Вы находитесь на странице: 1из 12

Optimising and revamping a refinery

hydrogen network
Pinch analysis and mathematical programming support hydrogen network
design and retrofit

Zhao Jianwei Luoyang Petrochemical Engineering Company


Lou Yuhang, Zhang Nan and Keith Guy Process Integration Ltd

A
n increased demand for
hydrogen often makes Make-up Purge
the supply of hydrogen (FM,YM) (FP,YP)
Recycle
a severe bottleneck for many Sink (FR,YR) Source
modern refineries. Much
progress has been made in the Liquid Separator
last few years on the techno- feed
logical development of refinery Reactor
hydrogen management, with Liquid
two categories of approach: product
simple hydrogen pinch analysis
for network targeting, and Figure 1 Simplified diagram of a hydrogen consumer showing hydrogen
detailed mathematical program- sources and sinks
ming for optimisation, design
and retrofit. However, when reactors. Such changes have a crude oils. To obtain the best
applied to the revamping of minimal impact on reaction value from these oils, refiners
existing refinery hydrogen hydrogen consumption, prod- must be able to convert heavy-
networks, these approaches uct yields and quality, while end compounds to lighter
show common limitations providing extra degrees of fractions that can be blended
caused by their rigid restric- freedom for refiners to exploit with gasoline or diesel. To
tions of constant hydrogen additional options for saving achieve this, refineries are using
purity and hydrogen partial hydrogen. The developed more hydrogen addition than
pressure in hydroprocessor approach has been applied to conventional carbon rejection
reactors. In industrial practice, a hydrogen network revamp- for a better production yield.
such restrictions are not ing project for Sinopec, The reason for such a selection
followed exactly and make it achieving a hydrogen utility is not only to produce better-
extremely difficult to modify reduction of 8.8% and annual quality transportation fuels, but
any complex hydrogen systems operational cost savings of also to process lower-quality
using existing techniques. over $9.7 million, with a crude oil, which is heavier than
In this article, a modified simple payback time of less conventional crudes and
hydrogen network model is than half a year. contains more sulphur and
proposed, which allows for nitrogen. All of these facts are
marginal changes in hydrogen Hydrogen addition driving refineries to increase
purity and hydrogen partial There is a worldwide trend their levels of hydroprocessing,
pressure in hydroprocessor towards processing heavier which places increasing

www.digitalrefining.com/article/1000417 REVAMPS 2011 1


crackers, hydrotreaters and
1.0 hydroprocessers. Hydrogen
0.9 purifiers convert lower-purity
+
0.8 hydrogen-containing gases into
– – higher-purity product, which
0.7
+ can then be reused in processes.
0.6
Purity, –

Typical hydrogen purification


0.5 technology includes membrane
0.4 purifiers, pressure swing
0.3 adsorption (PSA) and cryogenic
0.2
separation. Interaction among
hydrogen producers, consumers
0.1
and purifiers determines the
0 design of the hydrogen network
0 50 100 150 200 250 300
in a refinery, as well as the
Flow rate, MMscfd
demand for hydrogen.
In 1999, Alves1 presented
Figure 2 Hydrogen composite curve hydrogen network pinch analy-
sis and hydrogen network
optimisation methods. The
1.0 hydrogen network pinch analy-
0.9 sis method is based on heat
0.8 exchanger network pinch
0.7 analysis technology.2 This
method can determine the
0.6
Purity, –

bottleneck of the whole hydro-


0.5 gen network and can
0.4 systematically analyse hydro-
0.3 gen utilisation at different
0.2 purification levels.
0.1
To carry out hydrogen
network pinch analysis, it is
0
0 10 20 30 40 50 necessary first to identify
Hydrogen surplus, MMscfd sources and sinks of hydrogen,
which can be analogous to hot
and cold streams in heat
Figure 3 Hydrogen surplus diagram exchanger networks. Hydrogen
sources are streams containing
demands on the supply of Existing methodologies hydrogen, which can be used
hydrogen. As energy prices rise, Hydrogen pinch analysis to provide hydrogen to the
the cost of hydrogen production Generally speaking, a refinery’s system. Hydrogen sinks are
also increases. Currently, all hydrogen network consists of processes that consume hydro-
major hydrogen production three parts: hydrogen produc- gen. A hydrogen consumer is a
processes consume a significant tion, hydrogen consumption hydrogen source and also a
amount of energy and generate and hydrogen recovery through hydrogen sink (see Figure 1).
a large amount of greenhouse purification. Hydrogen produc- Recycle hydrogen, high-pres-
gases. Therefore, better hydro- ers generate hydrogen by means sure purge, low-pressure purge
gen management through of continuous catalyst regenera- and other purge gases can be
hydrogen network optimisation tion reformers (CCR), steam regarded as hydrogen sources,
is needed for energy savings reformers and and the reactor inlet should be
and hydrogen generation cost partial oxidation reformers. seen as a hydrogen sink. With
reduction Major consumers include hydro- hydrogen pinch analysis, all

