You are on page 1of 7

Improved stripper efficiency raises

upgrader production
Revamping a heavy oil upgrader’s steam stripping tower increased tray
efficiency and achieved record throughput

MIKE GOULDING and FRED ZHANG Suncor


MICHAEL KRELA Koch-Glitsch

S
uncor Energy is Canada’s
leading integrated energy
company, with a focus in Ejector
Steam
the oil sands, and has been (150 psig)
involved in heavy oil extrac-
tion and refining dating back
to the Great Canadian Oil PCW
Sands project in the 1960s. At
Condenser
the time this was the largest
Diluent recovery
private investment in Canadian unit No. 1 6
history. The Base Plant 1.0 to 2.0 vol% 5
4 VOK
diluent in bitumen
Upgrader, located in Fort 3
2
McMurray, Alberta, has been 1 LVGO
in operation since 1967.
Steam
This article reviews the use of Diluent recovery (50 psig)
unit No. 2
Superfrac trays in a successful
0.1 to 0.2 vol%
revamp of an underperforming diluent in bitumen
steam stripping tower in the HVGO

Base Plant. The increased tray ATB bitumen


stripper VTB
efficiency provided by the trays
allowed for debottlenecking of
the downstream vacuum over- 0.85 vol%
head condenser system, diluent in bitumen
Delayed
resulting in an overall upgrader Fired heater coke unit
production increase of 5% and
record throughput.
Figure 1 Simplified process flow diagram (the pink tower is the steam stripping
Background tower)
Suncor’s Base Plant Upgrader
operation recovers diluent stripping tower with six trays overhead ejectors (see Figure 1).
from diluted bitumen feed in was put into operation in an The vacuum tower is a dry
diluent recovery units (DRU). effort to minimise diluent slip tower design with a pre-con-
The recovered diluent is then from the DRU distillation denser and is particularly
sent to a vacuum distillation tower and to reduce the load sensitive to slippage of light
unit (VDU). In 1998, a steam on the downstream vacuum naphtha (diluent).

www.eptq.com PTQ Q2 2016 1


impact of improving the strip-
ping tray efficiency on the
overall unit. The existing tray
efficiency (~0%) was modelled
by treating the stripping
section as a flash column with
no steam flow. In addition, a
typical efficiency range for this
service was modelled. The
evaluation also reviewed the
stripping tower operation and
existing mass transfer equip-
ment, identifying serious
operational and design issues.
Both the bitumen and steam
Figure 2 Existing liquid feed arrangement feeds were found to be areas of
concern.
The existing bitumen feed
distributor is a 12in (305mm)
open nozzle flowing into a
centre false downcomer above
From h1 = 0.091m and V1 = 4.22m/s
to h2 = 0.53m
the top tray (see Figure 2). The
existing nozzle was undersized
for this type of feed arrange-
ment, with a false downcomer
inlet velocity of nearly 14 ft/s
(4.2 m/s). With this existing
Velocity 4.22m/s arrangement, hydraulic calcula-
Landing height h1 0.091m
tions showed that the jump
Height after jump h2 0.53m
Weir height ~0.4m
height would far exceed the
open channel (false down-
comer) height. The calculated
Figure 3 Hydraulic jump calculations for existing liquid inlet liquid height after the first jump
was 20.9in (0.53m), which is
Performance of the stripper Technical evaluation well in excess of the 16in (0.4m)
has been poor from the outset, In 2012, Suncor and Koch- tall false downcomer (Figure 3).
with very little, if any, stripping Glitsch performed a technical As a result of the hydraulic
being measured. Operations evaluation of the tower with a jump, the performance of the
saw a diluent slip of 0.85%, view towards improving top tray would be poor due to
corresponding to 857 b/d dilu- performance. A simulation the majority of the liquid short-
ent slipped to the vacuum unit study was done to assess the circuiting the tray flow path.
and recovered in the overhead To improve liquid distribu-
system as vacuum overhead tion, a perforated feed pipe
kero (VOK). The pre-condenser distributor was installed (see
is not designed to condense this Figure 4). The use of a feed
amount of light naphtha, with pipe would ensure that all feed
most of it slipping to the first- liquid would remain within the
stage ejector and impacting false downcomer, and that
vacuum. As a result, the exist- there would be uniform distri-
ing overhead system places a bution feeding onto both active
constraint on total upgrader area panels. A new false down-
production. Figure 4 New liquid feed distributor comer was designed to match

