Simple, Safe and Reliable KRES:

First Commercial Application

The first commercial installation of the Kellogg Reforming Exchange System (KRES) has been com-
pleted, which enhances plant operability, reliability and safety while offering advantages in capital
and operating costs and reduced emissions. KRES can contribute significantly to steam reforming in
ammonia and methanol production.

J.R. LeBlanc, Robert V. Schneider, m, and Kenneth W. Wright

M.W. Kellogg Company, Houston, TX 77210
Bob Lai
Methanex Corporation, Kitimat, BC V8C 267, Canada
Introduction tion section contains several coils that are used to
recover heat for both the process and utility systems in
the plant. Apart from being energy-intensive to oper-

T
he steam reforming of hydrocarbon feeds is the
predominant unit operation in the production of ate, the steam reforming section of the typical synthe-
synthesis gas, an intermediate used in produc- sis gas generating plant accounts for a significant por-
ing both ammonia and methanol. For ammonia, steam tion of the plant capital cost.
reforming accounts for about 75-80% of the world's Today, there is a commercially available alternative
production. In methanol, it would be even higher at to the classic steam reforming process — Kellogg
80-85%. Reforming Exchanger System or KRES. The KRES
As practiced by the petrochemical industry, steam unit replaces the complex fired steam reformer
reforming dates back to the 1930s. Early steam through the use of a unique process configuration and
reformers were primarily used to produce hydrogen. relatively simple heat-exchanger design. This article
During World War II, ammonia hi the U.S. was pro- describes the KRES system and its benefits, signifi-
duced through the steam reforming of natural gas. cant analytical and mechanical features of KRES and
From these early beginnings, steam reforming has the startup of the first commercial unit at Methanex
evolved as the overwhelming process of choice for the Corporation in Kitimat, BC, Canada.
production of hydrogen/carbon monoxide synthesis
gas in a fired furnace. KRES Application and Benefits
Fired steam reformers are, however, complex units:
they require close attention to operating safely and Process description
reliably. Steam-hydrocarbon reforming is accom-
plished in the radiant section of the fired furnace. The KRES process is described in Figure 1. The
Typically, this operation has a thermal efficiency of preheated mixed feed consisting of hydrocarbon feed
about 50%. To increase efficiency and improve energy plus steam flows in parallel to both the reforming
utilization, all steam reforming furnaces have a con- exchanger and an adiabatic or autothermal reformer.
vection section following the radiant zone; the convec- The oxidant, steam, and a portion of the hydrocarbon

AMMONIA TECHNICAL MANUAL 38 1996

 outside of the tube is minimized
being limited to the unit pressure
drop.
It is this open-tube arrangement
which makes KRES unique and
has resulted in two Kellogg U.S.
patents.

Benefits overview

Kellogg's activity in steam

reforming dates back to the origins
of commercial practice, with more
than 50 years of experience in both
analytical and detailed reformer
design. It is this experience which
in large part led to the KRES con-
Figure 1. Reforming exchanger diagram. figuration. Understanding the
severity of the reforming operation
feed are introduced as separate feeds to a specially from a designer's standpoint resulted in first address-
designed mixing zone at the inlet of the autothermal ing the mechanical aspects of the reforming exchanger
reformer. Inside the autothermal reformer, two basic design. By defining the mechanical and maintenance
reactions take place: the exothermic partial oxidation
of the feed hydrocarbon and the endothermic reform-
ing of the hydrocarbon feed with steam.
The effluent from the autothermal reformer, typical-
ly at about 1,000°C, flows to the reforming exchanger
shell inlet. The remaining portion of the hydrocarbon
feed and steam (that portion not fed to the autothermal
reformer) feeds the tube side of the reforming
exchanger. As the hydrocarbon-steam mixture flows
inside the tubes, it comes into contact with conven-
tional steam reforming catalyst, which is used to pro-
mote the endothermic reforming reaction. The heat TO HEAT
RECOVERY

required for this reaction is provided from the shell-

side gases, which consists of the combined effluent WATER JACKET
from the autothermal reformer and the tube side of the
reforming exchanger. The shell-side gases exit the
exchanger for further downstream processing after
heat exchange with the reacting tube-side gases. This KHFOOATED
OBntBUTOft
process configuration would basically remain the same
for either ammonia or methanol applications.
As seen in Figure 2, the reforming exchanger in the
KRES unit is of the open-tube design. The catalyst-
filled tubes are suspended from a single tube sheet
located at the cold end of the exchanger. With the
open-tube configuration, each tube is free to expand
individually without any restriction to its growth.
Also, pressure differential between the inside and the Figure 2. Open tube reforming exchanger.

