Debottlenecking hydrogen plant

production capacity
Retrofitting a steam methane reformer with oxygen enhanced reforming can
substantially boost hydrogen production capacity at low capital investment

Gregory Panuccio, Troy Raybold, James Meagher, Ray Drnevich and Venkatarayaloo Janarthanan
Praxair

S
tricter environmental emissions
Comparison of common SMR retrofitting methods with Praxair’s OER retrofit1,2
regulations and increased
processing of heavier and more
sour crude oil fractions are causing Method H2 rate Cap cost Comments
Modify WGS reactor +3–5% Medium Single HTS reactor changed to two-stage
hydrogen demand to grow across HTS, HTS+LTS, or MTS design
the refining industry. Costs associ- Upgrade reformer +5–15% Med-high New catalyst; replace tubes with better
ated with new hydrogen plant metallurgy; modify pigtail/tunnel design;
construction have been escalating, so upgrade controls
Install pre-reformer +8–10% Medium Requires significant changes to reformer
refiners are increasingly looking for convective section; considerable drop in
low capital cost retrofit solutions steam export
that can expand hydrogen produc- Install OER +10–15% Low Simple installation, equipment and controls;
tion capacity at their existing non-invasive
Install post-reformer +20–30% High Large footprint; capital intensive
facilities. This article discusses
Praxair’s oxygen enhanced reform-
ing (OER) technology, which can Table 1
expand hydrogen plant production
capacity by enriching the steam downstream of the existing high to higher temperatures. Taking
methane reformer’s (SMR) combus- temperature shift (HTS) unit, and these steps can increase hydrogen
tion air with oxygen. An OER adding a pre-reformer or a post- production capacity by 5–15%, but
retrofit can be installed with low reformer.1,2 Characteristics of these will require significant capital
capital investment, quick turnaround retrofit solutions are summarised in investment and plant downtime.
and minimal plant downtime. Table 1. The hydrogen production rate
The first concept usually evalu- from an SMR can be increased by
Options for increasing hydrogen ated is to increase the firing rate of 3–5% by modifying the water-gas
supply capacity the primary reformer by burning shift (WGS) reactor system in order
A refinery manager has many more fuel and increasing the heat to convert more of the residual
options to consider when opera- available in the radiant section of carbon monoxide in the syngas into
tions are limited by hydrogen the reformer furnace. As a result, hydrogen and thereby improve
supply from an on-site SMR. For more heat is transferred to the hydrogen output without increas-
example, hydrogen could be reforming reaction within the tubes ing the flow of process gas or flue
purchased from a liquid hydrogen and additional feed gas can be gas, or raising the reformer firing
supplier or the hydrogen supply processed. A new induced draft fan rate. A two-stage HTS configuration
capacity could be increased by may have to be purchased to can be installed with interstage
constructing an additional on-site accommodate the additional flow of cooling; the existing HTS reactor
SMR or by replacing the existing flue gases, and a significant increase can be replaced with a medium
SMR with newer and larger equip- in flue gas stack temperature can temperature shift (MTS) reactor; or
ment. These solutions can be costly, result unless the convective section a LTS reactor can be installed
so it is often most economical to heat recovery system is modified. immediately downstream of the
debottleneck the existing hydrogen Furthermore, the resulting increase HTS reactor. LTS units can be diffi-
plant. Approaches typically consid- in reformer tube wall temperature cult to operate and additional
ered for debottlenecking include will reduce the life of the tubes and make-up fuel is required to replace
increasing the firing rate of the significantly add to plant mainte- the lost PSA tailgas. Furthermore,
reformer, replacing existing tubes nance costs. Increased maintenance modifying the WGS design will
with ones with a larger diameter or costs can be defrayed by upgrading require a moderate capital invest-
improved metallurgy, adding a low the reformer with new tubes made ment and significant downtime for
temperature shift (LTS) reactor of better metallurgy that is tolerant installation.