2 REVAMPS 2011 www.digitalrefining.com/article/1000417


conditions for hydrogen
sources and sinks are fixed. 1.0
Once the conditions for all 0.9
hydrogen sources and sinks are 0.8
determined, the next step is to 0.7
draw the hydrogen composite
0.6

Purity, –
curve, which is a two-dimen- Pinch
0.5
sional plot with the flow rate of
total gas on the horizontal axis 0.4
and purity on the vertical axis 0.3
(see Figure 2). 0.2
Plotting the hydrogen demand 0.1
profile and the hydrogen supply
0
profile gives the hydrogen 0 10 20 30 40 50
composite curves. This purity Hydrogen surplus, MMscfd
profile contains the hydrogen
sinks and sources ordered by
decreasing purity. Separately, Figure 4 Targeting minimum hydrogen utility flow rate
the sink and source curves start
at zero flow rate and continue laps with the zero axes (see sources is imposed through
until the lowest purity is repre- Figure 4). The purity at which equality constraints. The objec-
sented. Where the hydrogen this occurs is defined as the tive function is to minimise the
supply curve is above the hydrogen pinch and is the total cost of the hydrogen
hydrogen demand curve, the theoretical bottleneck for how network.
area between the two profiles is much hydrogen can be used Hydrogen pinch analysis is a
marked as surplus (+), which from the sources to the sinks. graphical approach to finding
means the sources provide more The hydrogen utility flow rate the minimum hydrogen utility
hydrogen than is required by that results in a pinch is the in distribution networks. It can
the sinks. If the hydrogen minimum target and is deter- provide insights into hydrogen
supply is below the hydrogen mined before any network distribution and is easy to
demand curve, the area between design. access, so was quickly adopted
the two profiles is marked as With hydrogen pinch tech- for industrial applications.
deficit (-), which means sources nology, the minimum hydrogen However, it also has some
do not provide enough hydro- demand of a hydrogen network significant drawbacks.
gen to the sinks. Calculating can be determined with very The first is that the targets
these surpluses (+) and deficits basic information and simple are set based only on the flow
(-) and plotting them against the data collection. Hydrogen rate and purity requirements,
purity level constructs a hydro- pinch analysis technology also and pressure is ignored. The
gen surplus diagram, or provides certain principles for targeting method assumes that
hydrogen pinch diagram (see hydrogen network design: any streams containing hydro-
Figure 3). For any existing • No cross-pinch match gen can be sent to any
network, the surplus curve is between hydrogen sources and consumers, regardless of the
always on the right side of the sinks stream pressure. In reality, a
vertical axis. • Effective hydrogen purifica- source can only feed a sink if it
For an existing network, all tion should bring hydrogen is at a sufficient pressure level.
parts of the surplus curve are from below the pinch to above Thus, the targets generated
always positive. The hydrogen the pinch. may be too optimistic and
utility can be reduced by Alves also presented a math- unachievable in a real design.3
moving the curve towards the ematical method using linear Another drawback is that the
vertical axis until a vertical programming (LP) to design a selection of a hydrogen purifi-
segment between the purity of hydrogen distribution network. cation process is based simply
the sink and the source over- Mass balance for sinks and on the purities of the product