AMMONIA TECHNICAL MANUAL 39 1996

design features that would be the key in producing a the steam generated from process heat can, by and
unit adaptable to large-scale plants and would be low large, supply the steam necessary for all major turbine
in cost and easily serviceable, the development effort drives. With the KRES unit, flexibility is available to
initially focused on hardware. Typically in the process the designer to select drivers that take maximum
industries, a process scheme is devised and the equip- advantage of project specifics.
ment is designed to accommodate the process require- As mentioned above, one significant benefit of
KRES is reduced emissions of NO, and CÛ2- This is
Table 1. Relative Emissions From Conventional due to elimination of the fired steam reformer flue-gas
Steam Reforming Operation Compared to KRES stream. Kellogg's experience with large-scale ammo-
nia plants provided data indicating the relative emis-
sions from a conventional steam reforming operation
Conventional KRES
compared to KRES (see Table 1).
Total Emission Gases 100% 40%
In the emissions comparison (Table 1), included in
NO, 100% 27%
the KRES unit is a preheat furnace used to heat the
CO2 100% 25% feedstreams to reaction temperature. This preheat fur-
nace accounts for the total gas emissions from KRES.
In operating the conventional fired steam reformer,
ments. In this case, the key equipment characteristics significant attention is required to monitor and regu-
were defined and the process scheme was configured late the multiple burners contained within the unit.
to encompass the desired exchanger features. The Depending on furnace size, there can be as many as
result is KRES. Relative to the fired steam reforming 300+ burners in a large-scale single-train steam
approach, the benefits for utilizing KRES can be sum- reformer. The KRES unit itself has no direct firing.
marized as follows: The feed preheat furnace will have only a fraction of
the services required in a normal steam reformer.
• Lower capital cost Thus, operator attention is greatly reduced with
• Improved energy efficiency KRES. Experience at the Methanex installation indi-
• Easier and Faster startup cates that KRES can easily be operated by one panel
• Reduced NO, and CC>2 emissions operator and one part-time field operator.
• Less operator attention Serviceability is a key feature of KRES. Loading
• Easier to maintain and unloading of the catalyst in the exchanger are rela-
• Less plot space required. tively easy. Full access to the tube sheet and tube tops
is gained by removing the top head of the exchanger.
The KRES unit can be used in several applications, With such access, catalyst tubes can be loaded and
most notably ammonia and methanol production. The unloaded in much the same manner as is done in the
first commercial unit was installed in an ammonia conventional fired primary reformer. There are neither
plant for The Methanex Corporation at Kitimat, BC, tubes to cut and reweld, nor is bundle removal
Canada. This installation was a retrofit application. required as in other designs. The unit is serviced in
KRES is intended, however, for new grassroots plants place, making for quick, easy and inexpensive mainte-
as well as a retrofit to existing plants. nance.
Capital cost, relative to the conventional unit, is
lower due to lower equipment cost, reduced construc- Analytical Design
tion requirements, and a smaller plot. The KRES unit
utilizes heat from the autothermal reformer, which The obvious heart of the KRES unit is the reforming
provides greater flexibility in driver selection. This exchanger. The steam and reforming exchanger hydro-
can lead to reduced driver cost, e.g., electric motor vs. carbon feedstock is reformed, as these gases flow
steam turbine and/or integration of a gas turbine into through the catalyst-filled tubes. The catalyst used is
the power cycle, which can reduce energy consump- conventional nickel reforming catalyst that is in cur-
tion. In the conventional approach, heat downstream rent use. The heat supply for the endothermic reform-
of the autothermal reformer is generally used to gener- ing comes from the gas on the shell side. This gas is a
ate steam. In modern ammonia plants, for example,

