0th order reactions, the mixing conditions affect conversion of reactants in the
system.
In many cases, however, both the ideal CSTR and the ideal PFR are quite bad
approximations of the real mixing conditions in a specific open systems. Thus, in
those cases our models and reactor calculations can be misleading if we use on of the
ideal reactor models. This chapter describes the causes of non-ideal behavior, general
methods to describe the mixing conditions, as well as helpful non-ideal reactor models.
The most important new concept is residence time distribution.
76
6.2 Residence-time distributions 77
Figur 6.1: Illustration of how the water in a river or stream may be irregular,
so that some “packages” of water move faster while others move slower than
average.
there, tied up in the tissue of fish, though it would otherwise have been washed out
since long.
These are examples of situations in which a reaction system cannot be described
well with an ideal reactor model (CSTR or PFR). In summary, one can say that non-
ideal mixing conditions arise due to short-cutting, channel formation, the presence of
stagnant zones, or back-mixing.
2 ∞
Ms = Q(t) · c(t)dt (6.1)
0
where
Ms = amount of tracer mass
Q= volumetric flow rate volume time−1
c(t) = tracer concentration mass volume−1
Note that if Q(t) is constant, the factor Q can be placed outside the integral.
To get the residence time distribution, one should simply normalize the the mea-
sured concentration c(t). In this context, normalizing c(t) means that the integral of
c(t) with time becomes independent of MS .
The normalized c(t), the RTD, is called the E(t)-distribution. If Q is constant,
E(t) is calculated as:
Q c(t) 1
E(t) = c(t) =2 ∞ (6.2)
Ms time
c(t)dt
0
This is really quite obvious, since the similarity implies that all trace elements which
are introduced into a system must have a residence time between 0 and ∞.
Another way to interpret E(t) is to see what fraction of the injected tracer remains
after a certain time in the system. We may then split the integral Equation 6.3 into
6.2 Residence-time distributions 79
cin=0
Q c=0
c V
Q
t<0
cin=hög
Q
c=ct=0
c V t=0
Q
cin=0
Q c < ct=0
c V t>0
Q
cin=0
Q
c << ct=0
c V t >> 0
Q
c(0)
c(t)
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
tid/medeluppehållstid
Figur 6.3: c(t) for a CSTR. The value of c(t) for t = θ is indicated.
where Q is the volumetric flow into the reactor. However, since there is no tracer
in the input flow (the tracer is added as a pulse) and since tracers do not react (by
definition), we get cin = 0 and r = 0. hence, the mass balance is reduced to:
d(cV )
= −Qc (6.6)
dt
Since V /Q is the mean residence time θ we get:
1
dc = dt (6.7)
θ
2 2
c
dc 1 t
= dt (6.8)
c(0) c θ 0
c t
ln = (6.9)
c(0) θ
c = c(0)e−t/θ (6.10)
The obtained function c(t) thus shows that a tracer which is added to an ideal (i.
e., perfectly stirred) tank reactor, will be flushed out of the reactor according to an
exponential process (Figure 6.3).
6.2 Residence-time distributions 81
1/Θ
0.8/Θ
0.6/Θ
0.4/Θ
0.2/Θ
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
t /Θ tid/medelupphållstid
Figur 6.4: E(t) for a CSTR. The value of E(t) for t = θ is indicated.
e−t/θ e−t/θ
E(t) = = (6.12)
−θ[e−t/θ ]∞
0 θ
Figure 6.4 shows E(t) for the ideal CSTR. The figure indicates the value of E(t)
for t = θ, which is e−1 /θ ≈ 0.37/θ. This means that when one mean residence
time (θ) has passed, the concentration of tracer will amount to 37% of the original
concentration. This also means that 63% of the tracer has left the reactor.
1.0
0.8 F(t)
0.6
0.4
0.2 E(t)∗Θ
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
tid/medeluppehållstid
Figur 6.5: F (t) or a CSTR, which shows the proportion of the volume elements,
or tracer, which has left the reactor after a certain time.
We can now show that the hydraulic residence time θ = V /Q of a CSTR is identical
to the average residence time of the added tracer. We show that by inserting the E(t)
of the CSTR (Equation 6.12) into Equation 6.13:
2 ∞ 2 ∞ 2 ∞
e−t/θ 1 1 2 te−t/θ
tE(t)dt = t dt = te−t/θ dt = [θ (− 0 =
− e−t/θ )]∞
0 0 θ θ 0 θ θ
θ2
= (−0 − (−1)) = θ
θ
Figure 6.5 shows how F (t) increases, and F (t) =1 when all of the tracer has crossed
the lower boundary.2
2 One can also show that the trace element is added continuously from t = 0. The resulting E(t)
will look exactly like the F (t) obtained if the tracer is added as a pulse to the reactor.
6.2 Residence-time distributions 83
Calculate the mean residence time of the reactor from the resulting data.
Solution
The average residence time is obtained by going through the steps:
Spårämneskoncentration
7
0
0 10 20 30 40 50 60
Tid (minuter)
One can also directly see that the mean residence time must be greater than the time
at which the concentration of tracer in the output is greatest. This is because the
“tail” of the tracer gives a high contribution to the mean residence time.
