Académique Documents
Professionnel Documents
Culture Documents
temperatures, stream compositions, stream flow rates, and flow L and vapor molar flow V are coming from the stage j
heat-transfer rates at each stage by solving energy-balance, from a VLE. The feed stream F is in the 43rd tray and vapor
material-balance, and equilibrium for each stage. (SV) side stream are planned at the top tray for Whole
Unfortunately, these equations are strongly interacting non- Naphtha, 12th tray for SRK, 25th tray for LDO, 35th tray for
linear algebraic equations. So, solution procedure becomes HDO and 48th tray for AR. Normally, for obtaining the model
difficult. till now the value of vapor liquid equilibrium ratio is
A proper model is necessary in sizing and arrangement of considered to be constant.
processing equipment for performance analysis. Although
Fig. 1, represents vapor liquid equilibrium case where the
various methods are available for solving multi component
trays are numbered from top to bottom. Feed of molar
separation problems, approximate methods are generally used
component is entering stage j with a flow rate Fj, with
for design to establish optimal design conditions. Also,
composition zi,j in mole fractions of component i and molar
process synthesis is studied to determine optimal separation
enthalpy hFj.
sequences with initial approximations for rigorous, iterative
methods. The mathematical model describing the separation 2.1 Assumptions
process like distillation consists of four sets of basic
equations, called mesh equations which are Mass balance To obtain the model reported in Jackson et al., a 48
equation, Equilibrium relations, Sum (or conservation) tray column model is chosen.
equation and Enthalpy balance relations. The residence time on each plate is such that
Vj Liquid from
stage above equilibrium is attained between the liquid and the
Vapor side Lj-1 vapour. Hence Yn = Kn * Xn
stream
Wj The flow rate of the side streams is equal to the flow
rate of the reflux which is assumed from Seader,
xi, j-1
yi, j J.,(2011)
hL j-1
hV j From the example 10.1, sighted in Seader, J.,(2011)
the reflex ratio is considered to be 2.
Head
L The flow rate of the vapour throughout the column
Valve is considered to be the summation of the three side
Feed F Heat Transfer streams and reflux of the three side streams. i.e.,
Fj Stage j Vj =W12+ W25+ W35+reflex ratio*( W12+ W25+ W35).
zi, j qj
hF, j
The liquid flow rate is also considered to be constant
throughout the column.
xi, j
The mass balance doesn’t match for the inflow rates
hL j and out flow rates reported in Jackson et al,. Hence
Valve
V by considering only the feed flow rate and the feed
compositions new values of side streams are
yi, j+1 calculated and liquid flow rate are calculated.
h V j+1 With reference to Jackson et al., only vapour side
Liquid side streams are present. The liquid side stream flow rate
stream is zero.
Uj
Vj+1 2.2 Process variables
2
IFAC ALCOSP 2016
June 29 - July 1, 2016. Eindhoven, The Ganesh Muralidharan et al. / IFAC-PapersOnLine 49-13 (2016) 199–204
Netherlands 201
molar flow rate Vj to be sent to stage j - 1 or, if j = 1, to leave As sighted in Seader, J.,(2011), the overall material balance
as a product. The flowrates of various parameters and equation is given in Eq. 6.
variables withdrawn from the column trays considered are Vj + Fj-1 + F1 + F2 = L j-1 + U j-1 + Wj-1 + U 2 + W2
summarized in Table 1. (6)
+ U1 + W1 + V1
Table 1. Flow rates
L j-1 = U j-1 + Wj-1 + U 2 + W2 + U1 + W1 + V1 - Vj
(7)
Variables Values Trays (j) - Fj-1 - F1 - F2
F43 = 993.7 (m3/hr) 43
Feed flow rates j = 1,2,3,… L j-1 is represented in general form as shown in Eq. 8.