www.digitalrefining.com/article/1000417 REVAMPS 2011 3


place. The selection of hydro-
gen purifier, type and scale can
H2 plant Sink A be crucial to hydrogen utilisa-
360 psi 1600 psi
tion efficiency.6-10 Liu and
Zhang presented simplified
Source A Sink B models for membrane purifica-
1500 psi 2200 psi
tion and PSA. When integrated
with a mixed integer non-linear
Source B Fuel
1700 psi 80 psi programming (MINLP)
approach, the selection of
hydrogen purifier can be
AM AM considered in the optimisation
1600 psi 360 psi
of the hydrogen network super-
structure. Ahmad et al11
BM BM
2200 psi 360 psi improved this method by
considering typical operating
conditions for hydrogen
AR AR
1600 psi 1500 psi consumers in various operating
periods, and applying it with
BR BR multiple operating cases.
2200 psi 1700 psi
For all mathematical
programming approaches
discussed here, there has been
Figure 5 Example of the superstructure for a hydrogen network one common assumption: that
there are only hydrogen and
and the feedstock. In practice, Detailed mathematical methane components within
the selection of a purification programming the system. All impurities have
process relates to many practi- In 2001, an automated design been lumped as methane.
cal conditions, such as approach for hydrogen network However, the compositions of
recovery, pressure conditions, management was developed, hydrocarbon impurities can
payback rate and network based on the optimisation of a vary in hydrogen sources in
structure.4 reducible superstructure.3 With any refinery. Under a fixed
One more drawback of the superstructure, all possible hydrogen-to-oil ratio and
hydrogen pinch technology is connections within a network hydrogen partial pressure at
that it lumps all impurities as can be included, so that an the reactor inlet of a hydrogen
methane. In fact, for a hydrot- optimisation algorithm could consumer, if the composition of
reating process, even though select among all possible design hydrocarbon impurities
the purity and flow rate for scenarios. Pressure constraints changes in the make-up stream,
make-up hydrogen are fixed, are also included in the design. the vapour-to-liquid equilib-
the flow rate and purity of Compressors are considered rium in the downstream flash
recycle hydrogen and purges both as hydrogen sinks and separation is affected. Hence,
vary if the composition of sources, a compressor inlet as a the flow rate and composition
impurities changes. Therefore, sink, and an outlet as a source. for recycle and purge hydrogen
the minimum hydrogen Efficient utilisation of hydro- also change. Therefore, impuri-
demand target is affected.5 gen in refineries relies on ties can affect the results of the
To overcome the drawbacks hydrogen recovery. Most pinch analysis. Zhang et al5
of hydrogen pinch analysis, hydrogen resources with lower presented a more detailed
many researchers have purities cannot be directly used modelling approach for hydro-
presented different mathemati- in processes. Therefore, hydro- gen network simulation and
cal programming approaches gen purifiers (PSA, membrane optimisation, which uses a
for hydrogen network design and cryogenic separation proc- detailed model for a high-pres-
and optimisation. ess) need to be in the right sure separator of hydrogen.

4 REVAMPS 2011 www.digitalrefining.com/article/1000417


Due to the scale of this detailed could affect product yield, impacts caused by varying the
model, an iterative approach quality and hydrogen consump- hydrogen-to-oil ratio and
between simulation and opti- tion in reactors. Hence, great hydrogen partial pressure are
misation is used in solving the caution must be applied when reflected in the change of
model. adjusting them. composition and flow rate from
In summary, there are two To consider the impact of the consumer to the fuel gas
major categories for refinery varying the hydrogen-to-oil system. Hence, the fuel gas
hydrogen network optimisation ratio and hydrogen partial balance has been written as
techniques. One is based on pressure to the optimisation follows:
graphical analysis and the other result, a few modelling equa-
on mathematical programming. tions need to be modified.12 Σi Fi,j + FH2,j – F*j = ∆Fj, Fuel A
j (7)
Pinch analysis can quickly First, the mass balance equa-
determine the bottleneck of a tions need to be adjusted: Σi Fi,j • yi* + FH2,j • yH2 – F*j • yi*= ∆H2j, Fuel
hydrogen system and the mini- j (8)
A
mum hydrogen demand. Σi Fi,j + FH2,j = F*j A j (1)
Various mathematical program- where ∆Fj,Fuel is the flow rate
ming approaches can help in Σi Fi,j • yi* + FH2,j • yH2 = F*j • yj* A
j (2) change from sink j to fuel gas;
designing a practical hydrogen ∆H2j,Fuel is the pure hydrogen
network. Both approaches have where Fi,j is the flow rate from flow rate change from sink j to
certain advantages and disad- source i to sink j; FH2,j is the fuel gas. The objective function
vantages. Therefore, it is flow rate from the hydrogen is minimum total cost, and the
necessary to integrate the two plant to sink j; Fj* is the flow fuel gas value is calculated
approaches in a practical rate demanded by sink j; yi* is based on the heat value.
hydrogen network design and the purity of source I; yH2 is the Therefore, the change of flow
retrofit. purity of the hydrogen plant; rate and composition of fuel
On the other hand, for all the yj* is the purity demanded by gas caused by the changes in
approaches discussed here, sink j. the hydrogen-to-oil ratio and
there is a common constraint, To vary the hydrogen-to-oil hydrogen partial pressure in a
which is the fixed hydrogen ratio and hydrogen partial hydroprocessor can be reflected
flow rate and purity (or hydro- pressure within a certain range, in the objective function.
gen-to-oil ratio and hydrogen a few new parameters need to Jla provides a more detailed
partial pressure) at the reactor be introduced in modelling. description of the mathematical
inlet of each hydrogen Fj*L, Fj*U are the lower and upper modelling approach.13
consumer. This constraint is to bounds of the flow rate demand
make sure the operating condi- of sink j. yj*L, yj*U are the lower Case study:
tions of hydrogen consumers and upper bounds of the an industrial application
are not affected while the hydrogen purity demand of In this project, Luoyang
network is optimised. However, sink j. Equations 1 and 2 will Petrochemical Engineering
for the new design or retrofit of therefore be changed to Corporation (LPEC) collabo-
a hydrogen network, because inequalities 3 to 6: rated with Process Integration
of the uncertainties of new Ltd (PIL) for a hydrogen
processes, such a constraint Σi Fi,j + FH2,j ≥ F*j L j
A (3) network design optimisation
could be a barrier in determin- project in Sinopec M Refinery.
ing optimisation scenarios. Σi Fi,j + FH2,j ≤ F*j U j
A
(4) Sinopec M Refinery consists of
a refinery and a petrochemical
Modified hydrogen Σi Fi,j • yi* + FH2,j • yH2 ≥ F*jL • yj*L j (5)
A
complex. Currently, the refinery
network model has a processing capacity of 13.5
Allowing the hydrogen-to-oil Σi Fi,j • yi* + FH2,j • yH2 ≥ F*jU • yj*U A
j (6) million t/y of crude oil and
ratio and hydrogen partial ethylene production is 1 million
pressure to vary might be To ensure feasibility, the t/y. In 2009, the refinery
beneficial to network optimisa- mass balance of the network decided to implement a debot-
tion. However, changes in these needs to be maintained. The tlenecking project to increase