The necessary calculations are compiled in the following table:
i t (min) ci ∆ti ci ·∆ti Ei t·Ei ∆ti
0 0 0 2 - - -
1 2 1.1 2 2.2 0.0095 0.038
2 4 2.9 2 5.8 0.0250 0.200
3 6 4.8 2 9.6 0.0414 0.497
4 8 5.9 2 11.8 0.0509 0.815
5 10 6.2 2 12.4 0.0535 1.071
6 12 5.8 2 11.6 0.0500 1.202
7 14 5.1 2 10.2 0.0440 1.233
8 16 4.3 2 8.6 0.0371 1.188
9 18 3.7 4 14.8 0.0319 2.300
10 24 2.2 6 13.2 0.0190 2.736
11 30 1.2 8 9.6 0.0103 2.487
12 40 0.4 10 4 0.0034 1.382
13 50 0.2 10 2 0.0017 0.864
$ $
=115.8 = θ =16.0
i i
E(t)∗Θ
cin 2.5
Q
c1 2.0 ideal tub
Vtot/2
1.5
E(t)∗Θ
cin 2.5
Q c1
2.0 ideal tub
Vtot/5
c2
1.5
Vtot/5
c3
1.0 ideal tank
Vtot/5
c4 n=5
c 0.5
Vtot/5
c5 Q
Vtot/5 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
tid/total medelupphållstid
E(t)∗Θ
cin 2.5
Q c1
Vtot/10 c2 2.0 ideal tub
Vtot/10 c3
Vtot/10 c4
Vtot/10 c5 1.5
Vtot/10
E(t)∗Θ
cin 2.5
Q c1
Vtot/30 c2 2.0 ideal tub
Vtot/30 c3
Vtot/30 c4
Vtot/30 c5 1.5
Vtot/30
From this expression, it is not obvious how cout will be expressed. However, one
dimensional mathematics teaches us that:
3 4x
1
lim = e−k
x→∞ 1 + xk
The output concentration from this CSTR in series is thus exactly the same as for a
PFR with the residence time θ. This is of course in full accordance with Equation
E5–5.8.
dV dV dV
dV
dV dV
dV
dV dV
dV dV
dV dV
dV dV
dV dV dV
dV
dV dV dV dV dV
dV
dV
dV dV
dV dV
dV dV dV
dV
Figure 6.7 illustrates how the segregated elements, batches, add up to form the
reactor.
To put the segregation model into practice to calculate cout,A one needs to combine
three elements:
The first step is to combine the kinetic equation of mass balance for the batch
reactor. This gives, as before,
dc = rdt
For example, we have shown that for a 1st order reaction A → B we get
cA = c0,A e−kt
We must now recall that each tiny, segregated batch reactor has the concentration
cin,A at the moment that they enter the reactor. Thus, c0,A = cin,A . According to
the segregation model, the output from the actual, non-ideal reactor is calculated as:
2 ∞
cout,A = cin,A e−kt E(t)dt (6.16)
0
6.4 The segregation model 89
Solution
According to Equation 6.16 we know that:
2 ∞
cA
= e−kt E(t)dt
cin,A 0
Instationära CSTR
91
7.1 Svar på ändring i ingångskoncentration 92
A−→B r = kcA , k θ = 1
Qin Qin
[A]t≤0 [A]
[B]t≤0 [A]t≤0 [B] [A]
V [B]t≤0 V [B ]
Qut Qut
Som beskrivits i ekvation 5–2 och ekvation E5–2.5 kan koncentrationen cA i en sta-
tionär CSTR med en irreversibel 1:a ordningens reaktion beräknas med ekvationen:
cA 1
=
cA,in 1 + kθ
1.8
inkoncentration
1.6
1.4
1.2
koncentration
0.8
utkoncentration
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9 10
tid
1.8
inkoncentration
1.6
1.4
1.2
koncentration
0.8
utkoncentration
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9 10
tid
1.8
1.6
1.4
1.2
koncentration
1
inkoncentration
0.8
0.6
0.4 utkoncentration
0.2
0
0 1 2 3 4 5 6 7 8 9 10
tid
7.2 Arbetsmetodik
Arbetsgången för reaktorberäkningarna för dynamiska system skiljer sig inte mycket
från metodiken för att hantera stationära system. Skillnaden ligger naturligtvis i att
man måste lösa de differentialekvationer som uppkommer eftersom ’Ack’–termen finns
med. Metodiken för stationära reaktorsystem beskrivs på sidan 62.
Eftersom vi i detta avsnitt uteslutande skall använda numeriska lösningsmetoder
kommer begreppet simulera användas synonymt med ’lös differentialekvationerna’.
De olika momententen är:
1. Beskriv blandningsförhållandena i reaktorn med en reaktormodell och system-
gränser
2. Ställ upp materialbalanser för alla (oberoende) komponenter, och bestäm de
reaktorspecifika parametrarna
3. Beskriv de kemiska reaktionerna med en kinetisk modell och hastighetskonstanter
4. Kombinera reaktormodell och kinetisk modell i form av en differentiell materi-
albalans
Problemställning
I en tankreaktor sker den första ordningens reaktionen:
A−→B r = kcA
Lösning
1: Reaktormodell och systemgränser
Systemet beskrivs som en ideal tankreaktor, där in–och utströmmar passerar system-
gränsen:
[A]in,t≤0 [A] in,t≥0
Qin Qin
[A]t≤0 [A]
[B]t≤0 [A]t≤0 [B] [A]
V [B]t≤0 V [B ]
Qut Qut
2: Materialbalanser
Materialbalanserna tecknar vi som Ut + Prod = In + Ack, och eftersom volymen är
konstant så gäller:
In + Prod = Ut + Ack
dcA
A: Qcin,A + rA V = QcA + V
dt
dcB
B: 0 + rB V = QcB + V
dt
dcA Qcin,A Q
A: = − cA ( + k)
dt V V
dcB Q
B: = kcA − cB
dt V
eller, i matrisform der = Ac + Y :
5 dc 6
A
dt
der =
dcB
dt
! Q "
−( V + k) 0
A=
k − VQ
! "
cA
c=
cB
5 Qcin,A 6
V
Y =
0
5: Initialvärden
Initialvärdena får man genom att lösa ekvationen
0 = Ac + Y
Funktionsfilen blir nästan identisk med den som användes för att ta fram initialvär-
dena. Man kollar första att man verkligen har stationära betingelser, och kör med
ode45(’DynExempel’,[0 10],[9.0909;90.9091]). Man får följande figur, som visar
att man har stationäritet:
100
90
cb
80
70
60
koncentration
50
40
30
20
ca
10
0
1 2 3 4 5 6 7 8 9 10
tid
7.2 Arbetsmetodik 98
180
160
cb
140
120
koncentration
100
80
60
40
ca
20
0
1 2 3 4 5 6 7 8 9 10
tid
CHAPTER 8
In previous sections we have treated both ideal and non-ideal reactor models. By using
the CSTR, the PFR as well as the tanks-in-series model, we were able to describe,
analyze and perform calculations for a variety of open systems.
