Fuels For The 21st Century

Fuel Sulfur Solutions Hydrogen Solutions MTBE Solutions Benzene Solutions

If You're Burning H2 Hydrogen Solutions

Youre Burning Money
The continuing need to produce lighter, higher performance,
Selection and Revamp of and environmentally acceptable fuels requires more
Hydrogen Purification hydrotreating and hydrocracking, which in turn increases the
Process demand for hydrogen. How can you optimize the use of
hydrogen throughout the whole refinery network?
Polybed PSA Technology

Polysep Membrane
Technology
IF YOURE BURNING H2, YOURE BURNING MONEY
Use New H2 Management Techniques to Improve Refinery
Margins and Meet New Fuel Requirements

Alan Zagoria, Gavin P. Towler, and Brian M. Wood

UOP LLC
Des Plaines, Illinois

and

Fred M. Hibbs
UOP Limited
Guildford, U.K.

Print This Section

Prepared for presentation at the

1999 NPRA Annual Meeting
San Antonio, Texas
March 22 25,1999
 IF YOURE BURNING H2, YOURE BURNING MONEY
Use New H2 Management Techniques to Improve Refinery
Margins and Meet New Fuel Requirements

Alan Zagoria, Gavin P. Towler, and Brian M. Wood

UOP LLC
Des Plaines, Illinois

and

Fred M. Hibbs
UOP Limited
Guildford, U.K.

ABSTRACT

One of the most-critical issues in producing low-sulfur fuels is managing the impact of the
new requirements on the refinery hydrogen balance. The combination of low-sulfur fuel
specifications, reduced production of hydrogen in catalytic reformers, and the desire to
process heavier, more-sour crudes make hydrogen management a critical issue. Generating,
recovering, and purchasing hydrogen have a significant impact on refinery operating costs.
More importantly, overall refinery operations may be constrained by the availability of
hydrogen.

New transportation fuel specifications are forcing refiners to look carefully at their
existing hydrogen systems. This situation presents an excellent opportunity to identify ways
to actually increase the refinerys profitability, rather than simply minimizing the negative
impact of the new rules.

AM-99-08
1
 A methodology has recently been developed for systematically analyzing the refinery
hydrogen balance as a network problem. This methodology provides a means of setting
minimum consumption targets and getting some direction on where network improvements
can be made. It can lower refinery operating costs or new hydrogen plant capacity by
reducing overall hydrogen needs.

UOP has found that combining this methodology with a detailed understanding of the
role that hydrogen plays in hydrogen-consuming processes can increase refinery profitability
far beyond the benefit realized by simply reducing hydrogen costs. Hydrogen, specifically
hydrogen partial pressure, has such a strong impact on profitability because of its effect on
throughput, product quality, conversion, yield, and catalyst life. This paper uses commercial
examples to demonstrate the great potential to increase refinery margins by optimizing,
rather than minimizing, the use of hydrogen in the refinery.

INTRODUCTION

In a typical oil refinery, hydrogen is consumed primarily in hydrotreating and hydrocracking

processes for removal of sulfur and nitrogen, in heavy hydrocarbon cracking, and in
saturation of aromatic compounds. The primary source of hydrogen is by-product hydrogen
from the catalytic reformer. Typically, steam reformers produce any additional hydrogen
required.

One or more hydrogen headers form the distribution system that supplies hydrogen to
the consuming processes. Hydrogen-rich purge gases from the consuming processes are
either returned to this distribution system or burned as fuel. A complex series of makeup
and recycle compressors is required to move the hydrogen through the network. As the
hydrogen vented from one process is used as makeup to another, the pressure and purity
typically fall; hence, the distribution system often includes cascades of hydrogen from high
purity to low purity. Hydrogen purification processes such as pressure-swing adsorption
(PSA) and membrane separation, which upgrade the hydrogen purity of these offgasses, are
often added to the hydrogen distribution system.