AMMONIA TECHNICAL MANUAL 40 1996

mixture of autothermal reformer effluent and effluent uniform heat-transfer coefficients; therefore, this
from the tube side of the reforming exchanger. option yielded the lowest variation in tube tempera-
In developing the KRES analytical design, Kellogg tures and hence degree of reaction within the catalyst-
employed its proprietary steam reforming program filled tubes. The segmental baffle arrangement imparts
that models both the reaction kinetics and equilibrium significant strength to the bundle assembly and mini-
of the reforming and shift reactions. This program has mizes tube vibration in the unit during normal opera-
been used extensively to design numerous Kellogg tion. The small variation in tube-to-tube performance
direct-fired steam reformers. It has been checked with predicted for the preferred design allowed for a unit
field data over the years and consistently updated. design without a long unbaffled section at the bottom
The program simultaneously considers heat transfer of the exchanger. This resulted in significant cost sav-
and reaction kinetics/equilibrium in calculating the ings, since any long unbaffled section would have a
design. With such a tool, detailed temperature and relatively lower heat-transfer coefficient which would
stream data are made available to provide for a com- in turn result in a unit perhaps 15-20% longer than the
plete understanding of the reactor thermodynamics optimal design that was finally chosen for KRES.
and its impact on the mechanical design. Figure 3 shows the shell- and tube-side temperature
Extensive analysis revealed that the segmental no- variation for an ammonia syngas application using
tubes-in-the-window baffle option provided the most segmental no-tubes-in-the-window baffles. This con-
figuration allows for a high
degree of accuracy in predicting
shell-side gas temperature at the
baffle turns, and hence the tube
metal temperatures are calculat-
ed on a firm basis. The metal
SHELLStOE FLUID TEMPERATURE
temperature of the tubes is an
REACT AMTS TEMP • DEQ C
DEGC MOL % METHANE (DRY) important design constraint, and
accurate prediction leads to a
well established mechanical
A B C design.
50t 509 509 The example in Figure 3
0
932
932 932
shows that in the hottest baffled
«41 83« 613
t 56.0 59.« CZ4 segment at the bottom of the
2
tat 670 678 exchanger, the shell-side bulk
34.9 36.1 3S.t
721 709 705
gas temperature changes from
3
21.4 23.0 23.1 986°C to 959°C at the baffle
749 750
4
14.3 14.8
759
14.0
turn. Thus, the tubes in this baf-
5
803 792 791 fle section see a shell-side gas
829 9.18 9.07
temperature variation of 27°C.
837 637 847
e 5.08 525 4.69 The gas temperature of the tube
7
896
225
884
2.65
884
2.60
side varies from 933°C to
933 933 942
942°C, some 9°C difference; the
«
1.19 121 1.05
bulk gas temperature average is
about 938°C. This in turn leads
to reasonably small variations in
residual methane from tube to
tube. Similar analysis for each
section of the exchanger can be
made to understand temperature
and stream composition varia-
Figure 3. Actual variation reactants temperature and methane fraction. tions and the impact these vari-
ables have on the detailed unit

AMMONIA TECHNICAL MANUAL 41 1996

 type distributor is used to direct the flow of
100 autothermal reformer effluent upward the
gas exiting the catalyst-filled tubes. This
mixed stream then passes through a 180°
turn before making its first pass over the
tube bundle at the bottom of the reforming
exchanger. The distributor design perfor-
mance was checked by applying computa-
tional fluid dynamics (CFD); Kellogg uses a
45000 CFD program called FLUENT. FLUENT
input considers the specifics of the design to
be simulated including both physical geome-
35000
try and fluid conditions. Figure 6 shows
temperature isotherms predicted for the
25000 mixed gas stream as it first passes to the tube
bundle just above the bottom-most baffle.
The predicted temperatures in this stream
15000 range between 977°C and 992°C; the ideal
mix temperature is 985°C. The total range of
temperatures predicted for this stream is
15°C with deviation around the ideal temper-
2SO
10 20 30 40 50 60 70 «O 90 100 ature being +6°C and -9°C. This analysis
PERCENT OF ACTIVE TUBE LENGTH (TOP" O) indicates that the degree of mixing achieved
is quite good and that thermal gradients due
Figure 4. Reforming exchanger performance profiles. to lack of mixing are insignificant.
Currently, we have completed reforming
design and operation. The greatest temperature differ-
exchanger designs both for the first commercial appli-
ence between the shell and tube side is exhibited at
cation of KRES (the Methanex unit) and for a 1,000-
the top of the reforming exchanger where the
mixed feed enters the catalyst-filled tubes. It is at
this point that the local heat flux is the highest. The
flux, however, is relatively uniform over about 80%
of the tube length and, as can be noted, is about
40% of what could be expected in a fired reformer.
The majority of the reforming reaction is completed
in the top half of the tubes, whereas a close
approach to equilibrium conversion is achieved in
the bottom half. The overall heat-transfer rate is ^ TUBE CAPS

very uniform, and accurate prediction of this rate

using Kellogg's analytical tools allows us to confi-
dently size the reforming exchanger unit for the
\ PERFORATED
intended application. Variation in heat flux, temper- DISTRIBUTOR
1*. PLATE
atures, heat-transfer rate, and methane leak can also
be predicted for the entire unit. Figure 4 graphical- 1
ly presents such data for the previously mentioned
ammonia application. CA FALYST TUBE EFFLUENT 23.0«% *w e
AU rOTHERMAL REFORMER EFFLUENT W.84% ( MC
Mixing of incoming autothermal effluent gas- and
IDEAL MIXED STREAMS 100.0«% 9 ISC
tube-side effluent is a critical element of this
design. Figure 5 displays the distributor arrange-
ment used to mix these two hot streams in the bot- Figure 5. Mixing catalyst tube and autothermal
tom of the reforming exchanger. A multihole plate- reformer effluent.