www.digitalrefining.com/article/1000417 REVAMPS 2011 5


one membrane unit attached to
the 1.2 MPa(G) hydrogen main.
CCR-2
User 1
3950.0 Overall hydrogen generation
10000
3500.0 91.59% from the two CCRs is about 70
H2 rich purge
6# HTU 4200
000 Nm3/h, with hydrogen
2# HTU User 4 User 5
purity of about 91.59 vol%. In
50726.7 9060.0 3597.1 680 332 2005, M Refinery installed a
CCR-1 membrane process to recover
Membrane User 3
96321.5
2# HC
8360 73% Inlet: 10460 2500 60000
91.59% hydrogen from the feedstock of
Recovery: 70%
All LP purge 2000
a steam reformer (hydrogen
2100 66%
16397.74 plant). The feedstock of the
3# HTU
14910 membrane unit has a flow rate
5518.0 95%
45073
of around 10 200 Nm3/h and a
1# HC
purity of about 62.62 vol%.
All LP purge
10107.0 2970.2
Product hydrogen has a flow
5# HTU 1# HTU
rate of about 4782 Nm3/h and
19513
a purity of about 94.84 vol%.
User 2
207 Overall recovery is over 71%.
Membrane inlet and outlet
4# HTU pressures are 3.4 and 1.2
16000.0
4.5 MPa main 2.4 MPa main MPa(G). The hydrogen concen-
H2 from coal H2 from tration in the 1.2 MPa(G)
gasification ethylene
166215.2 97.5% plant hydrogen main is about 91.80
35720 94% vol%.

Figure 6 Base case: hydrogen distribution 2.4 MPa(G) hydrogen main


There are two sources of hydro-
gen to the 2.4 MPa(G) hydrogen
Base case: reactor inlet conditions for major hydrogen consumers
main: the ethylene plant and
the hydrogen plant. The ethyl-
H2 concentration Flow rate Pure H2 flow ene plant provides hydrogen at

1# HC 84.22 165 346 139 254 a flow rate of 35 720 Nm3/h,
1# HTU 87.03 11 733 10 211 and the hydrogen concentra-
2# HTU 87.02 123 102 107 123
3# HTU 84.46 128 094 108 188 tion is about 94 vol%. The
4# HTU 92.00 16 000 14 720 hydrogen plant generates 99.9
5# HTU 80.39 158 979 127 803 vol% hydrogen at a flow rate
6# HTU 94.19 409 977 386 157 of 43 500 Nm3/h. There is a
User 3 91.00 2484 2260
2# HC 94.89 503 921 478 171 reciprocating compressor to
User 1 _ _ _ compress steam reformer feed-
User 2 _ _ _ stock from 0.4 to 3.7 MPa(G).
The hydrogen concentration in
Table 1 2.4 MPa hydrogen main is
about 97.24 vol%.
crude oil processing capacity to utilisation in the new refining
18 million t/y. Many new proc- configuration. Base case for the new hydrogen
esses will be installed, including Currently, the hydrogen network
a crude distillation unit, fluid mains in M Refinery have two A basic hydrogen network
catalytic cracker, hydrocracker pressure levels: 1.2 MPa(G) and design has been produced for
and sulphur recovery unit. The 2.4 MPa(G). the new M Refinery at the
hydrogen network design and feasibility study stage for the
optimisation project was 1.2 MPa(G) hydrogen main debottlenecking project, which
adopted to improve hydrogen There are two CCR units and is therefore used as the