The above reactor models are just - models! They have evolved to be useful in
practice, i.e. decent descriptions of reality suited for mathematical calculations. We
have thus been able to handle relatively complex chemical system with quite simple
mathematics.
In the following sections, we shall focus on another important class of concepts
and models that often are utilized in the environmental field. One concept is dis-
persion which we shall use to calculate solute transport in porous media. Dispersion
is a mixing model that mathematically is described in the same manner as diffu-
sion. Accordingly, we hereby combine together the basic molecular mass transport
theories with environmental transport phenomena in the field scale. Applications in-
clude, among others, how organic pollutants move, sorb and degrade in groundwater
aquifers. This is a very important issue in society, and knowledge is needed in relation
to waste management, soil remediation and water conservation.
Figure 8.1 illustrates, in a simplistic fashion, how features related to substances
(pollution), soils (the reactor) and transport processes (flow) are linked. In mathe-
matical model, the chemical and physical processes involved can be linked and treated
quantitatively.
This section begins with the fundamentals of flow in porous media such as ground-
water aquifers. This leads in to the concept of dispersion. Then we take up how the
solutes, i.e. a dissolved substance and the solvent (in this case water) move together in
process called advection. The next section describes how solute transport is affected
by degradation and adsorption, introducing and using the important retardation fac-
tor.
It is inevitable to use partial differential equations in this context. The important
thing this context, however, is to understand what the equations represent, and what
boundary conditions account for. If one understands the fundamental equations and
boundary conditions, can easily perform amazing numerical model calculations using
software such as COMSOL (earlier called Femlab).
99
8.1 Water flow- Darcy’s law 100
R-värde
organisk halt
lipofilicitet
adsorption
porositet Kow
MARKEN ÄMNET
nedbrytning
vattenhalt
Grundvatten- hast.koeff
förorening
Darcy-hast. advektion hast.
ekv
VATTNET BERÄKNING
Darcy kontinuitets Femlab
dispersion ekvation
randvillkor
Figur 8.1: A simple mind map that illustrates how the properties of substances,
the soil and the hydraulic system interact to determine the rate of solute trans-
port in a porous soil system.
grun
-Δy dvat
te nyta
Δ
vattendrag
eller brunn
Δx
Figur 8.2: The principles behind Darcy’s law - gravity makes water flow from a
state with higher potential energy to a point with lower energy.
where ηe is the effective porosity. As shown in table 8.1. it may differ significantly
between the total and effective porosity. The difference is greatest for very fine-grained
materials, such as clay. This is because a large proportion of total porosity occupied
by immobile water that is very strongly tied to the earth by capillary forces. In the
fine pores, soil capillarity can be much larger than the gravitational force.
Problem
To determine the flow in an open groundwater aquifer, 3 observation wells were in-
stalled along a trans-sect of 400 m. The wells were thus at 200 m intervals. Ground-
water levels in the test point was measured at 90, 89.25 and 88.5 m relative to the
sea level, respectively. Aquifer consisted of sandy silty moraine with a hydraulic con-
ductivity of 2·10−3 cm s−1 . The effective porosity was 0.30.
Calculate:
1. Darcy–velocity
2. Flow velocity
3. The time it takes for water to be transported 400 m
Solution
The hydraulic gradient is constant along the trans-sect and amounts to:
∆y 88.5 − 90 1.5
= =− = −3.75·10−3 m m−1
∆x 400 − 0 400
The unit of K is changed, so that:
cm m s m
2·10−3 cm s−1 = 2·10−3 ·10−2 ·60·60·24·365 = 630
s cm year year
The Darcy–velocity can then be calculated as:
∆y m
vx = −Kx = −630·3.75·10−3 = 2.36
∆x year
The real flow velocity is:
vx 2.36 m
ux = = = 7.87
ηe 0.3 year
and the time it takes for water to be transported 400 m becomes:
∆x 400
t= = = 58 year!
ux 7.87
8.1 Water flow- Darcy’s law 103
lång väg
kort väg
8.2 Dispersion
8.2.1 Molecular diffusion - Fick’s law
The molecular diffusion of a a substance follows Fick’s law. In its original form, Fick’s
law is based on the assumption that the probability that an individual molecule
of a substance is identical in all directions. Each molecule does is therefore move
completely random, independent of the concentration, in any directions.
However, this reasoning leads to the conclusion that the probability is greater that
more molecules will move from a volume with many molecules to a volume with fewer
molecules, than in the opposite direction. We then conclude that there must be a
net flow of molecules from higher concentrations to lower concentrations. This means
that the net flow goes in the direction opposite to the concentration gradient.
This can be formulated as Fick’s Law, which says that, in the direction of L the
flux J will be proportional to the concentration gradient:
dc
J = −D mass m−2 s−1 (8.3)
dL
For molecular diffusion, the proportionality constant is the diffusion rate coefficient
D (e.g. m2 s).
