CHE3065S

Assignments

page 1

Assignment 4 : Solution of non-linear equations and

non-linear least squares
1

Root finding (isothermal and adiabatic flash)

The Rachford-Rice method is very efficient in solving isothermal flash calculations1 and is given
by
n
X

zi (1 Ki )
=0
1 + (Ki 1)
i=1
zi
= V /F, V = F , L = F V, xi =
, yi = Ki xi
1 + (Ki 1)
Q = HV V + HL L HF F (energy requirement)
f () =

C8
C9
component iC4 nC4 iC5 nC5 C6 C7
kmol/h
14 448 36 15
23 39.1 272.1 31
Assume that the K values are independent of composition.
(a) Determine the isothermal flash composition of the mixture at P=80 psia, T=82.2 C
(b) Solve the the energy balance and the Rachford-Rice equation simultaneously for the adiabatic flash composition. The feed now enters at 130 psia and 101 C (essentially saturated
liquid). What is the heat duty and final temperature/composition of the flashed stream if
the pressure in the flash tank is 80 psia?
The data for this system can be obtained from COCO/Chemsep data base. You can also use
COCO/Chemsep to check your answers.

Non-linear equations (adiabatic flash)

The Rachford-Rice method fails (becomes ill-conditioned) when K 1. An alternative method

is to formulate the flash calculation2 as follows ;
component flows :
li + v i f i = 0
P
P
P
L = li , V = vi , F = fi
equilibrium relations :
y i = Ki x i
V
vi Ki li = 0
L
energy balance :
V HV + LHL F HF + Q = 0
1
2

JD Seader and EJ Henley, (1998), Separation process principles, Wiley (Chapter 4, Table 4.4)
based on HA Kooijman, R Taylor, (1998), The Chemsep book.

CHE3065S

Assignments

page 2

Where Q is the energy removed from the system (Q = 0 is adiabatic). Here if T,P are
specified, the the value of Q will be obtained. If only P is specified (typical of an adaibatic
flash), then Q = 0 and the temperature needs to be calculated. These equations are to be
solved for the vector x = {l1 , ..., lN , vi , ..., vN , T or P or Q} for a N component system.
(a) Repeat the isothermal flash of problem question 1. Assume the feed enters at the same T
(and P)
(b) The feed now enters at 130 psia and 101 C (essentially saturated liquid). What is the heat
duty and final temperature/composition of the flashed stream if the pressure in the flash
tank is 80 psia?
This method is applicable to distillation column stage by stage calculations. The Rachford-Rice
method remains faster and more efficient for many species when the system is isothermal.

Non-linear equations : chemical equilibrium

Chemical equilibrium can be solved by minimising the gibbs free energy of the system. The
advantage of this method is that it does not require a knowledge of the chemical reactions. It
is sufficient to list the chemical species and their associated gibbs free energy. The theoretical
development3 is as follows :
Formulation of the element balance which form the constraints of the system;
N
X

ni aik Ak = 0

where N is the number of species, ni moles of species i, aik number of atoms of k th element
of the i th molecule and Ak is the total amount of element k present in the mixture. Introduce
the lagrange multipliers k (see optimisation course), one for each element and summing over
all elements;
!
w
N
X
X
i
ni aik Ak = 0
k

where w is the number of elements in the system. This equation is combined with the Gibbs
energy of the system;


G t T,P = G(n1 , ..., nN )
!
w
N
X
X
F = Gt +
i
ni aik Ak
k

A minimum of F (and thus G) occurs when the derivatives with respect to ni are zero;


F
ni

i +

w
X


=

T,P,nj

k aik = 0,

G t
ni


+
T,P,nj

w
X

k aik = 0

(i = 1, 2..., N)

k
3

JM Smith and HC van Ness, (1975), Introduction to Chemical Engineering Thermodynamics, 3rd ed.,
McGraw-Hill, Chapter 9, p424-428 (Also in Perrys Chemical Engineering handbook, its not in Sandler !)

CHE3065S

Assignments

page 3

Expanding the chemical potential;

ai )
i = Gi + RT ln(
i = Gi + RT ln(fi )
for gas phase reactions with the ideal gas state as 1 atm. Gi is set to zero for all elements
in their standard state, then for compounds Gi = Gfi , the standard Gibbs free energy of
formation. With fi = yi i P, then
i = Gfi + RT ln(yi i P )
Gfi + RT ln(yi i P ) +

w
X

k aik = 0,

(i = 1, 2..., N)

Gfi

is zero for elements and P is in atm. There are N equilibrium equations, one for each
species and there are w material balance equations giving a total of N + w equations. There
are N unknown ni0 s and there are w unknown 0i s that have to be found.
Write a programme that calculates the equilibrium composition at 1000K and 1 atm for a system
containing CH4 , H2 O, CO, CO2 and H2 . Initially there is present 2 gmol of CH4 and 3 gmol of
H2 O.
species
CH4
H2 O
CO
CO2
Data :

Gf i [cal/gmol] 4610 -46030 -47940 -94610

Solving mass balances

f6

Recycle

f5

f7
f1

Purge: (B)
purge ratio=beta

f2

mixer: (A)

f3

Feed
Inerts=0.004%
Reactor:
H2 conversion = 0.6
feed mole fractions
CO2 = 0.28
H2 = 0.70
Inerts = 0.02

Condenser:
Vapour:CO2,H2,Inerts
Liquid:methanol,water
separation factor=alpha

f4
Products (desired)
methanol=155kmol/h

Figure 1: Process block flow diagram of the poduction of methanol from carbon dioxide and
hydrogen.
Figure 1 is the block flow diagram of the methanol production process outlined in Felder and
Rousseau4 (2000). Methanol can be produced by the reaction of CO2 with H2 to produce
CH3 OH and H2 O as given by the following reaction
CO2 + 3H2 CH3 OH + H2 O
4

Richard M. Felder and Ronald W. Rousseau, (2000), Elementary principles of chemical processes, 3rd edition,
Wiley. Chapter 4, example 4.7-3 on page 139. Also see chapter 10 for a more elaborate discusion on computers
and process simulators.