6 REVAMPS 2011 www.digitalrefining.com/article/1000417


base case for optimisation
design. 1.05

Hydrogen concentration,
0.95
Hydrogen providers
0.85
Hydrogen demand for the
refinery after debottlenecking 0.75

vol.%
will be significantly increased. 0.65
As the feasibility study esti-
mated, overall hydrogen usage 0.55
will reach 270 000 Nm3/h. 0.45
Meanwhile, hydrogen produc-
0.35
tion in the refinery is facing
new limitations. According to 0.25
local environmental legislation, 0 500 1000 1500 2000 2500 3000
heavy fuel oil is no longer H2 flow rate, Nm3/h
allowed to be used as fuel in
furnaces and fired heaters, and Figure 7 Base case: hydrogen surplus diagram
desulphurised fuel gas is
required as the substitute. conditions for major hydrogen Figure 7. The hydrogen supply
Therefore, there will be a consumers are shown in is very close to the minimum
shortage of fuel gas to be used Table 1. target based on the current
as raw material for hydrogen hydrogen purifying strategy.
production. Moreover, local Pinch analysis for the base case However, Figure 7 also indi-
natural gas and naphtha prices The hydrogen surplus diagram cates that a large amount of
are high, so that hydrogen for the base case is shown in hydrogen (over 25 000 Nm3/h
production based on these
sources will be uneconomical.
After investigation, Sinopec CCR-2
decided to install a coal gasifi- User 1
3950.0
10000
cation-based hydrogen plant, 3500.0 91.59%
H2 rich purge
with a design capacity of 200 6# HTU 3417.26
2# HTU User 4 User 5
000 Nm3/h. The existing hydro-
gen plant based on fuel gas will 50726.7 10478.0 391.5 680 332

CCR-1
be used as a standby. The cost Membrane User 3 60000
of hydrogen generated from a 96321.5
2# HC
8360 73% Inlet: 10460
Recovery: 70%
2500
91.59%
All LP purge 2000
coal gasification-based hydro-
gen plant is estimated to be 3# HTU
31628.7

around $0.18/Nm3. The hydro- 2100 66%


All LP purge
5518.0 95%

gen pressure from the new 33047.7


1# HC
hydrogen plant is 4.5 MPa(G). 9060.0

Hence, there will be three 4540.0 2970.2


5# HTU 1# HTU
hydrogen mains (4.5, 2.4 and
25293.0
1.2 MPa(G)) in M Refinery. This
User 2
introduces new complexity to 207

the design of the hydrogen


4# HTU
network. 1160
4.5 MPa main 2.4 MPa main
H2 from coal H2 from
Hydrogen network gasification ethylene
The hydrogen network flow- 162066.2 97.5% plant
sheet for the base case is shown 35720 94%

in Figure 6, and the corre-


sponding reactor inlet Figure 8 Optimisation scenario 1: hydrogen distribution

www.digitalrefining.com/article/1000417 REVAMPS 2011 7


pure hydrogen) is lost to the
1.05 fuel gas system. Therefore,
Hydrogen concentration, hydrogen recovery through
0.95
additional hydrogen purifica-
0.85 tion could be an economical
0.75 option, and this was investi-
gated further. These streams
vol.%

0.65
have a lower hydrogen concen-
0.55 tration and cannot be reused
0.45
directly.

0.35
Optimisation scenarios
0.25 The objective of the hydrogen
network design is to minimise
0

0
0

00

00

00

00

50

00

50

00

50

00
25

50

75

the total operating cost:


10

12

15

17

20

22

25
H2 flow rate, Nm3/h
• Total operating cost = H2
generation cost + total
Figure 9 Optimisation scenario 1: hydrogen surplus diagram compression cost – fuel gas
value
• Fuel gas flow rate = total
Optimisation scenario 1: reactor inlet conditions for major hydrogen supply from all H2 providers –
consumers
net H2 consumption in all H2
consumers
H2 concentration Flow rate Pure H2 flow
• H2 concentration in fuel gas
Base case Optimised Base case Optimised Base case Optimised
1# HC 84.22 84.22% 165 346 167 462 139 254 141 036 = pure H2 supply from all
1# HTU 87.03 87.03% 11 733 11 732 10 211 10 210 providers – net H2 consump-
2# HTU 87.02 87.02% 123 102 124 638 107 123 108 460
3# HTU 84.46 84.54% 128 094 126 875 108 188 107 261 tion in all H2 consumers/fuel
4# HTU 92.00 92.10% 16 000 16 000 14 720 14 736 gas flow rate.
5# HTU 80.39 80.49% 158 979 158 785 127 803 127 809
6# HTU 94.19 93.91% 409 977 409 977 386 157 384 998
The net hydrogen consump-
User 3 91.00 91.59% 2484 2500 2260 2290 tion in each consumer is
2# HC 94.89 94.45% 503 921 503 921 478 171 475 956 assumed to be fixed. The impu-
User 1 _ 91.59% _ 3950 – –
User 2 – 94.00% – 207 – rities are lumped as methane,
so the net heating value of fuel
gas can be calculated based on
Table 2 hydrogen concentration and
fuel gas flow rate.