Fj = 0 (m3/hr)
….42,44,45…48
L j-1 = Vj + Fj-1 + F1 + F2 - U j-1 - Wj-1 - U 2 - W2 - U1
Overhead j = 1,2,3… (8)
Vj = 1275.9 (m3/hr)
Flow rate ….42,44,45…48 - W1 - V1
Bottom Flow j = 1,2,3,…
Lj = 310.03 (m3/hr) j-1
Rate ….42,44,45…48 L j-1 = Vj + (Fm -Wm -U m )-V1 (9)
Vapor Side j = 1,2,3,…11,13,…
Wj = 0 (m3/hr) m=1
Stream & ..24,26,….34,36...48 Similarly, Eq. 9 can be expressed for Lj as,
reflex flow W12 = 138.1(m3/hr) 12 j
rate(assumed W25 = 240.5(m3/hr) 25 L j = Vj+1 + (Fm -Wm -U m )-V1 (10)
to be same) W35 = 46.7(m3/hr) 35 m=1
Liquid Side Substituting the expression from Eq. 9 and Eq. 10 in Eq. 1,
Uj = 0 (m3/hr) j = 1,2,3,………48
Stream we get,
zi,j = 0
j-1
z1,43 = 0.260
i = 1,2,…5; j (Fm -Wm -U m )-V1 x i,j-1
V +
Feed z2,43 = 0.139 m=1
j=
Composition z3,43 = 0.242 Aj
1,……42,44,….48
z4,43 = 0.047
j
z5,43 = 0.312 - Vj+1 + (Fm -Wm -U m )-V1 +U j +(Vj +Wj )K ij x i,j (11)
The MESH equations for stage j is given from Eq. 1 to Eq. 5, m=1
The material balance: Bj
M ij = L j-1x i,j-1 + Vj+1Yi,j+1 + Fjz i,j - (L j +U j ) x i,j
(1) + Vj+1K i,j+1 x i,j+1 + Fjzi,j = 0
- (Vj +Wj ) yi,j = 0 Cj Dj
The equilibrium relationship for each stage is given as,
A j xi,j-1 +B jxi,j +C jxi,j+1 = D j (12)
Eij = yij - Kij xij =0 (2)
j-1
where, Kij is the Vapor Liquid Equilibrium (VLE) ratio. A j = Vj + (Fm -Wm -U m )-V1 ; 2 j 48 (13)
The summation equations are defined by, m=1
n j
(Sy ) j = yi,j -1 = 0 (3) j+1 (Fm -Wm -U m )-V1
V +
i 1 Bj= - m 1
; 1 j 48 (14)
n
+U j +(Vj +Wj )K ij
(Sx ) j = x i,j -1 = 0 (4)
i 1 C j = Vj+1Ki,j+1 ; 1 j 47 (15)
The energy balance equation is given as,
Dj = - Fz ; 1 j 47 (16)
H j = L j-1h L F-1 + Vj+1h V j+1 + Fjh Fj - (L j +U j ) h L j - j i,j
3
IFAC ALCOSP 2016
June 29 - July 1, 2016. Eindhoven, The Ganesh
202 Muralidharan et al. / IFAC-PapersOnLine 49-13 (2016) 199–204
Netherlands
Xi,1 =(D1 -C1xi,2 )/B1 (18) term. Here, j represents the corresponding tray
number of each component.
For simplification the Eq. 18 is rewritten in the form as, Step 4: VLE values are then estimated from Step 3 of
xi,1 = (Q1 -P1xi,2 ) (19) individual trays upto feed tray.
where, Step 5: The remaining coefficients in P matrix are
calculated based on the previous values of Qj-1 and
P1 =C1 /B1 (20) composition x.
Q1 =D1 /B1 (21) Step 6: With the determined values in Step 5, the values of
Q matrix is calculated recursively from feed tray
Similarly, second row of matrix in Eq. 17 is expressed as in
to the Nth tray.
Eq. 22,
Step 7: Now with P and Q matrices, the remaining values
D 2 -A 2 Q1 C2 of VLE (from Feed tray to Nth tray) are estimated
x i,2 = - - x i,3 (22) precisely.
B2 -A 2 P1 B2 -A 2 P1
From the matrix in Eq. 25, it is clear that the values of Q are
where, zero upto the feed tray. The remaining coefficients of Q
D -A Q varies from the feed tray that depends on Aj, Bj, Dj and the
Q2 = 2 2 1 (23) previous values of Qj-1 and Pj-1. Likewise, the values of P are
B2 -A 2 P1 obtained until the feed tray as given in Eq. 29,
C2 xj
Pj = - ; 1 j 42 (29)
P2 = (24)
x j+1
B2 -A 2 P1
Similarly, the remaining rows in the matrix shown in Eq. 17 It is assumed that the vapor is entirely refluxed in the top tray
are modified as in Eq.25, of the column. So, B1 can be expressed as in Eq. 30.
x i,1 Q1
1 0
x i,2 Q 2
P1 ... ... ... ... ... ... 0 B1 = -V1 (30)
0 1 P2 ... ... ... ... ... .. 0
x i,3 Q3 From Eq. 20, C1 is calculated as given below,
0 0 1 P3 0 0 ... ... ... 0 (25)
C1 = B1P1
: ... : (31)
: : nd
: ... : From Eq. 15, we determine the values of VLE for the 2 tray
×
: .... 0 as,
: :
:
.... 0
K 2 =-C2 /V2 (32)
: 0 1 PN-2 0
0 x i,N-2 Q N-2 From this the values of Aj, Bj, Cj and Dj are calculated till the
0 1 PN-1
x i,N-1 Q N-1 Feed tray.