In the past, refiners have rarely been limited in their production capacity or product
slate by the availability of hydrogen. However, current trends in the refining industry are
reducing the availability of hydrogen to the point where most refiners are concerned about
their future hydrogen balance. The specifications for low-sulfur fuels require increased
hydrogen consumption in hydrotreaters. At the same time, limits on the aromatics content
of gasoline and the requirements for oxygenates have led to lower severity in the catalytic

AM-99-08
2
reformer and so less hydrogen is produced in this unit. The use of heavier crude oils and
more bottom-of-the-barrel processing increase the hydrogen demand in hydrocracking and
heavy oil hydrotreating units.

Traditionally, refiners would have met this increased demand by increasing hydrogen
plant production (turning up production or revamping existing equipment), building a new
hydrogen plant, purchasing hydrogen from outside suppliers, or increasing recovery of
hydrogen that was going to fuel by installing new hydrogen purification capacity. These
options are all expensive because they require large capital outlays or high operating costs
or both. As a result, refiners are interested in the lower-cost alternative of optimizing and
revamping the hydrogen distribution network. Table 1 shows typical operating and capital
cost implications of these alternatives for meeting increased hydrogen demand.

Table 1. Typical Costs to Meet Additional H2 Requirements

Capital Cost, Operating Cost, Total Cost,

Options $/SCFD $/M SCF $MM/yr
Turn-up H2 plant 0 2 7
Expand H2 plant 0.2 1.0 2.1 8 - 10
Build new H2 plant 1 1.9 9
Add new H2 purification 0.1 0.4 0.5-1.0 24
Purchase over-the-fence H2 0 15 3 - 17
Modify the H2 network 0.1 0 0.3
Based on 10 MM SCFD additional capacity and 25%/year capital charges. Operating cost for
new purification includes debit for the heating value of hydrogen not going to fuel.

Many refiners are inclined to avoid capital investment in hydrogen production facilities
because the hydrogen plant is regarded as a utility and not a profit center. Over-the-fence
supply may be cost-effective in some cases, but it inevitably makes the refinery critically
dependent on an external supplier. Although the addition of purification capacity is usually
an attractive option, it involves complex decisions such as:

Which stream or streams to recover?

Which recovery technology to use?
What inlet and outlet pressures to use?
What trade-off to make between product purity, recovery, and capital cost?

AM-99-08
3
The complexity of the problem has created a need for systematic methods and tools for
hydrogen management.

PREVIOUS WORK

Alves and Towler1 proposed a method for analyzing refinery hydrogen networks that is
analogous to the well-established pinch analysis method for heat integration. The hydrogen
pinch method they describe requires collecting hydrogen balance data around individual
process units (sinks and sources) and generating a hydrogen purity diagram. In Figure 1, the
stream hydrogen purity is plotted against the gas flow rate as two composite streams, one
representing the hydrogen sources and the other representing the demands of the hydrogen
consumers (sinks). From this plot a purity surplus diagram, as described by Alves and
Towler2 and illustrated in Figure 2, is generated. The surplus must be zero or greater at all
points on this curve. If the surplus were negative, then some unit would not receive all the
hydrogen it needs or would receive hydrogen at a lower purity than required. If the surplus
at any point is negative, then hydrogen production from the swing unit (usually a hydrogen
plant) must be increased to make more pure hydrogen, thus shifting the curve to the right.
This diagram can be used to set targets for the minimum hydrogen consumption in the
network. The target is the hydrogen flow rate from the swing producers that creates a
purity surplus pinch, that is, a purity at which the surplus is zero. The hydrogen pinch
method can also provide guidance in selecting which offgas streams are most suitable for
recovery.

Although the hydrogen pinch method can identify network modifications that reduce
hydrogen consumption, the pinch method cannot be directly used to design the actual
network, because the pinch method specifies only the network as a whole, not flows
between the various process units. Another limitation of the method is that the feed flow,
feed composition, and performance of all purification units must be known before the
analysis begins, because the purifiers must be treated as a process unit and included in the
composite curves. Designing the network requires more-detailed models of the real network
and its components.