Optimisation scenario 2: reactor inlet conditions for major hydrogen Optimisation scenario 1
consumers The first optimisation scenario
aims to optimise the hydrogen
H2 concentration Flow rate Pure H2 flow network design without intro-
Base case Optimised Base case Optimised Base case Optimised ducing any new hydrogen
1# HC 84.22 84.22% 165 346 166 065 139 254 139 860
1# HTU 87.03 87.03% 11 733 11 732 10 211 10 210 purification process.
2# HTU 87.02 87.26% 123 102 122 767 107 123 107 124 Compared with the base case,
3# HTU 84.46 84.54% 128 094 126 875 108 188 107 261
4# HTU 92.00 92.10% 16 000 16 000 14 720 14 736 the operating pattern for one
5# HTU 80.39 80.41% 158 979 158 954 127 803 127 809 hydrotreater (4# HTU) has
6# HTU 94.19 93.91% 409 977 409 977 386 157 384 998
User 3 91.00 91.59% 2484 2500 2260 2290
been changed from once-
2# HC 94.89 94.45% 503 921 503 921 478 171 475 956 through to complete recycle. In
User 1 _ 91.59% _ 3950 _ _ the base case, the make-up
User 2 – 94.00% – 207 – –
stream for 4# HTU comes from
the 2.4 MPa(G) main, with a
Table 3 flow rate of 16 000 Nm3/h. This

8 REVAMPS 2011 www.digitalrefining.com/article/1000417


make-up pressure is high
enough for the reactor; there-
CCR-2
fore, the make-up compressor User 1
3950.0
10000
can be turned off. Moreover, 3500.0 91.59%

the recycle compressor in 4# 6# HTU


2# HTU User 4 User 5
HTU is off. High-pressure (HP)
purge leaving the separator of 50726.7 4625.0 680 332

CCR-1
4# HTU has a flow rate of Membrane User 3 60000
about 15 000 Nm3/h and 96321.5
2# HC
8360 73% Inlet: 10460
Recovery: 70%
2500
91.59%
All LP purge 2000
contains 93 vol% hydrogen at a
pressure of 1.2 MPa(G). This 3# HTU
31628.7
2100 66% 5518.0 95%
HP purge can be used in other All LP purge
1.2 MPa main
2921.0
hydrogen consumers. With 1# HC
4353.0 33439.2
such an arrangement, the
compression duty of the make- 4540.0
5# HTU 1# HTU
2970.2

up and recycle compressors in 30000.0


7790.5

4# HTU is saved in the base User 2


8200 38.5%
1# HC LP purge
207 PSA
case. After optimisation, the 1.2 MPa 5200 52.2%
recycle compressor of 4# HTU 4# HTU
Inlet: 22540 3# HTU LP purge
1160
is turned on. As a result, only 4.5 MPa main 2.4 MPa main
4200 87.3%
6# HTU LP purge
1160 Nm3/h make-up hydro- H2 from coal H2 from 4940 45.5%
gen is required, but the gasification ethylene Membrane
residue
plant
compression duty is increased. 151673.2 97.5%
35720 94%
The hydrogen distribution in
2.4 MPa(G) and 1.2 MPa(G) is
rearranged because of this Figure 10 Optimisation scenario 2: hydrogen distribution
operational change in 4# HTU.
Figure 8 shows the hydrogen

distribution system for optimi-
sation scenario 1. 
)ZES PHF O  D PO D F O U S B U J PO 

The total compression duty 


in optimisation scenario 1 is 
357.6 kW/h higher than the

base case, but the total hydro-
WPM  

gen supply has been reduced 


by 4149 Nm3/h. As a result, the 
total operating cost is reduced 
by $3.5 million/y.

The hydrogen surplus
diagram for optimisation 
scenario 1 is shown in Figure 9. 
The reactor inlet conditions for       
major hydrogen consumers are H2 flow rate, Nm 3 /h
listed in Table 2.
Total hydrogen supply is Figure 11 Optimisation scenario 2: hydrogen surplus diagram
very close to the target level.
The net hydrogen loss to the the hydrogen loss, a new process is introduced into the
fuel gas system is reduced to hydrogen purification unit is hydrogen network and the
21 202 Nm3/h, from 25 247 needed. existing membrane unit is still
Nm3/h in the base case. The used. The new PSA process
hydrogen purification capacity Optimisation scenario 2 consists of a PSA unit, product
is used up. To further reduce With this scenario, a new PSA hydrogen compressor and tail

www.digitalrefining.com/article/1000417 REVAMPS 2011 9


Figure 10 shows the hydro-
gen distribution system for
$$3
6TFS


optimisation scenario 2. The
  hydrogen surplus diagram for
)56 optimisation scenario 2 is
)56 6TFS 6TFS
shown in Figure 11. The reactor
.1BNBJO
    
inlet conditions for major
$$3
hydrogen consumers are listed
6TFS 

)$

 in Table 3.