AMMONIA TECHNICAL MANUAL 42 1996

 Catalyst Tube Outlet Temperature * 938.5 deg C
Autotheimal Reformer Outlet Temp. « 999.4 deg C
Ideal Mixed Shellside Inlet Temp. = 985.4 deg C
- -i 1 . _ r-._ T'

B&W KAOLITE 2300. This dual-

layer approach allows the unit to be
designed without a metallic liner
which is less expensive to fabricate
and eliminates difficulties that could
be experienced in removing the bun-
dle. Liners are used in this design
only in the nozzle areas where diam-
eters are small, and bundle removal
and reinsertion are not an issue. This
approach is now routinely offered
on Kellogg waste heat boiler
designs and has been successfully
used in a large methanol plant appli-
cation. Figure 7 displays a diagram
of the KRES shell with the dual-
Distributor Plate
layer refractory system and the girth
Figure 6. Isotherms indicating extent of mixing. flange at the top.
Catalyst tube outlet temperature is 938.5°C; autothermal reformer outlet tempera- Figure 8 is a diagram of the
ture is 999.4°C; ideal mixed shell side inlet temperature is 985.4°C. reforming exchanger tube bundle.
The tube bundle is constructed in an open tube
MTPD ammonia syngas application. Design investiga- arrangement where the reforming tubes are suspended
tions lead us to anticipate that a single reforming from a single tubesheet located at the cold end of the
exchanger would be possible for application in plants reforming exchanger. Typically, the tubes would be
up to a nominal capacity of 1,800 MTPD. 2.0 in. (50 mm) in outside diameter. The top head of
the reforming exchanger is a full opening closure that
Mechanical Design allows full access to the shell interior for bundle
removal, reinsertion, and access to the tubesheet.
The reforming exchanger shell is constructed of car- Stress on the KRES reforming exchanger tubes will be
bon steel and is refractory-lined as well as water-jack- much less than that in normal fired reformer tubes and
eted. The water jacket feature is added for both rea- also less than in other reforming exchanger designs
sons of safety and reliability. This feature is a standard where the autothermal reformer operates in series
on Kellogg secondary reformers and high-temperature with, rather than parallel to, the reforming exchanger
waste heat boilers such as those used in both ammonia itself. Stress at the bottom of the tube in the open tube
and methanol service. Use of a water jacket keeps design is based on the pressure drop across the
shell wall temperatures at close to 100°C even in the tubesheet and the catalyst-filled tube plus the full
event of internal refractory failure. Should this occur, weight of the suspended tube (loaded with catalyst).
water loss from the jacket increases markedly (as does Using stress-to-rupture data for the appropriate materi-
makeup water flow) due to steam generation, thus als selection, the expected tube life exceeds 100,000 h
alerting the operating staff to a potential problem and at the normally expected operating temperature.
allowing ample time for a safe shutdown. The interior Compared to fired reformer tubes, the opportunities
of the shell is refractory-lined using a dual-layer sys- for damage to the catalyst tubes due to an operating
tem. The hot face refractory is a dense layer of upset are much less in the case of a reforming
castable material such as HW CASTOLAST "G". exchanger since there is no possibility of direct flame
This physically rugged layer of refractory protects an impingement, and flue-gas distribution is no longer a
inner layer of insulating castable material such as relevant concern. The maximum tube metal tempera-

AMMONIA TECHNICAL MANUAL 43 1996

 The control system provides protection against heating
the tubes of the reforming exchanger without suffi-
•AGK-UP LAYER cient tube-side flow.
DBOCIAYDI The tube bundle is suspended from a single
tubesheet, which in turn is supported by a tubesheet
skirt (see Figure 9). This skirt has a top flange that
rests in a groove cut into the top shell flange which in
turn supports the entire assembly. A lifting ring is fur-
nished for handling of the bundle as necessary. The
geometry of the skirt has been checked using finite-
element analysis. Baffle plates are designed with ade-
quate clearance so that binding does not occur against
the refractory face when the bundle is removed for
inspection. Alloy tube caps will be welded to the bot-
tom of each tube as shown in Figure 10. The tube caps
will contain numerous outlet holes that will be ade-
quate to retain the ring-type catalyst that is specified
for use in the reforming exchanger tubes.
Although expected reliability of the unit is such that
tube failures should be far less common than in direct-
fired applications, allowances have been made for an
unexpected failure where operations could still contin-
Figure 7. Shell diagram. ue until a convenient time was found to shut down.
Accordingly, the top of each tube has been fitted with
ture is limited to the temperature of the shell-side gas an orifice ferrule assembly. The orifice holes are sized
rather than combustion flue gas (as in a radiant box). to limit flow through the empty tube in the event of a
rupture during operation. The CH^ leakage from the
tubes is designed to be 1.2 mol % on a dry basis.
TUBE BUNDLE Should a tube rupture, the orifice will limit flow
through the broken tube to approximately 5% of the
total inlet flow to the reforming exchanger. This would
*~ i; result in the tube-side CH^j. leakage increasing on aver-
age from 1.2 mol % to 2.3% (dry basis). Considering
the dilution effect of the autothermal reformer efflu-
ent, the net increase in CH4 leakage at the reforming
Vfl
exchanger shell exit would be from 0.3 to 0.54 dry
mol % which should still be acceptable. This allows
continued operation of the plant even after a tube rup-
9--- ture until the damaged tube is plugged, repaired, or
replaced.
rt
Materials selection for the reforming exchanger
tubes required careful consideration not only of the
normal environment the tube would be subjected to
(considering temperature and gas composition), but
also the potential for accelerated corrosion due to a
phenomenon known in the industry as "metal dust-
ing." In gas environments where high CO concentra-
tions are present, carbon may be formed that in turn is
Figure 8. Tube bundle. absorbed onto the metal surface leading to pitting or