It is obvious that the flux into a control volume can affect the concentration of a
substance in the volume. If the input flux equals the output flux, the concentration
in the volume will not change . We can, of course, conclude that the change in con-
centration per unit time in a control volume must depends on the difference between
the input and the output fluxes.
8.2 Dispersion 105
koncentrations- diffusions-
gradient riktning
dL A
c c+dc
L-riktning
Let us now establish a mass balance for the control volume shown in Figure 8.4.
Since the flux is specified per unit area, and volume of the control volume can be
written dV = AdL we get:
Ack = In − Ut
∂c
∂V = A·Jin − A·Jut
∂t
∂c ∂c ∂(c + ∂c)
∂V = −A·D − A·(−D )
∂t ∂L ∂L
∂c ∂2c
A∂L = A·D
∂t ∂L
∂c ∂2c
= D (8.4)
∂t ∂L2
Dispersion is thus a process that occurs in the macroscopic scale, cm and meters,
while diffusion takes place in a scale several orders of magnitude smaller.
Since dispersion is a stochastic process, we can describe it with the same equations
and mathematics as we use to describe diffusion. Thus, we can apply Fick’s law also
for dispersion. The only difference is that we replace the physical diffusion constant
D (e.g. m2 s−1 ) with the empirical dispersion coefficient E.
The condition ∂L
∂c
(0, t) = 0 expresses the fact that the curve is symmetrical around
the maximum value. The solution is:
M −L2 /4Dt
c(L, t) = 1 ·e (8.6)
(4πDt) 2
The spread around the symmetrical axis at x = 0 is given by the variance, σ 2 . It is
defined as the weighted deviation from x = 0. As the spread changes with time, the
variance is also time dependent:
2 +∞
L2 c(L, t)dL
−∞
σ (t) = 2 +∞
2
(8.7)
c(L, t)dL
−∞
This is a standard integral which (for positive values of a) has the solution:
2 +∞ √
2 π
b y 2 e−ay dy = 2b √
−∞ 4a a
so that:
√ √ √
π a π 1 4Dt
Numerator = 2b· √ = 2M √ · √ = 2M · = 2M = 2M Dt
4a a π 4a a 4a 4
Denominator The denominator can be treated in about the same way. The integral
from −∞ to +∞ may, because of symmetry, be divided into the two segments −∞ → 0
and 0 → ∞. Inserting the solution to the diffusion equation into the expression for
the denomimator we get:
Denominator =
2 +∞ 2 +∞ 2 +∞
M 2 M 2
c(L, t)dL = 1 ·e−L /4Dt dL = 2 1 ·e−L /4Dt dL
−∞ −∞ (4πDt) 2 0 (4πDt) 2
Koncentration av ämne
σ=0
σ = 0.25
σ = 0.50
σ=1
-2 -1,5 -1 -0,5 0 0,5 1 1,5 2
Längdskala (t ex meter)
Koncentration av ämne
σ=0
σL= √2EL/u
σ =√2E4L/u=2σL
σ =√2E16L/u=4σL
0 L 4L 16L
Transportsträcka L=u•t (t ex meter)
Figur 8.6: Illustration of how the concentration profile develops during coupled
advection and dispersion.
L = uL · t (8.9)
Combining equations 8.9 and 8.8 we see that when the peak is at L, σL is:
8
σL = 2EL/uL
Solution
The spread can be√ quantified in
√ terms of the standard deviation σ, which can be
calculated as σ = 2Dt or σ = 2Et. 95% of the pulse is contained within 2·2σ.
We can then calculate the width of the pulse (in meters) as:
√ 8
4σdif f = 4 2Dt = 4 2·(1·10−9 ·3600·24·365)·58 = 7.6 m (E8–2.10)
√ 8
4σdisp = 4 2Et = 4 2·(1·10−6 ·3600·24·365)·58 = 242 m (E8–2.11)
We can thus conclude that dispersion is the process responsible for the spread, simply
because the dispersion coefficient is greater than the diffusion coefficient.
8.3.1 Dispersivitet
The dispersion phenomenon is strongly linked to the flow velocity. The higher the
flow velocity is, the higher the E-value becomes. If one assumes that E is proportional
to uL , one can define a proportionality constant, the dispersivity α:
EL = α·uL (8.10)
The dispersivity has the unit meter, or any other unit of length.
8.3 Dispersion coefficients in groundwater aquifers 111
Empirical values 1 Table 8.2 shows example values of the dispersivity. As the
table shows, the dispersivity α is scale dependent. The greater the distance of travel,
the higher value does the dispersivity attain. The reason for this is that the larger
the distance of transport is, the greater the chance that the water encounters hetero-
geneous zones. It may be impervious lerlinser, permeable deposits and similar.
Empirical values 2 Another data set has been published by Logan (2000). here as
well, the dispersivity is dependent on the spatial scale. Uncertainties are significant,
and one must not forget that the axes are in log - scale.
8.3 Dispersion coefficients in groundwater aquifers 112
Empirical correlation In the university text book Applied Hydrogeology one finds
the following correlation:
α = 0.0175L1.45 (8.11)
where L should be given in meter. This correlation gives similar results as the empir-
ical values described above.
Pe-correlation Yet another possibility is to use correlations that uses a dimension-
less property, the Pe-number (after Péclet).
The Pe-number is defined as the ratio between a term that reflects the advective
transport, divided by the dispersive mixing.
uL L
Pe =
E
In the literature, one finds empirical equations such as:
E
= 8.8Pe1.17 , for Pe > 0.5
D
from which it is possible to calculate E by iteration. The usefulness of such equations
varies, but in any case it is possible to get an idea of the magnitude of the dispersion
coefficient. That is, provided that the systems from which they have been derived are
similar to the system of application.