)56
 Optimisation scenario 3
In this scenario, a larger PSA

process is introduced into the
)$ 
hydrogen network and the
  existing membrane unit is
)56 )56

turned off. Such an arrange-

 ment gives a simplified
6TFS
 14"
)$-1QVSHF
hydrogen network design. The
.1B 
*OMFU )56-1QVSHF optimised PSA unit capacity is
)56


16 000 Nm3/h. The capacity for
.1BNBJO .1BNBJO )56-1QVSHF compressors involved in this
)GSPNDPBM )GSPN
HBTJýDBUJPO
 PSA process is larger than for
FUIZMFOF )56-1QVSHF
 QMBOU 
optimisation scenario 2.
 )$-1QVSHF Figure 12 shows the hydro-
gen distribution system for
optimisation scenario 3. The
Figure 12 Optimisation scenario 3: hydrogen distribution hydrogen surplus diagram for
optimisation scenario 3 is
gas compressor. The optimised from various consumers and shown in Figure 13. The reactor
capacity for the PSA unit is 10 the membrane residue stream. inlet conditions for major
000 Nm3/h. The new PSA proc- The total hydrogen supply can hydrogen consumers are listed
ess makes it possible to recover be reduced by 14 500 Nm3/h in Table 4.
hydrogen from LP purge gases compared with the base case. A comparison of results for
the base case and the three
optimisation scenarios are
 shown in Table 5. The improve-
 ment for optimisation scenario
Hydroge n  con ce n t r at i on , 

1 is limited by the hydrogen



purification capacity. Scenarios
 2 and 3 offer great improve-
 ment at relatively low capital
v ol %

 investment. Eventually, optimi-


sation scenario 2 was chosen

for implementation.

 Conclusions
 • After optimisation of the
 hydrogen network design, the
       total operating cost for M
H2 flow rate, Nm 3/h Refinery has been reduced by
$10.2 million/y, compared with
Figure 13 Optimisation scenario 3: hydrogen surplus diagram the base case design

10 REVAMPS 2011 www.digitalrefining.com/article/1000417


• Selection of the type and
Optimisation scenario 3: reactor inlet conditions for major hydrogen
scale of hydrogen purifiers can
consumers
be crucial to hydrogen utilisa-
tion efficiency
• Fully understanding practical H2 concentration Flow rate Pure H2 flow
Base case Optimised Base case Optimised Base case Optimised
constraints is very important in 1# HC 84.22 84.28% 165 346 165 233 139 254 139 254
refinery hydrogen network 1# HTU 87.03 87.03% 11 733 11 732 10 211 10 210
2# HTU 87.02 87.27% 123 102 122 749 107 123 107 125
design or retrofit. Pinch analy- 3# HTU 84.46 84.54% 128 094 126 875 108 188 107 261
sis can quickly determine the 4# HTU 92.00 92.10% 16 000 16 000 14 720 14 736
5# HTU 80.39 80.40% 158 979 158 967 127 803 127 808
bottleneck of a hydrogen 6# HTU 94.19 93.91% 409 977 409 977 386 157 384 998
system and minimum hydro- User 3 91.00 91.59% 2484 2500 2260 2290
gen demand, and can identify 2# HC 94.89 94.45% 503 921 503 921 478 171 475 956
User 1 _ 91.59% _ 3950 – –
a direction for improvement. User 2 – 94.00% – 207 – –
Mathematical programming
approaches can then help in
designing a practical hydrogen Table 4
network. It is necessary to inte-
grate the two approaches in
Comparison of results
practical hydrogen network
design and retrofit.
Base case Optimisation Optimisation Optimisation
scenario 1 scenario 2 scenario 3
References Hydrogen supply
1 Alves J J, Analysis and design of 1#CCR, Nm /h 3
60 000 60 000 60 000 60 000
2#CCR, Nm3/h 10 000 10 000 10 000 10 000
refinery hydrogen distribution systems, Ethylene H2, Nm3/h 35 720 35 720 35 720 35 720
PhD thesis, Department of Process Coal gasification H2, Nm3/h 166 215 162 066 151 673 151 588
Integration, University of Manchester, Total H2 supply, Nm3/h 271 935 267 786 257 393 257 308
1999. Total pure H2 supply, Nm3/h 259 749 255 704 245 571 245 488
Total H2 consumption, Nm3/h 234 502 234 502 234 502 234 502
2 Linnhoff B, Mason D R, Wardle I, Net H2 loss, Nm3/h 25 247 21 202 11 069 10 986
Understanding heat exchanger networks, Effective H2 usage, % 90.23 91.71 95.49 95.53
Computers and Chemical Engineering, 3, H2 generation cost, MMRMB/yr 2489.7 2451.2 2354.8 2354
1979, 295. Hydrogen purifier(s)
Membrane inlet, Nm3/h 10 460 10 460 10 460 0
3 Hallale N, Liu F, Refinery hydrogen Membrane outlet, Nm3/h 5518 5518 5518 0
management for clean fuels production, PSA inlet, Nm3/h 0 0 22 541.9 28 060
Advances in Environmental Research, 6, PSA outlet, Nm3/h 0 0 10 711 15 477
Fuel gas
2001, 81–98. Fuel gas flow, Nm3/h 37 433 33 284 22 891 22 806
4 Liu F, Zhang N, Strategy of purifier H2 concentration in fuel gas, % 67.40 63.70 48.40 48.20
selection and integration in hydrogen Fuel gas value, MMRMB/yr 265.5 247.2 201.3 200.9
Compression
networks, Chemical Engineering Research
Total compression duty for
and Design, 82, 2004, 1315–1330. make-up H2 compression, kW 38 627 37 740 39 205 39 062
5 Zhang N, Singh B B, Liu F, A systematic Total compression duty in
approach for refinery hydrogen recycle compressors, kW 17 647 18 005 18 005 18 005
Total compression cost, MMRMB/yr 270.1 267.6 274.6 273.9
management, proceedings of PRES2008/ Total operating cost, MMRMB/yr 2494.3 2471.6 2428.1 2427
CHISA2008, Prague, 2008, 4, 1201. (H2 generation cost + total
6 Miller G, Stoecker J, Selection of a compression cost - fuel gas value) Base Base-22.7 Base-66.2 Base-67.3
Capital investment, MMRMB/yr 32 46
hydrogen separation process, NPRA Pay back time, year 0.48 0.69
Annual Meeting, San Francisco, CA, 1989.
7 Spillmann R W, Economics of gas
separation membranes, Chemical Table 5
Engineering Progress, 1989, 41–62.
8 Winston H W S, Sirkar K, Membrane
Handbook, Chapman and Hall, London, 10 Peramanu S, Cox B G, Pruden B B, 11 Ahmad M I, Zhang N, Jobson M,
1992. Economics of hydrogen recovery processes Modelling and optimisation for design
9 Ruthven D, Farooq S, Knaebel K, for the purification of hydroprocessor of hydrogen networks for multi-period
Pressure Swing Adsorption, VCH, New purge and off-gases, International Journal operation, Journal of Cleaner Production,
York, 1994. of Hydrogen Energy, 24, 1999, 405–424. 18, 2010, 889–899.