REDEFINING THE PROBLEM

Applying hydrogen pinch analysis and a network model is sufficient to substantially reduce
overall hydrogen requirements and the related compression costs. However, a more-
powerful approach is to look at the role hydrogen plays in the hydroprocessing units,

AM-99-08
4
because hydrogen significantly affects the performance of hydrotreaters and hydrocrackers.
What every refiner is really concerned about is increasing refinery profitability. Modifying
the hydrogen network to reduce hydrogen requirements without considering the impact of
hydrogen on the overall refinery does not maximize refinery profitability. Therefore, the
hydrogen network analysis must be expanded to consider the overall refinery, including the
hydroprocessing processes, catalytic reforming processes, and refinery economics.

The hydrogen pinch method assumes that the purity of hydrogen (makeup plus recycle)
fed to the reactor of every consuming process remains constant. Although this assumption
simplifies network analysis, significant process and economic benefits can be obtained by
varying the hydrogen partial pressure in hydroprocessing reactors. The key to this approach
is to understand its effects on process performance. Although the hydrogen saved by
network optimization can be valuable, using the hydrogen more effectively within the
refinery provides a greater impact on profitability than simply minimizing its consumption.

Increasing the hydrogen partial pressure in a process improves conversion, yield,

product quality, catalyst life, or throughput. Depending on the specific network, these
factors may or may not affect the overall hydrogen consumption in the refinery. If one of
these factors is constraining the overall refinery, then the increased costs of the hydrogen
system are usually inconsequential compared to the economic benefits realized. UOP has
developed detailed process models that can be used to determine the relationship between
reactor hydrogen partial pressure and process performance.

On the supply side of the hydrogen balance, the catalytic reforming process and its
economics should also be considered. Typically, the economics of the gasoline pool, not
hydrogen requirements, govern the desired severity of the reformer. The reformer feed rate
and severity are not normally adjusted to meet hydrogen demand. However, when major
changes in the hydrogen balance are anticipated and capital expenditures of some kind are
required, the catalytic reformer may be a cost-effective source of additional hydrogen.
Process revamps to increase naphtha charge rate or hydrogen yield may be a practical
alternative to increasing hydrogen plant capacity.

INCREASING PROFITABILITY WITH HYDROGEN

Optimizing the use of hydrogen in the refinery consists of three major steps. First, the
network itself is optimized. Second, existing hydrogen purification is modified, or new
purification capacity is added, if appropriate. Third, process changes are made that improve

AM-99-08
5
refinery gross margin. Because the process changes require redesign of the hydrogen
network, the analysis becomes iterative.

STEP 1. NETWORK OPTIMIZATION

Hydrogen-related field data for each consuming and producing unit are used to generate the
composite curves, hydrogen purity surplus curve (pinch diagram), and a detailed hydrogen
network model of the existing refinery. The surplus curve is used to generate a target
hydrogen consumption. The detailed hydrogen network model is used to find modifications
to the network (connections and cascades) that will bring consumption down to the target
level. Usually, these modifications can be made without any major capital projects.

STEP 2. IMPROVED PURIFICATION

Modifying the feed flow, feed composition, performance, or recovery of existing

purification equipment changes the target hydrogen consumption and therefore may be a
relatively low-cost way to reduce hydrogen consumption below the target calculated in the
network optimization step. New purification reduces the target in the same way that
modifying existing purification does. The capital cost of the purification must be justified by
the hydrogen savings, and the hydrogen lost in the purifier waste stream must be justified by
the reduction in hydrogen makeup requirements.

Many choices must be made in selecting an improved purification scheme. Guidance on

which streams should be recovered can be found in the hydrogen purity surplus curve (pinch
diagram). Selecting between membrane and PSA systems requires a series of trade-offs in
capital cost, purity, recovery, and flexibility. Because operating pressures have a very strong
impact on purifier performance, selecting the optimum feed and tail gas pressures requires
trade-offs in capital cost and performance of the purifier relative to compression cost. Once
a PSA or membrane system is chosen, variations of the purifier process, in essence trade-
offs of recovery vs. purity and recovery vs. capital cost, must still be considered.

STEP 3. PROCESS MODIFICATIONS

Modifying the hydrogen partial pressure in the reactor may significantly affect the
performance of the hydroprocessing unit. Detailed process models predict the effect of
changing partial pressure on conversion, yield, throughput, product quality, and catalyst life.
The economic impact of these changes is determined by a refinerywide economic model.