CHAPTER 9
113
9.1 The continuity equation - advection and dispersion 114
dL A
t = litet
t = större
t = ännu större
corresponds to a situation where the advective transport adds substance to the control
∂c
volume, causing a positive .
∂t
Equation 9.3 should be solved with three conditions:
Boundary condition I c(0, t) = cin
∂c ∂c ∂2c
∂c
= −uL +E 2 Boundary condition II (∞, t) = 0 (9.4)
∂t ∂L ∂L
∂L
Initial value I c(L, 0) = 0, L ≥ 0
The equation has the complete solution:
! "
cin L − uL t u L/E L + uL t
c(L, t) = erfc( √ )+e L ·erfc( √ ) (9.5)
2 4Et 4Et
However, with an error of a few percent, it can be simplified to:
cin L − uL t
c(L, t) = ·erfc( √ ) (9.6)
2 4Et
The simplified solution makes it obvious that at L = uL t (i.e. at the mean convective
travel distance), the concentration is always c = cin /2! Often the development of the
concentration, as illustrated in example 9–1, is seen as a front that moves forward in
the flow direction. The curve is often referred to as a breakthrough curve.
Pulse
What happens then if a substance added to a system as a single event, a pulse. In
this case, we have the situation described in figure 8.6. The results is that the pulse
is transported while it is spreading due to dispersion.
One can easily produce an equation for the process, from equation ??. The equa-
tion gives c(L, t) for a system without advection. With advection, the shape of the
concentration curve will be the same after a certain time, but the whole curve has
moved along the direction of advective transport. We can view this as if the entire
coordinate system is moved. After time t, the whole pulse has moved a distance uL t
from the starting point at L = 0. We can now modify equation 8.6 by replacing the
spatial coordinate L with L − uL t to obtain an expression for the concentration c at
any point L and at any time t:
M 2
c(L, t) = 1 ·e−(L − uL t) /4Et (9.7)
(4πEt) 2
as a solution to:
Boundary condition I c(∞, t) = 0
∂c ∂c ∂2c ∂c
= −uL +E 2 Boundary condition II (uL t, t) = 0
∂t ∂L ∂L
∂L
Initial value I c(0, 0) = M/δ
Note that (L − uL t)2 always is positive, hence the c is symmetrical around L = uL t.
How should then the amount M be interpreted? M is the amount that is added
per m2 at a certain instant. The follows from the fact that the left hand side has the
1
unit mass per unit volume (e.g. g m−3 ) while the denominator (4πEt) 2 has the unit
length (e.g. m).
9.2 Advection, dispersion and reaction 116
0.9
0.8
Koncentration (andel av inkoncentration)
tid=5.0 år
0.7
0.6
0.5
tid=2.0 år
0.4
tid=1.0 år
0.3
tid=0.5 år
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
Längdkoordinat
Figure 9.2 shows a calculation corresponding to example 9–1, with the difference that
the chemical reaction is included.. The time considered is 5 years. Note that for
k = 0, the results are identical to the correspond calculation in example 9–1.
Normally, one can disregard the second term. Then c is:
uL L(1 − ν)
cin L − uL tν
c(L, t) = ·e 2E ·erfc( √ ) (9.13)
2 4Et
If 4kE/u2L < 0.0025 then ν ≈ (1 + 2kE/u2L ), and equation 9.13 becomes:
cin −kL/uL L − uL t(1 + 2kE/u2L )
c(L, t) = ·e ·erfc( √ ) (9.14)
2 4Et
0.9
0.8
Koncentration (andel av inkoncentration)
0.7
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10
Längdkoordinat
Figur 9.2: Calculation with equation 9.11 for the time t = 5 years.
(9.15)
c(L, t) = (9.16)
−kL ! "
cin L − uL t(1 + 2kE
u2L
) L − uL (t − tf )(1 + 2kE
u2L
)
·e uL · erfc( √ ) − erfc( √ )
2 4Et 4Et
Now, let us concider the extreme cases. If the pulse is infinitely short, i.e. tf → 0,
then the two erfc-terms will be equal and cancel out, causing c = 0 for all tf . On the
other hand, if tf → ∞ than we have a continuous source. In that case the second
term is erfc(∞) = 0. That leaves us with the first term, and the final expression for
c(L, t) is identical to equation 9.14!
9.2.3 Steady–state
If there is no chemical reaction in a system that is continuously replenished, the
concentration will eventually be equal to the input concentration to the system. In a
system of chemical reaction, however, a steady–state situation will develop, where c
is lower than cin .
Figure 9.3 shows how the concentration profile develops for a reference case (E =
1.43, uL = 0.86, k = 0.2). we see that when t → ∞, the concentration will approach
a steady–value along the entire ”reactor”, just as in the ideal PFR.
Now let us consider what happens if the dispersion coefficient E is large or small.
The most simple case is obviously E = 0.If we do not have any mixing in the direction
9.3 Adsorption and transport 119
0.9
0.8
Koncentration (andel av inkoncentration)
0.7
0.6
0.4
0.3
tiden = 5 år
tiden = 2 år
0.2
tiden = 1 år
0.1
0
0 1 2 3 4 5 6 7 8 9 10
Längdkoordinat
of flow we get:
√
uL L(1− (1+4kE/u2 )
L
lim e 2E = e−kL/uL for E → 0 (9.17)
t→∞
√
L−uL t (1+4kE/u2L )
lim erfc( √
4E
) = erfc(−∞) = 2 for E → 0
t→∞
Furthermore, if we compare the spatial dimension (L) with the temporal dimension
(t), med tidsskalan, we can again use that the time it takes for the water to move L
is t = L/uL . If t = L/uL is substituted into equation 9.13 we can conclude that:
cin −kt
c(L, t) = 2 e = cin e−kt for E → 0 (9.18)
t→∞ 2
In other words: when the dispersion is negligible, this rather complicated model is
reduced to the simple PFR model. This result had also been possible to obtain by
”removing” the dispersion term from equation 9.9.
weak, while metal as well as neutral organic molecules can bind very hard. Metal
ions sorb mainly on negatively charged surfaces of organic carbon in the soil and in
crystal lattice of clay minerals. Organic molecules, however, also binds to soil organic
9.3 Adsorption and transport 120
matter. This soil organic carbon, SOM, found in soils is a non-degradable residue of
litter and root biomass.