2 PTQ Q2 2016 www.eptq.com


the inlet panel dimensions of • Applying a single valve
the Superfrac tray. To minimise layout (open area) across the
the turnaround time, a stab-in entire zone even though there
arrangement was used to elimi- are large changes in vapour
nate welding to the vessel wall. flow from tray to tray
The existing steam distribu- • Poor liquid feed distribution
tor was modified from a to the top stripping tray
V-baffle deflector plate to a • Poor steam distribution
perforated pipe distributor (see under the bottom stripping
Figure 5). While the velocity tray
through the nozzle is relatively • Fouling that can negatively
low, it was determined that the affect performance if not
increase in pressure drop asso- accounted for in tray design
ciated with flow through an Figure 5 Modified steam distributor (valve selection, orifice size,
orifice would help ensure and so on)
uniform vapour flow to both cation between the selected • Tray mechanical design not
active area panels of the vendor and the EPC company, suitable for potential upsets,
bottom tray. The new feed resulting in poor equipment which can result in loss of
distributor design eliminated selection and inadequate trays.
welding to the vessel wall. design. Inherent low tray effi-
ciency is further compromised. Applying the high performance
Tray efficiency The following are common Superfrac tray to heavy
Inherently low tray efficiency design errors that could lead to hydrocarbon stripping
in the stripping section of reduced efficiency: High performance trays tradi-
heavy oil towers has been well • Having the same diameter tionally are only considered for
documented in previous litera- for the flash zone and stripping increasing capacity, with a
ture.1,2 A well designed tray for section, which leads to severely reluctance to use in grassroots
this service could obtain oversized trays projects for the majority of
upwards of 25-40% efficiency;
however, values below 10% are
common.3,4 A typical grassroots 9.88e0
project will specify 4-8 trays 8.82e0
based on an assumed 25% 7.76e0
efficiency. 6.70e0
There are many factors that 5.64e0
contribute to stripping of 4.59e0
hydrocarbons having an inher-
3.53e0
ently low efficiency, including:
2.47e0
• Insufficient steam/oil ratio
1.41e0
• High relative volatility
3.51e−1
• High liquid viscosity.
Because grassroots projects −7.08e−1

will often have multiple equip- −1.77e0

ment vendors bidding, an effort −2.83e0


is made to standardise the offer- −3.88e0
ing and compare based on −4.94e0
price, working off little more −6.00e0
than data sheets, and without −7.06e0
proper review of overall tower −8.12e0
layout, diameter and feed
arrangements. This process
leads to inadequate communi- Figure 6 Computational fluid dynamics (CFD) study for single VG-0 valve