AMMONIA TECHNICAL MANUAL 44 1996

 syngas operations and is initiated by the deposition of
Boudouard carbon formed through the following reac-
tio
2 CO ^. CO2 + Cs
The Boudouard reaction equilibria are favored by
relatively lower temperatures. However, as tempera-
ture decreases, reaction kinetics slow down. Thus,
there is a distinct region where both reaction equilibri-
um and kinetic potential favor carbon formation.
VTHBMTH WtlOTD
Propensity for the occurrence of metal dusting is pri-
marily a function of CO/CO2 ratio, gas temperature,
HYDHAUUC
EXPANDED N. and materials of construction. KRES is designed with
a relatively low shell outlet CO/CO2 ratio and a high
enough exit gas temperature such that the normal
operation is well outside the area where failures are
likely. Materials were selected for the KRES design
that are best suited for resisting attack from metal
• TUBESHEET SKIRT DESIGNED TO ACCEPT PRESSURE. THERMAL, AND DEAD
dusting, if indeed such a condition should exist.
LOADS. THIS GEOMETRY HAS BEEN SUCCESSFUL!Y CHECKED BY FINITE
DIFFERENCE STRESS ANALYSIS.

• STRAIGHT/SHORT TUBESHEET SKIRTS HAVE BEEN SUCCESSFULLY USED

Methanex / Kitimat: First Commercial
IN 1Î1-C AMMONIA CONVERTER FEED EFFLUENT EXCMANOERS
Application of KRES
Figure 9. Tubesheet skirt.
Project overview

KRES was first proposed to Ocelot Ammonia com-

pany (a joint venture between Methanex corporation
and Devco in Tulsa, OK) in the early 1990s as the sec-
ond part of a two-step plan aimed at increasing
the Kitimat plant's ammonia production capacity by
40%. In November of 1992, Kellogg's new ammonia
synthesis technology, the Kellogg Advanced Ammonia
Process (KAAP), was successfully started up at Ocelot
(Methanex) reducing energy consumption and
enabling the plant to produce up to 40% more ammo-
nia when additional syngas was made available. The
original Ocelot ammonia plant was a high-efficiency
Kellogg loop designed to produce 600 STPD of
ammonia from methanol plant purge gas captured
from the adjacent Methanex facility at Kitimat.
Because N2 is produced at the Kitimat site for ammo-
r i nia production by an air separation plant, a waste O2
stream was available for enriching air which made
Figure 10. Tube cap detail. KRES a natural application for this site. Enriched air
is the preferred oxidant for KRES in an ammonia
"dusting" of the outer layers. This physical loss of application to keep the H2/N2 ratio within prescribed
metal can eventually result in loss of structural integri- boundaries while maintaining the heat balance around
ty which can be particularly troublesome in thin- the reforming exchanger and autothermal reformer
walled tubing. The problem is well documented in

AMMONIA TECHNICAL MANUAL 45 1996

 STEAM

TUBEStDEFEED

PURQE GAS FEED

FROM HEM« PLANT

WATER WASH

C02 REMOVAL

Figure 11. Ocelot/KRES expansion project block flow diagram.

tandem. relocated from a shutdown CFI plant at Terre Haute,

The design of KRES for Methanex was under way Indiana. A block flow diagram for the expansion pro-
by 1993. In July 1994, the reforming exchanger, man- ject is detailed in Figure 11. Note from Figure 11 that
ufactured by Kawasaki in Japan, was received at the shaded equipment is being added by the expansion
plant site ready for erecting into place. The project at project, while existing equipment is shown as non-
Methanex was organized such that Kellogg was given shaded.
responsibility for the KRES area only (reforming Precommissioning of the unit began in August 1994;
exchanger, autothermal reformer, and feed preheat fur- to support this effort, the Kellogg startup team assem-
nace), while the client and a Canadian contractor bled at the site at this time. Smaller-than-usual, but
assumed responsibility for the remainder of the front- otherwise conventional, BASF reforming catalyst
end expansion that consisted primarily of equipment (G-125S 8 mm X 8 mm rings) was loaded by a