CHE3065S

Assignments

page 4

The process contains both a purge and recycle streams. The design information for the process
can be summarised as follows:
1. The feed contains only CO2 , H2 and inerts (I) and no methanol or water. The composition
of the feed is unknown, other for the condition that it will contain 0.4mol% inerts. (very
unusual, but so be it)
2. The feed to the reactor will have the following composition : 28mol% CO2 , 70mol% H2 ,
2mol% inerts and no methanol and water. The conversion of H2 across the reactor will
be 60%.
3. The condenser will carry out a perfect separation between the gases and liquids, thus the
liquids, methanol and water leave in one stream and the gases, CO2 , H2 and inerts leave
in another stream.
4. The gas stream is recycled to the process on order to re-use the unreacted CO2 and H2 .
However, since the inerts are not reacting away, a purge stream is also needed to remove
the inerts.
This information has been entered into the block flow diagram in Fig 1.
Solve the mass balance by developing a system of linear and non-linear equations that represent the unit operations, stream splitters and mixers. To make this work, define the following
functions
function f=splitF(Fin, Fout1, Fout2,Fratio) where Rratio is the stream split ratio
function f=splitX(Fin,Fout1,Fout2,Xratio) where Xratio is a vector of split ratios
function f=reactor(Fin,Fout,conversion) were conversion is a vector of conversions for each
reaction and the flows are connected by a stoichiometric table
function f=mixer(Fin1,Fin2,Fout)
Each function must be written in the form f (F ) = 0. Now simply go around the flow sheet
connecting all the streams together using the defined function to yield an overall function block
[fspliF fspliX fr eactor ...]T = 0 that will solve for the missing flow rates. Note that since the solver
is non-linear, it is possible to determine the non-linear factors such as conversion and split ratios,
but be warned, an initial guess is still needed and you may specify flow rates of a output stream
that cannot be obtained from feasible non-linear factors, for example you might need negative
conversions or conversions greater than 1.
At this point you are only one step away from a flow sheet simulator like COCO, all that is still
needed is an energy balance and a phase calculation procedure all of which are also part of this
assignment!!!!

CHE3065S

Assignments

page 5

Solving highly non-linear equations - adsorption equilibrium

The Nitta multicomponent isotherm5 is given by

q1

P qi ni n1
K1 Q 1
Q
q2


y2 P =
P qi ni n2
K2 Q 1
Q
q1
x1 =
q1 + q2

y1 P =

where yi mole fraction in gas phase, xi mole fraction in liquid phase, P total gas phase pressure
[kPa], qi amount adsorbed on the solid [mol/L], Ki equilibrium adsorption constant [kPa1 ], ni
number of sites occupied by one mole of molecules, Q number of sites per volume of adsorbent
[mol/L] and qsi = Q/ni the saturation loading for species i . Make a xy diagram (x1 vs y1 ) as
function of pressure just like in distillation using the following data Q = 2 [mol/L]. Use the
following total pressures : 1, 10, 100 kPa
parameter
K
n
5

component
1
2
1
2
2
0.5

T Nitta, T Shigetomi, M Kuro-oka, T Katayama, (1984), An adsorption isotherm of multisite occupancy

model for homogeneous surface, J.Chem. Eng. Japan, 17(1), p39. & DD Do, (1999), Adsorption analysis :
equilibria and kinetics, Cambridge University Press.

CHE3065S

Assignments

page 6

Non-linear least squares : The single component Nitta isotherm6

The complete Nitta isotherm is given by

q
n EXP

p=
q
K n qs 1
qs



E
qn
RT

where the notation is the same as problem 5 and p is the partial pressure of the adsorbing
species. Determine K, n, qs , E from the data below at a temperature of 50 C. Estimate the
variance in the parameters in each case and comment on the results.
Partial pressure (atm)
0.0000
0.0010
0.0021
0.0032
0.0043
0.0055
0.0068
0.0081
0.0096
0.0111
0.0129
0.0148
0.0169
0.0194
0.0224
0.0260
0.0307
0.0373
0.0479
0.0705
0.5246

Nitta et al., (1984), ibid

Loading on the solid surface (mol/L)

set 1
set 2
0.0000
0.0250
0.0500
0.0749
0.0999
0.1249
0.1499
0.1748
0.1998
0.2248
0.2498
0.2747
0.2997
0.3247
0.3497
0.3746
0.3996
0.4246
0.4496
0.4745
0.4995

0.0000
0.0251
0.0504
0.0754
0.1012
0.1238
0.1544
0.1816
0.2061
0.2268
0.2477
0.2853
0.3022
0.3269
0.3528
0.3754
0.4012
0.4163
0.4574
0.4735
0.5035