AM-99-08
6
 Increasing the hydrogen partial pressure may or may not increase the overall hydrogen
requirements of the network, depending on the location of the process on the pinch diagram
and other factors. Even if overall hydrogen requirements do not increase, network
modifications must be made, and so the whole design process becomes iterative. Analysis
will show that the economic benefit of some process changes more than justifies increasing
the overall hydrogen demand.

TOOLS REQUIRED

To effectively maximize refinery profits through hydrogen management, the following

analytical tools are required:

Calculation of hydrogen pinch curves and target hydrogen consumption.

Detailed model of the hydrogen network: calculation of flows and purity to each
unit, including the purifier.
Detailed process models of hydroprocessing units: estimate performance as a
function of hydrogen partial pressure.
Detailed models of membrane and PSA purification processes: estimate
performance and capital cost as a function of feed flow, feed composition, pressure
levels, and various process options.
Refinerywide economic model: calculate impact of changing yields, product quality,
and throughput on refinery gross margin. This model is usually implemented using
linear programming (LP) techniques.

COMMERCIAL EXAMPLES

LARGE ASIA-PACIFIC REFINERY

These hydrogen management methods can be illustrated by an example based on a study

performed for a refinery in the Asia-Pacific area. The hydrogen study was part of a
refinerywide consulting study. The base case (current operation) is shown in Figure 3. The
customers stated objective was to reduce the direct operating costs associated with
hydrogen production. UOP recommended adopting the broader objective of increasing
overall refinery profitability.

AM-99-08
7
 The hydrogen plant production in the base case (current operation) is only slightly
greater (3% of total hydrogen plant production) than the target value based on the pinch
analysis. Hydrogen is lost to the vent only from HT 10 and the membrane nonpermeate
(Figure 3). Therefore, hydrogen demand can be reduced only slightly unless the feed to the
purification unit (membrane) or the performance of the purification unit are modified. A
detailed network model was used to identify the network modifications that are required to
reduce hydrogen requirements to the target value. Only minor piping modifications were
required.

An analysis of the refinerywide LP economic model and detailed process models

showed a great benefit from increasing the hydrogen partial pressure in two
hydroprocessing units (HT 4 and HT 10 ). This increase in partial pressure results in more
production of high-value distillates through greater conversion with the existing reactors
and catalyst. The existing run length was maintained. Network improvements were
therefore identified to provide higher purity in these two reactors with the lowest possible
hydrogen production requirements. Because changing the purity in these reactors makes the
previously optimized network suboptimal, network improvements were made to reduce the
overall hydrogen production requirements.

Various options to improve the network were developed (Table 2). Each has a
different impact on hydrogen requirements and refinery profit.

Case 1 shows the improvements that could be achieved by applying the basic hydrogen
pinch methodology (that is, modifying connections and purification only). The total savings
reported in this table includes the costs of additional compression power, membrane
replacement, and additional hydrogen as well as the value of increased distillate production.
Although the savings realized in Case 1 are significant, improving overall refinery
profitability has much more potential.

Case 2 includes additional hydrogen purification capacity along with network

optimization.

AM-99-08
8
 Table 2. Summary of Alternative Network Modifications

H2 Savings, Total Savings, Capital Cost,

Case Description Nm3/hr* $MM/yr** $MM
Base Current operation 0
1 Network optimization only 1,300 0.5 < 0.05
2 Network optimization & 1,950 0.7 0.7
purification
3 Process improvement, no (21,660) 2.8 < 0.05
network optimization
4 Process improvement, network (5,640) 8.1 3.1
optimization & purification
5 Process improvement, easy (3,470) 4.3 < 0.05
optimization only
* A negative number in the H2 savings column indicates that overall hydrogen requirements
increased.
** Savings include incremental cost of hydrogen + compression costs + increase in overall
refinery margin as determined by the refinery LP model. The refinery margin is increased by
increasing the amount of distillate produced in hydrotreaters HT 10 (Cases 3 to 5) and HT 4
(Cases 3and 4).