AMMONIA TECHNICAL MANUAL 46 1996

Kellogg-specified socking technique into the reform- equilibrium conversion of methane for both the
ing exchanger tubes under the supervision of the reforming exchanger as well as the autothermal unit.
Kellogg startup team. Following the conclusion of the test run, the KRES
The unit was heated up with steam and natural gas unit was fully accepted by Methanex.
prior to the introduction of enriched air into the
autothermal unit on October 6, 1994. Following addi- Operations overview
tion of the oxidant stream, combustion proceeded as
expected. Gas analyses indicated that near-equilibrium The KRES has simpler operating requirements than
conversion of methane to syngas was being achieved the typical reforming furnace and secondary reformer
in both reformers. Rates during the initial startup peri- system. The use of the fired feed preheater greatly
od ranged from 50-85% of nameplate capacity limited reduces the number of burners used, and heat distribu-
primarily by compressor availability. With improving tion in the heater is much simpler to control than in a

Table 2. Process Flowsheet

Autothermal Autothermal Reforming Reforming

Reformer Exit ReformerExit Exchanger Exit Exchanger Exit
Composition Composition Composition Composition
(Actual) (Flowsheet) (Actual) (Flowsheet)
Comp./mol.%
«2 4o.u 44.4v 04.14 sl.ss
C02 10.67 9.89 9.31 9.84
Ar 0.58 0.59 0.44 0.46
N2 33.24 33.78 25.75 26.18
CO 9.51 11.23 10.31 11.67
CH4 <0.01 .03 .05 0.30

compressor performance, 100% + rates have now been large reforming furnace. The required operator action
achieved in the KRES unit. Note that for the KAAP- during startups and shutdowns at levels above grade is
based loop at Methanex to achieve a full 40% increase minimal other than a one-time prestartup setup and
in ammonia production (850 STPD), it is necessary for routine condition monitoring thereafter. Operator
KRES to be operated at 65% rates; the stand-alone attention during normal operation requires only rou-
nameplate capacity of the KRES unit at Methanex is tine checking, usually an occasional burner or water
sufficient syngas to produce 350 MTPD of ammonia. jacket adjustment. With all that constituting the down-
On November 26, 1994, KRES syngas was first stream operating systems in a modern ammonia plant,
introduced into the Methanex KAAP-based loop. It this portion of the process is pretty well self-sufficient,
now operates at ammonia production rates of up to rarely requiring significant attention.
882 STPD. During the successful KRES performance
test run, analytical data in Table 1 was gathered which Startup
compares performance of the autothermal reformer
and the reforming exchanger to that expected accord- The startup is relatively straightforward with the
ing to the process flowsheet shown in Table 2. establishment of small low-pressure nitrogen flows
The performance test run data indicates essentially at- through the heater's three principal coils (air, natural