www.digitalrefining.com/article/1000471 REVAMPS 2010 17

 either delivery method, OER
3YNGASTOBACK END increases oxygen concentration and
decreases the concentration of inert
nitrogen in the combustion air.
Since the oxygen concentration is
&LUEGASTO higher than in normal air, the
HEATRECOVERY furnace firing rate can be increased
without raising the volumetric flow
rate of flue gas that is processed in
the convective section of the
&IRED
REFORMER reformer. And because the flue gas
flow rate is unchanged, no modifi-
2EFORMER cations to the flue gas heat exchange
TUBES equipment or the ID fan are
required to maintain the reformer’s
performance. A secondary effect of
increasing the oxygen concentration
in the combustion air is that the
"URNERFUEL additional heat that is generated is
0ROCESSFEED released over the same distance or
#OMBUSTIONAIR
even over a shorter distance than
,ANCED/ with air. This results in an increase
3PARGED/
in the heat flux into the tubes near
the inlet of the reformer where the
Figure 1 Simplified schematic of an SMR furnace radiant section with an OER retrofit. tubes are coldest. Process gas flow
Combustion air is enriched by either premixing oxygen with air in the burner air feed to the reformer tubes is increased to
duct (sparger) or directly injecting oxygen into the burner flame (lance). This illustration utilise the additional heat that
is for a bottom-fired upflow cylindrical reformer, but OER can be used in any reformer becomes available and to maintain
furnace design (eg, terrace wall, side-fired, top-fired)
the maximum tube wall tempera-
ture at the desired value. Typical
Retrofitting an existing SMR with reformer. However, installation of reformer tube heat flux and maxi-
a pre-reformer can boost produc- the retrofit will require significant mum tube wall temperature profiles
tion capacity by 8–10%. The investment of capital, plant down- with and without OER are shown
pre-reformer reactor utilises heat time and plot space. in Figure 2.
from the hot flue gas that was The characteristics of an OER The extent to which oxygen
previously used for steam produc- retrofit are also shown in Table 1. enrichment of combustion air can
tion to instead reform a portion of With OER, oxygen is used to enrich increase hydrogen production
the hydrocarbon and steam feed- combustion air in the reformer capacity in an SMR depends on
stock prior to introducing it into the furnace, which can improve hydro- many factors, including whether
reformer tubes. A pre-reformer gen production capacity by 10–15% other bottlenecks exist in the equip-
retrofit will require a moderate without increasing maximum ment upstream or downstream of
capital investment and significant reformer tube wall temperatures, the reformer furnace. Increasing
downtime, as both a new catalytic modifying the induced draft (ID) the oxygen concentration in the
reactor and a flue gas convective fan or convective section heat combustion air to between 22% and
section reheat coil must be installed. exchanger design, or appreciably 23% can typically result in a 10–
Pre-reforming will also result in affecting by-product steam produc- 15% increase in hydrogen
lower production rates of valuable tion rates. Further-more, an OER production capacity. Field testing
steam by-product. retrofit can be installed with little and simulation results show that
A post-reformer retrofit (for plant downtime and capital between 14 and 18 tonnes of
instance, a secondary autothermal investment. oxygen are typically required to
reformer or product gas heated produce 1 million standard cubic
reformer) can increase hydrogen Oxygen enhanced reforming feet of incremental hydrogen from
production by as much as 30% if A simplified schematic of the radi- a bottlenecked SMR. One of the
the SMR is not bottlenecked by the ant section of an SMR furnace that other advantages of OER technol-
downstream syngas processing includes OER is shown in Figure 1. ogy is the operational flexibility
equipment. With a post-reformer, a The combustion air is enriched with inherent to the retrofit. If incremen-
significant portion of the feed is oxygen by one of two means: either tal hydrogen production is not
being reformed outside of the oxygen is premixed with the air via required for a period of time, the
primary reformer, so hydrogen sparger in the air feed ductwork or flow of oxygen can be turned off
production is increased without the oxygen is injected directly into and the SMR will operate as it did
increasing the firing rate in the the burner flame via a lance. For prior to the retrofit.

18 REVAMPS 2010 www.digitalrefining.com/article/1000471

 control skid and an oxygen injec-
Max. tion device (a sparger or lances).
OER OER The oxygen flow control skid is
sized for the required oxygen flow
Heat Flux