0 0 ... ... ... ... 0 0 0 1 j
x i,N Q N
Hence, in general the expression for computing P j and Qj are B j =-[Vj+1 + (Fm -Wm -U m )-V1 +U j
m=1 (33)
given in Eq. 26 and Eq. 27 respectively.
Cj +(Vj -Wj)K i,j ] ; 1 j 43
Pj =
B j -A jPj-1
(26) From the predetermined values of Bj and Pj, the remaining
VLE constants are estimated as,
D j -A jQ j-1 K j+1 =Pj (Bj -A jPj-1 )/Vj ; 1 j 42 (34)
Qj=
B j -A jPj-1
(27) By using the values obtained in Eq. 34, the Cj are obtained
for first tray upto the tray above the feed as in Eq. 35,
xi,j =Q j - (Pjxi,j+1 ) (28) C j =VjK j+1 ; 1 j 42 (35)
4. PROPOSED METHOD USING REVERSE After the feed tray the values of P depends upon the previous
CALCULATION PROCEDURE IN THOMAS values of Q as given in Eq. 36,
ALGORITHM Pj = (Q j -x j )/x j+1 ; 43 j 47 (36)
Steps involved in the proposed algorithm for estimation of Q j = (D j -A jQ j-1 )/(B j -A jPj-1 ) ; 43 j 48 (37)
VLE constants using Reverse calculation of popular Thomas j
algorithm is described below: B j = -[Vj+1 + (Fm -Wm -U m )-V1 + U j
Step 1: Obtain the steady state composition of each m=1 (38)
component.
Step 2: From the steady state find the values of P matrix + (Vj -Wj )K i,j ] ; 44 j 48
until feed tray, which depends only on the VLE constants are precisely calculated as given in Eq.39 for
compositions. the trays below the feed by incorporating all the previous
Step 3: From matrix P obtained in Step 2, the values of Aj, values of Pj, Qj and Bj obtained in Eq, 36, Eq. 37 and Eq. 38
Bj, Cj and Dj matrices are calculated upto the feed respectively.
4
IFAC ALCOSP 2016
June 29 - July 1, 2016. Eindhoven, The Ganesh Muralidharan et al. / IFAC-PapersOnLine 49-13 (2016) 199–204
Netherlands 203
Fig. 4. Composition of SRK on each Tray 1-48 Atmospheric Residue (AR) 0.13
5
IFAC ALCOSP 2016
June 29 - July 1, 2016. Eindhoven, The Ganesh
204 Muralidharan et al. / IFAC-PapersOnLine 49-13 (2016) 199–204
Netherlands
Table 3: VLE values estimated by proposed method The variability of VLE ratio in each tray of the column
Tray VLE estimated by reverse calculation procedure is shown in Fig. 8.
No. Naphtha SRK LDO HDO AR It is observed that the VLE values are not constants for a
1 0.00 0.00 0.00 0.00 0.00 component in each trays and at the feed tray, the values are
maximum for five components of the crude distillation
2 1.60 0.73 0.98 0.99 0.98
column and a precise values of VLE on individual trays are
3 1.01 0.69 0.43 0.99 0.97
obtained.