AMMONIA TECHNICAL MANUAL 47 1996

gas to the secondary reformer, and natural gas to the reasons. There are two basic types of unplanned shut-
KRES exchanger) and lighting burners as needed to downs. One is the sudden loss of air flow requiring
obtain the desired temperature rise rate. Heat-up rates quick reductions of the natural gas feed rate and heater
of 55°C/h have been recommended during this initial firing to 30% of design conditions, and holding there
period with higher rates expected in future startups. until the air flow can be restored or the decision is
Once 230°C is reached at the stack, a methane flow is made to shut the unit down. The other is the total shut-
established though the desulfurizer preheat coil and down of the unit, which includes the heater fuel sup-
minimum steam flows through the three major coils ply. In this case, the unit is depressured, maintaining
are established. These minimum flows are provided to the minimum coil steam flow until moderate tempera-
protect the coils if feed flows are lost or tripped off. ture levels are reached and then either securing the
Heating continues, and steam rates are adjusted to unit as described above or restarted by lighting burn-
flow rates suitable for feed initiation. Hydrogen flow ers.
for desulfurization is established, and sulfur concen-
tration samples are taken when the desulfurizer tem- Controls
perature approaches its design operating temperature.
When the feed preheater exit temperature levels suit- Due to the conditions that may be encountered in
able for feed initiation are reached, the temperatures operating this type of process, the control system is
are stabilized and sulfur-free natural gas containing the extremely important. The control system was a major
appropriate feed initiation hydrogen level is started to issue in its design, aided by valuable input by
both reformers at 30% of the design flow rate. The Methanex. Ratio controls and ratio deviation trips are
heater firing rate is simultaneously increased, and the the key parameters in this operating system. The basic
air flow is started to the secondary reformer. Once the controls and trips are the steam/carbon ratio to either
secondary reformer exit temperature rise is detected, reformer, low steam flow to either reformer, feed split
the air flow is brought up to the same percent of between the two reformers, high secondary reformer
design flow as the natural gas. After adjustments, the exit temperature, high or low burner gas pressure, and
reforming unit is brought up to the design rate along low steam drum level. The air flow is monitored or
with the balance of the process. As soon as the gas at tripped by high air/gas feed ratio or low air flow to the
the suction of the synthesis gas compressor is of suit- secondary reformer. A comprehensive burner manage-
able quality, it may be introduced to the ammonia syn- ment system completes the control package of the
thesis loop. process.
This innovative process has proven itself to be effi-
Planned shutdowns cient, flexible (turndown rates are much lower than
those obtainable by typical direct-fired furnace
Planned shutdowns are relatively simple. The feed reforming processes), and simple to operate. Operation
rate is reduced to 30% of design level, and at that of the KRES unit at Methanex has been for the most
point the unit may be either tripped off or shut down part uneventful. The initial installation of this process
in a steady reduction. The tripped shutdown will be is operating at a site which uses a large reformer in a
described later. A steady reduction shutdown would be methanol plant; the observation of the operating staff
accomplished by stopping the air flow and, after at Kitimat is that the KRES unit is simpler and easier
adjusting the heater firing, stopping the natural gas to operate.
feed flow. The heater firing and the steam flows would
be lowered to a desired level for the duration of the Minor problems and solutions
shutdown period, and then the equipment would be
blocked in and isolated. The preheat furnace has been designed to burn both
natural gas and purge fuel in the form of membrane
Unplanned shutdowns separator reject gas. Some burner instability has on
occasion been noticed particularly with the use of the
Unplanned shutdowns can occur for a number of reject gas fuel (the unit was originally started up using

AMMONIA TECHNICAL MANUAL 48 1996

 Table 3. Comparison of Methanex Reforming Exchanger Unit to Full-Size Unit Capable of Producing
Syngas For Nominal 1,000 MTPD Facility
1000 T/D Unit Methanex Reforming
Exchanger

Capacity 1,000 MTPD 350 MTPD

Duty 100% 29%

Surface Area 100% 27%

Number of Tubes 100% 33%

TubeOD 2.0 in. (51 mm) 2.25 in. (57 mm)

Tube Length 35 ft (11 m) 25 ft (7.6 m)

Pressure Shell ID 100 in. (2,540 mm) 75 in. (1,950 mm)

Catalyst Volume 100% 31%

only natural gas fuel until KRES syngas was intro- Grassroots application
duced into the ammonia loop). Additionally, draft con-
trol has been achieved by a stack-mounted semicircu- The Methanex unit is about one-third the size of a
lar flap damper that was originally intended to provide full-scale 1,000-MTPD reforming exchanger. The tube
manual control. Changing over to automatic control diameter and length are different to accommodate pro-
may require the use of a louvered-type damper mecha- ject-specific pressure drop constraints in the case of
nism for more positive control under various ambient Methanex. The basic design approach for the 1,000-
conditions. Further investigations into the feed pre- MTPD (or larger) ammonia plant, however, is the
heater burner design may be required to eliminate the same. Table 3 compares the Methanex reforming
encountering of any transient instability while firing exchanger unit to a full-size unit capable of producing
purge fuel along with natural gas. syngas for a nominal 1,000-MTPD facility.
The feed controls to the reforming exchanger and the With the first commercial unit now in successful ser-
secondary reformer were originally designed to run in vice, Kellogg is prepared to offer full-size reforming
automatic mode. Some instability has, on occasion, exchanger designs for clients that seek to obtain the
been observed with respect to the steam flow control lowest-cost, highest-reliability reforming systems
valves, which has resulted in operating the unit in available today and therefore enhance their ability to
manual control mode. This has heretofore been be a low-cost producer.
accomplished without difficulty. There has been one
erroneous trip of the system experienced due to a high Summary
secondary reformer outlet temperature reading. This is
more of an isolated nuisance than an ongoing problem The KRES design is a marked improvement in
and is a result of the fail-safe thinking that went into reformer technology that has now been commercial-
the original design of the controls system. No action in ized. The design approach employed for KRES was to
this area is presently planned. look for a mechanical design that met the goals of the
Steam rates to the parallel reformers were initially concept and, accordingly, we "fit" the process around
somewhat higher than design. Since its startup, the our choice of the best design. Throughout the develop-
flexibility of the KRES unit has allowed us to opti- ment of KRES, the watchwords adhered to consistent-
mize the operation for the operators such that steam ly have been: simplicity, reliability, and maintainabili-
rates have been running at about the design level. ty. We believe that these goals have been achieved in