www.eptq.com PTQ Q2 2016 3


would allow us to maximise
Diluent slip vs tray efficiency comparison the stripping efficiency using
design techniques that will
Stripping efficiency Diluent slip, vol% Diluent to Reduction in diluent maximise plug flow, give a
pre-condenser, BPH slip, BPH
0%1 0.850 36 - more uniform horizontal liquid
20% 0.796 33 3 velocity profile and minimise
25%2 0.750 32 4 stagnant zones. This was
30% 0.696 29 7
37.5%3 0.650 27 9 accomplished through proprie-
tary design features, including:
1: Current operation 2: Estimated for conventional tray 3: Estimated for Superfrac tray
• Multi-chordal downcomers
to maximise flow path length
Table 1 • Optimised quantity and loca-
tion of push valves and other
refinery columns. However, • Long residence times and directional flow devices (see
application of technology stagnant areas at high liquid Figure 6)
should be based on economic flow rates resulting from use of • Anti-fouling and mechanical
impact. A hydrocarbon strip- conventional tray technology upset resistant tray features.
ping tower is the perfect • Excessive weir loadings Improvements in tray effi-
example of this concept. From • High downcomer exit veloci- ciency that increase recovery of
a purely hydraulic capacity ties leading to hydraulic jump a higher value product or lead
standpoint, it is rare that a high across the active area to a reduction in energy
capacity tray is justified. That • Short flow path lengths consumption can have fast
being said, if the maximum through the use of large, payback. Modelling the strip-
obtainable tray efficiency is straight downcomers ping section using actual trays
considered when doing the • Fouling due to coke forma- rather than theoretical stages
grassroots design, their use tion, which is a function of can provide clarity to the
would be much more common- long residence times. impact of the efficiency of vari-
place. While the operating Superfrac tray technology ous tray types. While an
point of a typical hydrocarbon allows for a customised tray assumed efficiency of 25% for a
stripper tray is in the range of design, selecting design well designed conventional
30-60% flood (including the features that best address the cross-flow tray is reasonable, a
tower in this article), the liquid and vapour flow regime Superfrac tray can provide an
following conditions conspire of a given application. In this additional increase in efficiency
to reduce the efficiency of tower, the use of high capacity of 10% over any other cross-
conventional trays: rather than conventional trays flow tray on the market. This
has been proven at the
Fractionation Research Inc.
(FRI) test facility using light
hydrocarbons,4 in a low rela-
tive volatility petrochemical
application,5 and in a heavy
hydrocarbon stripping applica-
tion similar to the one
discussed in this article.6 Koch-
Glitsch has successfully used
the Superfrac tray technology
in over 1700 columns
worldwide.
For this application, the
corresponding reduction in
Figure 7 Liquid flow distribution comparison between conventional trays and diluent slip proved to be attrac-
multi-chordal Superfrac tray tive as it would allow for

4 PTQ Q2 2016 www.eptq.com


debottlenecking of the Tower performance pre- and post-revamp
upgrader (see Table 1).
Before After
Vacuum feed stripper – Tray type Conventional Superfrac tray
operating conditions Deck type Moveable valves Minivalve fixed valves
Stripping steam 23 000 lb/hr (10 433 kg/hr) 23 000 lb/hr (10 433 kg/hr)
The Suncor vacuum feed strip- Flow path length 24.3in (617mm) 34.8in (884mm)
per is a 10ft-0in (3048mm) ID Open area, % Uniform, 10% Variable, max 6.5%
Tray mechanical design Standard 2 psi uplift
column equipped with six Tray efficiency ~ 0% 37.5%
two-pass trays. The L/V mass VOK to pre-condenser 182 BPH 150 BPH
flow ratio ranges from 23-100
across the stripping section, Table 2
with the most severe ratio
occurring at the bottom tray. Minivalve fixed valves, with more discussion regarding the
The weir load on the trays is the open area on each tray influence of flow enhancement
very high, at upwards of 230 varying to match the changing devices on residence time.
gpm/ft (171 m3/hr/m). vapour profile. The actual tray Lastly, because the stripping
Although there is not a defined vapour and liquid loadings section is prone to upset, the
maximum value for weir load, were determined to optimise trays’ mechanical features
various tray vendor design the pressure drop across the were upgraded to be able to
manuals recommend increas- valves on each tray. This is an withstand a 2 psi uplift condi-
ing the number of passes when important step to take for this tion. While it is difficult to
weir loads exceed a range of application, as the large quantify the individual impact
84-156 gpm/ft (62-116 m3/ change in vapour rate across of each design change, the
hr/m)8. However, many strip- the section is not captured cumulative effect resulted in a
ping columns with two-pass with precision using only major improvement in perfor-
trays run at excessive weir equilibrium stage outputs. mance, with the Superfrac tray
loads to avoid the complica- Proprietary design features operating at around 37.5% effi-
tions of having to use a were employed to minimise ciency post-revamp.
three-pass tray at low vapour stagnant areas and ensure A summary of the perfor-
velocity. good liquid distribution. mance of the tower pre-revamp
Please refer to the ‘liquid flow and post-revamp is presented
Results pattern comparison’ insert for in Table 2.
The focus of the revamp was
to increase the amount of light
hydrocarbon stripped; there- 50
fore, special attention was
45
Suction pressure, mmHGA