Max TWT
rate and may be integrated with an
existing control system. The flow
Base Base
skid can be installed in almost any
convenient location, it has a small
footprint and can be built for
Tube length Tube length compliance with most electrical
classifications. The nature of the
Figure 2 Approximate profiles of radiant zone reformer tube heat flux and maximum oxygen supply system will depend
tube wall temperature (Max TWT) with and without OER. With OER, hydrogen production on the size of the application and
can be increased without increasing Max TWT and decreasing tube lifetimes location of the refinery. If the refin-
ery does not currently have oxygen
Oxygen delivery: lance vs sparger Incompatibility issues arising available in bulk on-site, a liquid
Oxygen can be injected into the from air ductwork materials may storage tank or an oxygen genera-
combustion air by either premixing limit the air enrichment level to tion plant may be installed.
it with the air via a sparger in the 23.5% oxygen for the sparger case. Depending on the refinery location,
air feed ductwork, or by directly According to Compressed Gas it may be possible to draw product
injecting oxygen into the combus- Association (CGA) standards, air from a nearby oxygen pipeline
tion zone within the furnace via a that contains more than 23.5% network. The supply system and
lance. There are advantages and oxygen is subject to different safety flow control skid can typically be
disadvantages to both methods. standards compared with normal installed without shutting down
A sparger is simpler to install air. For lances, the enrichment is hydrogen plant operations.
than a lance. A sparger requires a not limited to the equivalent of Installation of the sparger or lances
single penetration into the air duct- 23.5% oxygen in air because could take up to one day of SMR
work, whereas lancing requires oxygen is added to the furnace outage.
multiple penetrations into the SMR separately from the air supply. Safety and environmental reviews
furnace near the burners. It is more Therefore, there are no incompati- of the proposed design are
difficult to install all the piping bility issues with air duct materials conducted before construction
required to feed oxygen to multiple because the air duct is only exposed begins. Standards and procedures
lances than the single pipe required to ordinary air. Hydrogen produc- are applied to ensure that the OER
to feed oxygen to a sparger. tion capacity can usually be retrofit is installed and operated
Furthermore, the lance nozzle’s increased to desired levels without safely. New interlocks and alarms
orientation may have to be adjusted raising the concentration of oxygen are programmed into the control
once it is installed, to optimise in air beyond 23.5%, so either a system to monitor critical operating
operating conditions. sparger or lances would be parameters. The flow of oxygen to
One disadvantage of the sparger suitable. the furnace will automatically be
is that NOx production within the The oxygen injection method is shut down if design limits are
furnace will increase. The adiabatic custom designed for each retrofit. exceeded.
flame temperature within the burner Although a sparger is much easier An OER retrofit with a liquid
rises as the concentration of oxygen to install than lances, NOx emis- oxygen supply system can be
in the air increases, which in turn sions or oxygen enrichment limits installed and commissioned within
rises the production of thermal NOx. may dictate that lances are the best six to 12 months from contract
The extent to which NOx emissions option. completion. Detailed SMR design
increase will depend on the design and operating data are required to
of the burners and the recirculation OER implementation engineer the retrofit, since each
patterns within the furnace. Field OER is a low-cost solution because, OER installation is custom config-
testing has confirmed that lances can unlike most other reformer retrofit- ured for the SMR in which it will
be installed so that NOx emissions ting solutions, there is no new be used. To expedite the project
(on a lb/mmbtu fired basis) do not expensive equipment to install schedule, an assessment of retrofit
increase as oxygen is added to the (such as catalytic reactors) and no performance can be performed
SMR furnace. A lance installation significant modifications to the prior to contracting and the results
can be NOx emission neutral because existing reformer furnace are can be reviewed along with the first
the lances can be designed to stage required as part of the installation. draft of the contract.
the oxygen consumption within the OER retrofit equipment is simple,
combustion zone and increase recir- safe and reliable. New equipment OER demonstration
culation of cooler flue gases within to be installed includes an oxygen OER technology was field tested at
the furnace. supply system, an oxygen flow one of Praxair’s cylindrical, up-