AMMONIA TECHNICAL MANUAL 49 1996

the design we now offer to the industry. Simplicity in fired applications and likely better.
was achieved through the use of a single tubesheet Conventional, proven nickel-on-alumina catalyst is
located at the cold end of the unit that will operate at a specified for use in KRES. When changeout time does
low differential pressure. There is no bayonet system arrive, unloading and reloading will be straightforward
or attendant bayonet insulation, and the use of expan- using techniques already in practice in more conven-
sion joints has been avoided. We believe the system tional tubular reformers. Catalyst loading and unload
will prove to be extremely reliable. Use of a dual-layer ing may be accomplished with the reforming exchang-
refractory system without a liner and with a full water er bundle in place.
jacket will insure that this unit is intrinsically safe.
Further, operation in the unlikely event of a tube rap- Acknowledgment
ture is still possible through the use of simple, but
effective, inlet orifice ferrules. We expect KRES to be We would like to recognize Methanex Corporation,
very user-friendly with respect to maintenance. The particularly, the Kitimat plant's operating staff, for
inlet channel is easily accessible via a breakout elbow their support in the development of this new syngas
at the top of the unit, and the use of a girth flange technology and help in bringing onstream the first
allows for full channel removal. The tube bundle itself commercial KRES unit.
is removable, and individual tubes can easily be without modification, would not be suitable for naph-
plugged or replaced should the need ever arise. tha. Possibly with a prereformer in front of it, it would
Catalyst life expectancy in KRES is at least as good as be appropriate.

DISCUSSION
S.Thomas, Pequiven: Mr. Schneider, what metallurgy Schneider: No you can't. I can't comment further on
do you use for the tubes? that right now, however, we will be recommending
Schneider: I use an Inconel series material. that the client inspect the exchanger at the first catalyst
Thomas: Can you specify the diameter and thickness changeout, which will probably be within the next two
of these tubes? years.
Schneider: Yes. The tubes in the Ocelot unit are 50- S. Stalin, Southern Petrochemical Industries: With the
mm (2.0 in.) ID. The tube thickness is about 3.2 mm KRES located at the exit of the secondary reformer,
(1/8 in.). there is a possibility of catalyst dust or refractory par-
Thomas: Okay. Are the tubes expanded and welded? ticles entering along with the gas stream. Will this
Do you use a seal weld or a strength weld? affect the heat transfer over a period? Has this been
Schneider: I believe it's a strength weld. taken into account in the design? What is the actual
S. Stalin, Petrochemical Industries Corp. Ltd: How operating experience?
much is the pressure drop in the reforming heat Schneider: Yes, as a matter of fact it has. As in any
exchanger? heat-exchanger design, we did not use a clean coeffi-
Schneider: The pressure drop in the unit is about 16 cient. We used a fouled coefficient for sizing the
PSI across the catalyst. exchanger and that's one of the reasons why we think
Philip Snyder, EQE International: How do you plan the heat transfer right now is much better than antici-
to inspect the tubes in future shutdowns for long-term pated due to an apparent lack of fouling.
creep? Stalin: You have incorporated this system as an add-
Schneider: We are going to unbolt the head and bring on unit to enhance the capacity of the plant?
a crane over and lift the bundle out. Schneider: That's correct.
Snyder: The tubes are grouped like a conventional Stalin: Are you developing a similar system for the
exchanger bundle. Are you going to be using eddy cur- naphtha based Ammonia plants?
rent or another method of inspection? You can't get to Schneider: I think for a naphtha based plant we would
the inner tubes for x-rays, right? have to pause and think about how we wanted to
design something like this. This particular design,

AMMONIA TECHNICAL MANUAL 50 1996

without modification, would not be suitable for nqph- Schneider: Well, actually with the inlet ferrules, we
tha. Possibly with a prereformer in front of it, it would limit the flow. Let's say we had a tube completely bro-
be appropriate. ken. I believe the calculated flow would be about 5%
Stalin: You mentioned a suitable material to avoid of the total. The increase of the methane leakage at the
metal dusting. Would you please name the material bottom of the tubes would be from about 1.2% to
being used? about 2.3%, but the net methane leakage increase at
Schneider: We use an Inconel series material. the shell side exit, after you've blended the thermal
Kanwer Khan, Engro Chemical Pakistan Ltd.: From exit gas and reforming exchange gas, would only be
the fundamental standpoint about the repairs, if the about 0.5%. So you could actually run for some time
KRES is leaking in the tube, should it be blocked or with that sort of leakage.
start again or should we change the tube itself?

AMMONIA TECHNICAL MANUAL 51 1996