The content of organic material can vary from a few percent in a sandy mineral
soil to nearly 100 % in an organogenic soils such as peat soils.
The sorption reactions are quite fast in relation to the transport processes. It
is therefore reasonable to model sorption processes as partitioning equilibrium reac-
tions1 , involving the substance in dissolved form and adsorbed to some type of solid
compound.
Classic physical chemistry suggests a number of adsorption isotherms, with dif-
ferent underlying mechanistic assumptions. In practice, however, one has to assess
adsorption isotherms empirically.
A common form of adsorption isotherm is the Freundlich isotherm. It is based
on the assumption that the amount of adsorbed substances is proportional to the
concentration of the substance in the soil solution. If the relationship involves direct
proportionality, we get:
S = Kd c (9.19)
where S is the amount of sorbing solid substance (sorbate) in the soil, expressed in
unit such as mg kg−1 . Since the concentration is given in mass of substance per
unit mass of soil, the partitioning constant Kd will have a unit of the type m3 kg−1 .
However, if we want to express how much sorbed substance that is present in the soil
in relation to the amount of water in the soil, we have to take into account:
1. soil bulk densityρ ( kg
m3
soil
)
soil
m3water
2. soil water content, e.g. the effective porosity η ( m3soil
) in the saturated zone
1 http://en.wikipedia.org/wiki/Partition coefficient
9.3 Adsorption and transport 121
∂c −uL ∂c E ∂2c k
= + − c (9.27)
∂t R ∂L R ∂L2 R
The reason why R is called retardation factor is obvious. If there is an adsorption
process, then R > 1 and the adsorption process will slow down all other processes,
including the advection, the dispersion as well as the reaction!
Pulse
In the case a pulse of substance is added we get, in analogy with equation 9.7, that:
(L − uL t/R)2
M −
c(L, t) = ·e 4Et/R (9.28)
1
(4πEt/R) 2
9.3 Adsorption and transport 122
c(L, t) = (E9–2.29)
! "
cin uL L(1−ν) L − uL tν/R uL L(1+ν) L + uL tν/R
e 2E ·erfc( 8 ) + e 2E ·erfc( 8 )
2 4Et/R 4Et/R
8
Note that in the exponent uL L(1−ν)
2E as well as in ν = (1 + 4kE/u2L ) R ”disappears”
from both the numerator and the denominator.
We calculate R as:
Kd ρ 0.6·1.5
R = (1 + ) = (1 + ) = (1 + 2.53) = 3.43
η 0.35
Applying the above equations and data gives:
1
0.9
dispersion lite dispersion
0.8
Koncentration (andel av inkoncentration)
0.7
dispersion+reaktion
0.6
0.5
0.4
0.3
0.2 dispersion+reaktion+adsorption
0.1
0
0 1 2 3 4 5 6 7 8 9 10
Längdkoordinat
Something strange to think about: shouldn’t the curves for only dispersion cross at
c/cin =0.5?
9.4 The continuity equation in multiple dimensions 123
2 http://logkow.cisti.nrc.ca/logkow/
9.4 The continuity equation in multiple dimensions 124
Bioreaction engineering -
biofilms
125
10.2 Kinetic equations 126
cell
biomassa
tillväxtreglerande metaboliter
substrat metabolism
energi
övriga
substrat
.. .. ..
..
..
.. .. ..
.. ..
.. ..
..
One can therefore conclude that calculations of biological systems requires models
of both populatationens dynamic that substrate dynamics.
In the following sections, we will develop general models for the sub-systems, and
then combine the equations representing these parts to describe the whole. Mass units
used will always be in grams or kg, not molar quantities.
Reaction rate can be defined in several ways, such as per unit volume reactor per
unit volume of cells or per unit mass of cells. This section defines:
gsubstrate
µ Reaction rate per unit mass of cells
gcells ·tidsenhet
gsubstrate
rS Reaction rate per unit volume of reactor
mreactor ·unit time
3
gsubstrate
rB Reaction rate per unit volume of cells
mcells ·unit time
3
One can formulate two rate equations for these two reactions:
r1 = k1 [S][E] − k1% [SE]
r2 = k2 [SE]
One can now deduce a general expression for µ. Fits we eliminate [E] by introducing
[E] = Etot − [SE] in both rate equations:
r1 = k1 [S](Etot − [SE]) − k1% [SE]
r2 = k2 [SE]
Then, one uses that r1 = r2 :
k1 Etot [S]
[SE] =
(k1 [S] + k1% + k2 )
10.2 Kinetic equations 128
Finally, one can insert the expression for [SE] in the most simple of the rate equations,
together with µ = r2 = k2 [SE], to get:
k2 k1 Etot [S]
µ=
(k1 [S] + k1% + k2 )
As we see, the final rate equations does only include two parameters , µmax and
K1/2 . The interpretation of these parameters is shown in figure 10.3.
µmax is the maximum reaction rate. Regardless of the substrate concentration
can beµ never be greater than µmax . It is convention to use the µmax to denote this
coefficient.