paid to employing design


40
techniques that would increase
tray efficiency. The use of a 35
Current
vapour tunnel downcomer 30
Post-revamp
resulted in the flow path 25
length increasing by 10.5in 20
(267mm). While the number of
15
eruption pools on the tray is
low due to the tower operat- 10
ing pressure, the net effect of 5
an increase in flow path length 0
is still positive. Where the
0

0
00

00

00

00

00

00

00

00

00

00
40

80
12

16

20

24

28

32

36

40

44

48

existing trays employed an


economical design with a Equivalent water vapour load at 70ºF, lb/h
single valve open area, the
new trays were designed with Figure 8 Vacuum overhead pressure comparison

www.eptq.com PTQ Q2 2016 5


Air/water residence time test data for two-pass trays (seconds) hydrocarbon stripping services.
From a commercial standpoint,
effective communication
Conventional Superfrac tray among EPC and equipment
Flow to centre downcomer 36.0s 12.0s vendors is critical to ensure a
Flow to side downcomer 7.0s 9.0s
Residence time ratio 5.1 1.3
sound engineering design.

Table 3 Liquid flow pattern comparison


and its impact on fouling
Performance after modification sent 4-8 actual trays potential
• Diluent slip to the pre-con- ■ There is a large change in Koch-Glitsch has previously
denser reduced by 8 bph vapour flow from stage to studied two-pass tray liquid
• VOK reduced by 32 bph stage that needs to be flow profiles using its 7ft-0in
• Vapour load to the vacuum accounted for when designing (2134mm) Air/Water Pilot
first stage ejector reduced by each physical tray Plant column. This column is
20%; the vacuum tower oper- ■ Koch-Glitsch has in-house of a similar size to the Suncor
ates at record lower vacuum knowledge to determine vapour vacuum stripper and was
(see Figure 8) and liquid load at each physical tested under similar weir load-
• Vacuum lift improved by 2% tray ings. A dye was injected into
vol • Superfrac tray technology the water to measure the resi-
• Upgrader reached record will maximise the strip-out rate dence time for both flow
high production, with an for a given steam/oil ratio. directions using a conventional
increase of up to 5% compared valve tray and the Superfrac
to pre-revamp operation. Conclusion tray (see Figure 9).
Optimised design aspects The results of the test show a
Steam stripping tray design – applied at the grassroots large differential in residence
keys to success project could have been time between side and centre
• Consider tray geometry achieved by proper joint design flow conventional trays due to
issues that arise when dealing review with participation from stagnant zones and retrograde
with high liquid to vapour EPC, vendors and owner’s flow (see Table 3). Conversely,
mass flow ratios, especially for engineers. All of the design the Superfrac tray uses push
multi-pass trays concepts presented, and tech- valves and other directional
• Avoid using parallel baffles nology used, have been well devices to provide a more
to reduce the effective flow established in the marketplace. uniform velocity profile.
path width in fouling From a technical standpoint, Stagnant zones around the
applications due attention should be paid to periphery of the side flow tray
• Consider the use of an inter- check the feed device design are minimised, as shown by
nal can or shroud when and ensure proper tray selec- the large reduction in residence
revamping a tower that has the tion to deal with heavy time of the Superfrac tray
same diameter in the stripping versus a conventional tray for
section and flash zone (This side flow.
was not necessary in this case, Stagnant liquid pools
as the Suncor ATB bitumen promote multiple types of foul-
stripper is a separate, dedi- ing mechanisms including
cated, reduced diameter thermal cracking (coking),
tower.) solids deposition and polymer-
• Do not rely solely on simula- isation. In some heavily fouling
tor equilibrium stage data to refinery applications, the use of
perform hydraulic design baffle trays is preferred due to
calculations their low residence time and
■ Simulations typically use lack of stagnant zones.9
1-2 theoretical stages to repre- Figure 9 Two-pass Superfrac tray However, baffle trays are noto-