www.digitalrefining.com/article/1000471 REVAMPS 2010 19

 Cost of incremental hydrogen
Summary of OER field testing results
produced
Incremental hydrogen from an OER
Operating condition Base OER OER retrofit can be produced on a cost-
O2 injection device None Lances Sparger
H2 production rate 100% 111% 110% competitive basis with the baseline
O2 in air1 20.8% 21.9% 22.1% hydrogen. For example, assume
O2/H22, tonne/mmscf N/A 20 18 that economic parameters (cost of
Natural gas feed + fuel 100% 109% 108% NG, power, make-up water and
∆Maximum TWT3, °F Base -1 +2
∆Reformer outlet temp, °F Base +1 +4 credit for steam) are such that the
PSA feed CH4 4.9% 5.6% 5.4% total variable cost of producing the
NOx emissions2, lb/mmbtu 100% 88% 120% baseline quantity of hydrogen from
the plant above is $2.25/mscf
1 Equivalent O2 enrichment level for OER cases with lances
2 Corrected for equivalent flue gas flow rate and flue gas O2 concentration 3 Calculated (not measured) hydrogen (thousand standard cubic
feet). Based on current market
Table 2 prices for liquid oxygen, the cost of
producing incremental hydrogen
fired, up-flow SMRs that utilises a lance and sparger cases were able to from the OER retrofit during the
PSA for hydrogen purification. Both increase SMR production capacity field demonstration would be
premixed (sparger) and lanced by approximately 10% by enriching approximately $2.90/mscf hydro-
oxygen delivery methods were the combustion air to an equivalent gen. Furthermore, previous
tested. Most of the OER equipment 22% oxygen. Enrichment oxygen experience has shown that the total
and controls were installed over a was consumed at a rate of 18–20 installed cost of an OER retrofit for
period of several weeks. Only a tonnes per mmscf of incremental the SMR operator can range from
single day’s outage was required to hydrogen produced; it was esti- $500 000 to $1.5 million, depending
make the final oxygen connections: mated prior to testing that it would on the size and design of the retro-
the sparger into the air duct and take 19.4 tonnes of oxygen per fit. The low capital investment
the lances into the furnace floor. mmscf hydrogen. The SMR hydro- along with the reasonable cost of
The results of testing are summa- gen output increased by 10–11% in incremental hydrogen produced
rised in Table 2. The results for each the OER tests, but the natural gas leads to short payback times on the
operating condition are from exper- (NG) consumption rate only retrofit investment.
iments representative of those increased by 8–9%. The efficiency of
operating conditions. Each of the using OER to produce incremental Conclusions
values reported is given from a heat hydrogen is higher than the baseline Demand for hydrogen within the
and mass balance process simula- hydrogen production efficiency. The refining industry will continue to
tion of the SMR that was reconciled calculated maximum tube wall grow as emissions limits for trans-
to the experimental plant data. temperature and observed reformer portation fuels become more
Some of the values that are reported outlet syngas temperature changed stringent and the nature of the
in Table 2 (for instance, NOx emis- only slightly from test to test. crude oil supplies becomes more
sions in lb/mmbtu fired) are not However, reformer methane slip severe. The economics associated
directly measured, but are deter- increased because the residence time with building new on-site supply
mined from a combination of direct of the process gas within the or replacing old hydrogen produc-
measurements and calculations reformer tubes decreased and the tion equipment dictate that refiners
made in the plant process simula- pressure in the reformer tubes rose focus on debottlenecking existing
tion. When practical, observed and as the hydrogen production rate SMRs. Simple, reliable equipment,
calculated values were renormal- increased. low capital investment, incremental
ised to equivalent operating At equivalent excess airflow, NOx hydrogen production capacity and
conditions within the correspond- emissions actually decreased competitive incremental hydrogen
ing process model for the executed slightly when the lances were used production costs make OER tech-
experiment, as it was much easier to inject oxygen directly into the nology a good choice for retrofitting
to hold key operating parameters burner flame. As expected, NOx an existing bottlenecked SMR.
constant in a process simulation emissions increased for the sparger
than it was during actual SMR experiment, where oxygen was
operation. Note that these results premixed with the combustion air
may not reflect achievable outcomes upstream of the burner inlet. Stable
for other OER installations. Actual furnace operation was observed References
OER operating conditions will vary, during the OER testing as well as 1 Drnevich R F, Fenner G W, Kobayashi H,
dependent on the SMR design and during the transitions into and out Bool L E, 2006, Production Enhancement for a
the way in which the SMR is oper- of the OER operating condition. Reactor, US Patent 6 981 994.
ated, but the results shown here Transitions into and out of OER 2 Ratan S, Vales C F, 2002, Improve your
hydrogen potential, Hydrocarbon Processing,
should be fairly representative. were accomplished within 20 to 60
81, 3, 57–64.
Field testing showed that both the minutes.

20 REVAMPS 2010 www.digitalrefining.com/article/1000471

Gregory J Panuccio is an R&D Development James P Meagher is a Senior Development industry for Praxair’s Global Hydrogen Business
Associate in Praxair’s H2 and Energy Process Associate at Praxair’s R&D Department in New in Houston, Texas. He holds a MS degree in
Technology group, Tonawanda, New York. York. He has a BS in mechanical engineering material science from Annamalai University,
He holds bachelors degrees in chemistry from the University of Notre Dame, USA, and India, and a PhD in polymer engineering from
and chemical engineering from Youngstown a MS in chemical engineering from the State Indian Institute of Technology, Bombay.
State University, USA, and a PhD in chemical University of New York at Buffalo. Email: [email protected]
engineering from the University of Minnesota, Email: [email protected]
USA. Email: [email protected] Ray Drnevich is Corporate Fellow for Praxair’s
Troy M Raybold is a Process Engineer in Hydrogen and Energy Technology organisation.
Links
Praxair’s Global Hydrogen Project Execution He holds a BS in chemical engineering from the
organisation. He holds a BS in chemical University of Notre Dame and an MS in water More articles from the following
engineering from Penn State University, resources engineering from the University of category:
USA, a PhD in chemical engineering from the Michigan. Email: [email protected]
University of Delaware, USA. Venkatarayaloo Janarthanan is a Senior Global Revamps, Shutdowns and Turnarounds
Email: [email protected] Business Development Manager for the refining

www.digitalrefining.com/article/1000471 REVAMPS 2010 21