www.digitalrefining.com/article/1000417 REVAMPS 2011 11


12 Liu F, Hydrogen integration in oil for Process Integration at University of Institute of Applied Catalysis, as a Partner
refineries, PhD thesis, Department Manchester, UK. and Director of Spiritus Consulting, as
of process integration, University of Zhang Nan is Process Integration Ltd’sVice Chairman of the Management Board of
Manchester, 2002. President for Business Development in Impact Faraday Partnership and as non-
13 JIa N, Refinery hydrogen China, with extensive industrial experience executive director of Webaspx. He is a
network optimisation with improved as a process engineer with Sinopec in Visiting Professor at the University of Bath
hydroprocessor modelling, PhD thesis, refinery process design, specialising in the and Imperial College London, a Fellow and
Centre for Process Integration, University design of FCC units. His specialisations Vice President of IChemE and Chairman of
of Manchester, 2010. include refinery optimisation, refinery its Technical Board, a Fellow of the Royal
hydrogen management, refinery process Academy of Engineering and Chairman
Zhao Jianwei is a Deputy Chief Engineer modelling, energy system optimisation of the academy’s President’s Membership
with Luoyang Petrochemical Engineering and integration, emissions reduction and Group, and a member of the Engineering
Company, Henan, China. A professor- reliability, availability and maintainability. and Physical Research Council.
grade senior engineer, he has over 25 He holds a BSc in chemical engineering
years’ experience in hydrogen plant from Tsinghua University, China, and a PhD
design, refinery engineering and project from the Centre for Process Integration at
management. University of Manchester.
Lou Yuhang is a consultant with Process Keith Guy is Chairman of Process Links
Integration Ltd, Manchester, UK. Her Integration Ltd. Following a 28-year
main specialisations include energy career in Air Products, which included the More articles from the following
conservation, refinery hydrogen network positions of Engineering Director, Director categories:
management and CO2 capture and of European Marketing and Business Hydroprocessing
storage (CCS). She holds a BSc from Development Director, his appointments Revamps, Shutdowns and
Zhejiang University, China, and a PhD in include Chairman of Process Systems Turnarounds
chemical engineering from the Centre Enterprise Ltd, serving as CEO of the

12 REVAMPS 2011 www.digitalrefining.com/article/1000417

Вам также может понравиться