6 PTQ Q2 2016 www.eptq.com


riously low efficiency devices References upgrading optimisation, debottlenecking,
and require high vapour veloc- 1 Kaes G, Refinery Process Modeling, trouble-shooting and projects, as well as
Athens Printing Company, 2008, 68. support to infrastructure projects. Prior to
ities (Cs >0.2 ft/s (0.06 m/s)) to
2 Watkins R N, Petroleum Refinery joining Suncor in 2001, he held positions
drive mass transfer. Typically a
Distillation, Gulf Publishing Co., 1st ed, at Anglo American and Sasol in South
heavy hydrocarbon stripping 1973, 17. Africa. With over 30 years of process and
section has low vapour veloci- 3 White S, Barletta T, Special Report: operational experience in oil refining,
ties (Cs ~0.02-0.10 ft/s Refiners processing heavy crudes can petrochemicals and mineral processing,
(0.006-0.03 m/s)), and a baffle experience crude distillation problems, he graduated in chemical engineering
tray may obtain only a quarter Oil & Gas Journal, 12 Nov 2002, 34-40. from the University of Natal, South Africa,
of the efficiency of a well 4 Hanson D W, et al, Low-capital crude and with an MBA from the University of
designed cross-flow tray. unit revamp increases product yield, Oil Cape Town. He is a registered professional
Furthermore, they are not foul- & Gas Journal, 24 Mar 2003. engineer in Alberta, Canada.
ing-proof, as there have been 5 Nieuwoudt I, Penciak J, Best of Both, Email: mgoulding@suncor.com
Hydrocarbon Engineering, July 2007, 85- Fred J Zhang is a Process Engineering
multiple cases reported of coke
89. Specialist with Suncor Energy Inc. and
formation occurring on the
6 Nieuwoudt I, et al, Revamp and has over 20 years’ process engineering
underside of baffle trays.10 Retune, Hydrocarbon Engineering, July and operational experience in oil refining,
We would suggest that the 2009, 56-60. petrochemical and bitumen upgrading
flow enhancement devices that 7 Remesat D, Improving performance in oil sands. He graduated in petroleum
are part of the Superfrac tray through low-cost modification of tower refining engineering from the University
toolbox would allow for the internals, PTQ, Q3 2010, 37-38. of Petroleum, China, and is a registered
best of both – maximisation of 8 Kister H, Distillation Operation, professional engineer in Alberta, Canada.
tray efficiency while employing McGraw-Hill, Inc., 1990, 167. Email: fzhang@suncor.com
a design that minimises the 9 Kolmetz K, et al, Case studies Michael Krela is a Senior Process
fouling potential resulting from demonstrate guidelines for reducing Designer with Koch-Glitsch Canada and
fouling in distillation columns, Oil & Gas has over 11 years’ experience providing
stagnant liquid zones.
Journal, 23 Aug, 2004, 46-51. detailed solutions for new and revamp
10 Kister H, Distillation Troubleshooting, mass transfer designs in refining and
Acknowledgement
John Wiley & Sons, Inc., 2006, 279. petrochemical plants. He holds a BASc in
We would like to thank Param
chemical engineering from the University
Parameshwaran and Bala
Mike Goulding is Manager, Process of Waterloo, Canada, and is a member of
Subramanyam at Suncor along
Development, for Upgrading with Suncor the Koch-Glitsch Global Refining team.
with the Process and Mechanical Design
Energy Inc. In this role he is responsible for Email: Michael.krela@kochglitsch.com
Team at Koch-Glitsch Canada.

www.eptq.com PTQ Q2 2016 7

You might also like