K1/2 is the substrate concentration at which µ is 50% of µmax . Therefore, K1/2
is named the half rate constant. As clear from equation 10.1 and figure 10.3, the rate
equation becomes ”flater” as K1/2 increases.
Enzyme catalyzed reactions therefore follows very special kinetics. Nevertheless,
we can distinguish two special cases. At low substrate concentrations, µ depends on
[S], i.e. it is 1st order with respect to [S]. At high values of [S], the enzymatic systems
of the cells work at full capacity. Thus the conversion rate is independent of the
substrate concentration, i.e. the reaction is of 0 order. This is illustrated in Figure
10.4.
From equation 10.1 we can identify the two extremes:
µmax [S]
Low substrate concentration µ"
K1/2
High substrate concentration µ" µmax
One should be aware that the experimentally determined equations and parame-
ters can not be extrapolated to other system. Although micro-organisms of the same
type and strain can behave differently in different environments, e.g. by forming
aggregates affecting the mass transport.
10.2 Kinetic equations 129
Reaktionshastighet µ
1.0
µ = 1*[S] / (1+[S])
µ = 1*[S] / (4+[S])
0.5
K1/2=4
K1/2=1
0.0
0 2 4 6 8 10
Substratkoncentration [S]
Figur 10.3: Plot of the kinetic equation for two cases. Note that the parameter
K1/2 corresponds to the substrate concentration where rS is 50% of µmax .
Reaktionshastighet µ
1.0
0:e ordningen:
µ ≈ µ max
0.5
1.a ordningen:
µ ≈ µ max[S]/K1/2
µ = µ max[S] / (K1/2+[S])
0.0
0 2 4 6 8 10
Substratkoncentration [S]
Figur 10.4: Illustration of the rate equation assumes the form of a 1st-order
expression at low substrate concentration, and a 0th-order reaction at high sub-
strate concentration.
10.2 Kinetic equations 130
µmax [S]
rS = −µ·[X] = − · [X] (10.2)
[S] + K1/2
Solution
For a batch reactor, Prod = Acc, which means that:
d[X] d[S]
= rX ; = rS
dt dt
where
µmax [S]
rX = (µ·Y − kd )·[X] =( ·Y − kd )·[X] (E10–1.4)
[S] + K1/2
µmax [S]
rS = −µ·[X] =( )·[X] (E10–1.5)
[S] + K1/2
These equations are put in a MATLAB-file:
100
90
80
Substratkoncentration S
70
Koncentration (g per m3)
60
50
40
Cellkoncentration X
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10
Tid (dygn)
You can clearly see how the microorganisms grow as long as the supply of substrate
is high. As [S] is decreases, the growth becomes lower than the mortality, and [X] is
decreases as well. Note that the initial value of X must be> 0 for the reaction to get
started.
10.3 Biofilms 132
10.3 Biofilms
In many technical applications the reaction takes place in biological films. It is very
common in water treatment contexts. These biofilms may grow on carriers made out
of plastics or beds of stones in a reactor.
Biofilms offer some practical advantages. They counter that the level of suspended
solids becomes too high in the water, they are surprisingly resistant to harmful pol-
lutants and they are physically stable.
A basis for biofilm process is that the substrate is transported from the surrounding
liquid bulk into the film. Here it is converted into products, e.g. biomass. Thereby
, a oncentration gradient is created from the biofilm surface in towards the carriers
wall. This creates a driving force for a flux from the liquid to the biofilm (Figure
10.7). This flux in each point is proportional to the gradient magnitude.
Intuitively, we may also find that the faster the substrate is transported into the
biofilm, the stronger the gradient and the flux become
Figure 10.7 shows schematically how the concentration profile in a biofilm can
look. The main principle is that substrates diffuse from the biofilm surface, through
the film toward the carriers surface. The driving force is the concentration gradient
as incurred by the substrate consumed in the biofilm.
The part of the biofilm that is supplied by substrate is called the penetration depth,
LP . The fraction of the film that is supplied is called the degree of penetration and is
defined as the ratio between the penetration depth and the thickness of the biofilm,
LP /LB .
10.3 Biofilms 133
C=C0
C=0
L=0 LP LB
(a) Deep film, non-penetrated. Parameter B < 1.
C=C0
C=0
L=0 LP=LB
(b) Exactly penetrated. Parameter B = 1.
C=C0
C>0
L=0 LB
(c) Thin film, penetrated. Parameter B > 1.
C=C0
L=0 LB
1. The reaction rate should be defined in terms of units of volume biofilm. Hence,
we use rB ( m3 gsubstrate
·unit time
)
biofilm
d[X]
2. Cell mass is constant, so that = 0 everywhere in the reactor.
dt
As previously mentioned, the kintetiska parameters µ and K1/2 specific to a par-
ticular system. In this section, we will simply assume that µ have been determined
for the current biofilm system.
If we denote cell mass per unit volume of biofilm with XB the rate equation for a
biofilm becomes:
µmax [S]
rB = −µ·XB = − XB (10.4)
[S] + K1/2
This is analogous with equation 10.2. In practice, however, one determines the prod-
uct µ·XB , and the rate equation can be written:
kB [S]
rB = − (10.5)
[S] + K1/2
where the rate constant kB , has the unit g of substrate per unit volume of biofilm
and time.