In Case 3, the hydrogen purity (partial pressure in the reactors) in the two units is
increased without changing the hydrogen distribution system by simply increasing purge
flows. Although this increase in hydrogen purity improves process performance, it also
tremendously increases hydrogen demand and sends so much purge gas to fuel that the
refinery fuel balance may be upset. Therefore, this solution is not practical.

Case 4 represents optimizing the hydrogen distribution system, including new

purification capacity, after incorporating the process changes to the two reactors.
Additional purification was clearly warranted in this case to reduce the amount of hydrogen
lost to fuel from the purge of these two reactors. All of the trade-offs in operating
conditions and processing schemes described earlier for both PSA and membrane systems,
were considered. For this particular application, the most cost-effective solution is
membrane process with compression of the hydrogen product. Figure 4 shows the trade-off
between purifier capacity (capital cost) and hydrogen savings. The capital costs shown for
Case 4 reflect the estimated installed capital cost of the new membrane and compressor.

AM-99-08
9
Clearly, the capital project in Case 4 pays out quickly. The modified network for this case is
shown in Figure 5.

Because the improved network requires a capital project, an additional optimization

was carried out to identify a low-capital solution that would capture the benefit of the
process improvements but could be implemented immediately. In Case 5, only the purity in
HT 10 is increased, because too much venting was required to increase the purity in HT 4.
The only modifications are simple network modifications such as varying flows and
connections. This option still improves profitability over the base case, and network-only
improvement (Case 1) requires negligible investment and can be implemented until the new
purification unit can be installed.

This study demonstrates that solely minimizing hydrogen consumption does not
maximize refinery profitability. Figure 6 represents this case study from a hydrogen-
consumption point of view. The left bar shows the hydrogen savings benefit of improving
the network only (Case 1). The middle bar shows the benefit of improved network plus new
purification (Case 2), and the right bar shows the result of process optimization (Case 4). If
a refiner is concerned only with minimizing hydrogen consumption, Case 2 would be the
best he could do because it is at target, and purifying more hydrogen does not reduce
hydrogen consumption. From a hydrogen-consumption point of view, the proposed process
changes are clearly a bad idea. However, the real benefit of hydrogen management is
derived from viewing it as a means to increase refinery profit. When the options are judged
based on overall refinery economics (Figure 7), the profit improvement achieved by
combining network analysis with process know-how and refinerywide economics is far
superior (more than $7 MM/year) to the option selected based on network analysis alone.

EUROPEAN REFINERY

Summarizing the results of the hydrogen management study of a much-simpler European

refinery demonstrates that the previous example is not an isolated case. The refiners
primary concern here was reducing future hydrogen plant requirements as new fuel
specifications come into effect. Figure 8 shows that a suboptimal solution would have been
selected if process concerns and refinerywide economics were not considered. Process
improvements can increase profitability by an additional $1.2 MM over network analysis

AM-99-08
10
alone. In addition, 29 MM SCFD of potential additional hydrogen production was made
available through revamps of the existing catalytic reformers. The increased hydrogen
production results from both increasing naphtha charge rate and reducing reactor pressure.

IMPLEMENTING THE H2 NETWORK MODEL

WITHIN A REFINERY LP MODEL

Linear programming, a mathematical technique that has been used to optimize the operation
of refinery and petrochemical complexes for more than 30 years is the method used by most
refiners to make economic decisions. The refinery LP model includes simplified process
models that include, for example, product yields and properties, utility consumption rates,
unit capacity limits, and blending rules. Hydrogen is typically modeled on an extremely
simplistic basis. Until now, refinery hydrogen systems have commonly been modeled by
assuming fixed hydrogen purities for each process stream. Makeup and vent flows do not
change with makeup purity. The LP forces a hydrogen material balance so that excess
hydrogen ends up in the fuel system and hydrogen shortages force increased production in
the hydrogen plant.

The application of pinch analysis to refinery hydrogen management has highlighted the
need for more-rigorous LP modeling. The refinery hydrogen network and all other aspects
of the refinery operation should be optimized simultaneously. To achieve overall optimum
refinery operation, new insights and an understanding of the refinery hydrogen network
problem need to be incorporated into the conventional LP model of the refinery complex.