4 1.01 0.85 0.75 0.98 0.99
5 1.01 0.80 0.82 0.21 0.99 Table 4: Comparison of composition values for each
6 1.01 0.72 0.77 0.75 0.97 component
7 1.01 0.79 0.75 0.85 0.99 Composition
Sl. Tray of
8 1.01 0.90 0.79 0.93 0.98 Component Jackson Proposed
No. Withdrawal
9 1.03 0.85 0.91 0.91 0.94 et al. Algorithm
10 1.02 0.51 0.30 0.97 0.93 1 Naphtha 1 0.1750 0.2611
11 1.03 0.80 0.66 0.95 0.75 2 SRK 12 0.1690 0.1742
12 0.97 0.66 0.39 0.94 0.04 3 LDO 25 0.2560 0.2783
13 1.15 1.00 1.16 0.93 12.38 4 HDO 35 0.0570 0.0572
14 1.09 1.05 0.82 0.90 0.67
5 AR 48 0.2530 0.2621
15 1.05 1.01 0.89 0.91 0.79
16 1.05 1.01 0.90 0.90 0.76
Table 4, shows a comparative study of composition for five
17 1.10 1.03 0.91 0.91 0.81 components mainly, Whole Naphtha, SRK, LDO, HDO and
18 1.06 1.07 0.91 0.89 0.72 AR withdrawn on trays 1, 12, 25, 35 and 48 respectively. It
19 1.08 1.08 0.90 0.91 0.11 is observed that with variation in VLE parameters, more
20 1.06 1.09 0.87 0.90 0.81 precised values are obtained using proposed algorithm.
21 1.08 1.11 0.70 0.86 0.83
6. CONCLUSIONS
22 1.08 1.04 0.85 0.86 0.77
23 1.11 1.08 0.87 0.75 0.78 This work summarizes a new algorithm to develop a model
24 1.12 1.06 0.86 0.80 0.79 for multicomponent distillation column process. An
25 1.02 0.97 0.70 0.88 0.69 appropriate composition profile for various mixtures mainly,
Naphtha, SRK, LDO, HDO and AR were estimated by
26 1.25 1.10 1.02 0.87 0.78
obtaining the VLE values from the MESH equations by using
27 1.27 1.13 1.12 0.85 0.77
reverse calculation of Thomas algorithm for each tray of the
28 1.27 1.11 1.22 0.85 0.78 distillation column. A comparative study is carried out for
29 1.27 1.33 1.31 0.85 0.75 the steady state composition profile of the crude distillation
30 1.35 1.44 1.63 0.83 0.73 column with the results reported in Jackson et al.
31 1.36 1.65 1.43 0.83 0.75
REFERENCES
32 1.41 1.96 1.76 0.82 0.75
33 1.56 1.79 1.66 0.83 0.74 AKPA, Jackson Gunorubon and UMUZE, Omonigho
34 1.52 1.80 1.65 0.73 0.75 Diepriye, (2013), Simulation of a Multi-component
35 1.48 1.74 1.56 0.67 0.74 Crude Distillation Column, American J. of Scientific
36 1.63 1.89 1.76 0.90 0.79 and Industrial Research, vol.4, no.4, pp.366-377.
Charles, D. Donald, (1997), Fundamentals of
37 1.62 2.00 2.02 0.93 0.74
Multicomponent distillation, McGraw-Hill.
38 1.69 2.27 2.55 0.96 0.73 Seader, J. D, Ernest J. Henley, Keith Roper D., (2011),
39 1.75 2.33 6.19 1.15 0.72 Separation Process principles: Chemical and
40 1.81 2.81 6.63 1.21 0.70 biochemical operations, 3rd edition, John Wiley & Sons.
41 1.85 3.07 7.48 1.21 0.73 Abdullah, Z., Aziz, N. and Ahmad, Z. (2007), Nonlinear
42 1.86 3.71 8.37 1.40 0.73 Modeling Application in Distillation Column, Chemical
43 1.87 4.85 9.59 1.81 0.72 Product and Process modeling, vol.2, no.3, pp.1-15.
44 0.66 2.29 1.86 1.03 0.61 Bandyopadhyay, S. (2002), Effect of feed on optimal
45 0.65 2.25 1.97 1.18 0.61 thermodynamic performance of a distillation column,
46 0.65 2.30 1.82 1.06 0.59 Chem. Eng. J., vol.88, pp.175-186.
Bennett, D. L. and Kovak, K. W. (2010), Optimize
47 0.65 2.34 1.72 1.17 0.62
Distillation Columns, American Institute of Chemical
48 0.64 2.38 1.60 1.32 0.49 Engineers Web.
Rosendo, M. L., (2003) Simulation of Multi-component
Table 3 summarizes the variation in VLE ratios determined Multistage Vapor-Liquid Separations, An Improved
by proposed method in each tray of the column for each Algorithm using the Wang-Henke Tridiagonal Matrix
component. Method, Ind. Eng. Chem. Res. vol.42, no.1, pp.175-182.