As mentioned earlier, the conversion of substrate depends on substrate concentra-
tion only at low values of mconc (S). We may therefore identify two special cases
with respect to kinetics:
kB [S]
Low substrate concentration: rB " − = −k1 [S]
K1/2
High substrate concentration: rB " −kB = −k0
rB = −k0
∂2c
D = k0
∂L2
10.3 Biofilms 136
Now dimensionless quantities are introduced for concentration and for length. If the
concentration at the surface of the film is c0 and the biofilm thickness is LB , then:
c
c∗ = ; c = c∗ ·c0
c0
L
L∗ = ; L = L∗ ·LB
LB
If we combine all these equations, the continuity equation becomes:
∂ 2 c∗ ·c0 ∂ 2 c∗ L2B ∂ 2 c∗ k0 ·L2B
D = k0 ; D = k0 ; =
∂(L∗ ·LB )2 ∂L∗ 2 c0 ∂L∗ 2 D·c0
It is now possible to ”clean up” the right hand side by introducing the variable:
2D·c0
B2 = (10.6)
k0 ·L2B
and we get:
∂ 2 c∗ 2
= 2 (10.7)
∂L∗ 2 B
We will now solve equation 10.7 with two boundary condition. We utilize that:
L∗ = 0 c∗ = 1 Concentration known at the surface of the biofilm
∗
L∗ = 1 ∂c
= 0 Flux at the carriers surface is zero
∂L∗
Two integrations of equation 10.7 gives:
∂c∗ 2 ∗
= L + H1 (10.8)
∂L∗ B2
L∗2
c∗ = + H1 ·L∗ + H2 (10.9)
B2
We use the second boundary condition with equation 10.8, and thereafter the second
boundary condition with equation 10.9 to get:
2 2
0 = 2 + H1 H1 = − 2
B B
1 = H2
This is put into equation 10.9 and we get at complete expression for the concentration
profile in the biofilm:
L∗2 2L∗
c∗ = 2
− 2 +1 (10.10)
B B
Obviously, the concentration profile is a relatively simple 2nd order equation.
There is one very important special case. In a situation where the substrate just
is consumed in the film, so that the substrate concentration c∗ = 0 exactly when
L∗ = 1. In that case, equation 10.10 gives:
L∗2 2L∗ 1
c∗ = 2
− 2 +1 ; 0=− 2 +1 ; B =1
B B B
2D·c0
This means that exactly when B 2 = = 1 (i.e. when B=1) the substrate
k0 ·L2B
concentration is 0 at the surface of the carrier.
10.3 Biofilms 137
J = k0 ·LB (10.11)
The flux depends only by the rate constant and the film thickness. Hence the flux, i.e.
the diffusion rate into the film is independent of the diffusivity of the substrate in the
biofilm itself, provided that the reaction is independent of the substrate concentration.
∂ 2 c∗ 2
= 2 (10.12)
∂L∗ 2 B
For a deep film, this transfer equation should be solve with three conditions:
L∗ = 0 c∗ = 1 Concentration known at the surface of the biofilm
L∗ = LP c∗ = 0 Flux at the carriers surface is zero
∗
L∗ = LP ∂c = 0
Flux at the penetration depth is zero
∂L∗
One integration gives:
∂c∗ 2
= 2 L∗ + H1
∂L∗ B
10.3 Biofilms 138
Tabell 10.1: Equations for fluxes into biofilms for different cases.
B≥1 Tunn film J = k0 ·LB
0:e ordningen √
B≤1 Tjock film J = 2D·k0 ·c0
√
1:a ordningen Generellt J = c0 D·k1
L∗2
c∗ = + H1 ·L∗ + H2
B2
If we first insert the first boundary condition and then the second boundary condition
in the above equation we get:
1 = 0 + 0 + H2 H2 = 1
2LP 2LP LP 2
0= 2
− 2 +1 0=− +1 LP = B H1 =
B B B2 B
This can be summarized in the polynomial equation describing the concentration
profile in a deep film:
L∗2 2L∗
c∗ = − +1 (10.13)
B2 B
The flux depends on all possible parameters, except the thickness of the film, which
is logical.
10.3 Biofilms 139
Solution
The unit mg L−1 is equivalent to g m−3 . Since c0 >> K 12 we can assume 0th order
reaction in the entire biofilm. and thus k0 = kb . This is, however, an approximation
as clear from the figure below, showing the concentration profile: c = c0·L∗ 2 − 2L∗ + 1
(always valid when B = 1):
25
20
Substratkoncentration [S] (mg/L)
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionslös längd L* = L/LB
For a 0th order reaction, the rate coefficient k0 = kb ., and since the film is fully
penetrated, B ≥ 1. Since:
2D·c0
B2 = ≥1
k0 ·L2B
r = −J · A
where A is the number of m2 biofilm surface per m3 reactor. Thereby, we can easily
develop a relevant mass balance equation:
Solution
The concentration in the reactor will be 20% of the input concentration, i.e. 10
mg m−3 . In the absence of external mass transfer resistance, this is also the concen-
tration at the surface of the biofilm. c0 . First, one must check if the film is penetrated
or not:
: :
2D·c0 2 · 3·10−10 · 10
B= = = 1.41
k0 ·L2B 0.3(100·10−6 )2
When you count on other than pipes, one must be careful to define the Re in the right
way. In a packed bed the characteristic length of Re is defined as the particle radius.
If the particles in the bed are of different sizes one base the equivalent particle radius
on the material surface area:
7
1 Ap
L= (10.20)
2 π
10.4 External mass transfer 142
Entity: Unit
η Effective porosity –
v Darcy–velocity m s−1
L Characteristic length m
D Diffusivity m2 s−1
ν Kinematic viscosity m2 s−1
km Mass transfer coefficient m s−1
J=
= Flux into the laminar boundary layer
= Flux out from the laminar boundary layer
= Flux into the biofilm
10.4 External mass transfer 143
Although there are countless possibilities to perform calculations on this type of sys-
tems, the discussion is left at this point.
10.4 External mass transfer 144
6
Gränsfilmstjocklek (m)
Här är Re > 55
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Darcy-hastighet (m/minut) genom bädden