The results from optimizing the refinery hydrogen network obtained to date by LP
modeling have confirmed many of the old heuristic rules of refinery hydrogen management,
for example, that hydrocrackers should be fed high-purity hydrogen and low-purity
hydrogen streams are best used as hydrogen plant feedstock. New insights, particularly with
respect to the optimum use of the medium-purity hydrogen streams, have also been
obtained. Hydrogen management and the individual process units can thus be optimized at
the same time. Addressing the hydrogen network within the LP model ensures that the
hydrogen management study is not carried out in isolation from the overall refinery
optimization.

AM-99-08
11
 CONCLUSION

Improved hydrogen management offers the refiner a means of not only cost-effectively
meeting the increased hydrogen demands of low-sulfur fuels, but also improving
profitability. The hydrogen pinch methodology can provide significant savings in hydrogen
production and compression costs. However, benefits far in excess of what hydrogen pinch
alone can achieve are realized by considering the effect of hydrogen on hydroprocessing
performance and using increased overall refinery profitability instead of reduced hydrogen
consumption as the objective. This approach takes into account the effect of hydrogen
partial pressure on throughput, yield, product quality, and catalyst life in critical reactors.
The methodology described in this paper has been tested on a number of refineries and has
delivered significant profit improvement in every case.

REFERENCES

1. J.J. Alves, and G.P. Towler, Design of Refinery Hydrogen Networks, workshop
presented at UMIST PIRC Hydrogen Consortium meeting, Manchester, April 1998.
2. J.J. Alves, and G.P. Towler, Analysis and Design of Refinery Hydrogen Distribution
Systems submitted to Ind. Eng. Chem. Res., in preparation.

AM-99-08
12
 Figure 1. H2 Composite Curves
100
95
90
85
H2 Purity, %

80
75
70
65
60 Sources
Sinks
55
50
0 5,000 10,000 15,000 20,000 25,000 30,000
Gas Flow Rate, mol/hr
UOP 3125-1

Figure 2. H2 Purity Surplus Curve

100
95
90
85
H2 Purity, %

80
75
70
65
60
55
50
0 50 100 150 200 250 300
H2 Surplus, mol/hr
UOP 3125-2

AM-99-08
13
 Figure 3. Initial H2 Network

Isom Fuel
HT 1
Reformer 1 HT 2
HT 3

HT 4

HT 5 Fuel
H2 Plant 1
HT 6 HT 7

Reformer 2 HT 8
Nonpermeate
Permeate
Membrane Fuel

HT 9 HT 10 Fuel
H2 Plant 2

HT 11
UOP 3125-3

Figure 4. Relationship of Capital Investment

and H2 Savings
25,000

20,000
H2 Saved, Nm3/hr

15,000

10,000

5,000

0
0 .5 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Total Capital Investment, $MM
UOP 3125-5

AM-99-08
14
 Figure 5. Revamped H2 Network

Isom Fuel
HT 1
Reformer 1 HT 2
HT 3
New Increased Reactor Purity
Membrane HT 4 Fuel
Fuel
HT 5 Fuel
H2 Plant 1
HT 6 HT 7

Reformer 2 HT 8
Nonpermeate
Permeate
Membrane Fuel
Increased Reactor Purity
HT 9 HT 10 Fuel
H2 Plant 2

HT 11
UOP 3125-4

Figure 6. Hydrogen Savings

10
H2 Saved, % of Base Case

-5

-10

-15
Improved New Process
Network Purification Optimization

UOP 3125-6

AM-99-08
15
 Figure 7. Potential Profit Improvement
10.0

8.0 Process Improvements

Profit, $MM/yr

6.0

4.0

2.0
Network Only

0
0 1 2 3
Investment, $MM
UOP 3125-7

Figure 8. Potential Profit Improvement

for Second Refinery
5.0
Process Improvements
4.0
Profit, $MM/yr

3.0 Network Only

2.0

1.0

0
0 1 2 3
Investment, $MM
UOP 3125-8

AM-99-08
16