Identifying poor heat

exchanger performance
Identifying root causes of poor crude unit debutaniser reboiler performance and low-cost
solutions that help meet exchanger original design conditions. Of primary importance is
the impact factors such as fouling have on the service heat-transfer coefficient

Mark Fernsby and Abe DuPont Natref

Tony Barletta and Steve White Process Consulting Services Inc

H
eat exchangers are designed
with engineering tools that
allow the designer to evaluate
the influence of various parameters Unstabilised
on projected performance and select naphtha
the optimum combination. However,
computer model results are just
calculations and only reflect actual
performance if the model assumptions
correctly predict what actually happens
inside the exchanger. Consequently,
when an exchanger does not perform Gas oil 42 kg/cm2
product steam
per design, the model assumptions
are not correct. Heat exchanger
performance, just like that of any other
refinery process equipment, depends on Stabilised
naphtha
specific equipment design and not on
the ideals of the computer model.
Figure 1 Crude debutaniser reboiler system
Reboiler system
In 2000, National Petroleum Refiners of was operating satisfactorily. The new case, theories included inherently poor
South Africa (PTY) Ltd (Natref, a joint steam reboiler was designed to supply heat transfer when using U-tubes instead
venture of Sasol and Total South Africa) the balance of the debutaniser heat of straight tubes, insufficient shell-side
commissioned a major crude/vacuum requirements during normal operation fluid circulation resulting in excessive
unit expansion. As part of the revamp, a and the total column heat requirements, vapourisation and two-phase flow
new reboiler was installed on the crude so the gas oil reboiler could be taken irregularities in the reboiler return
debutaniser to provide increased duty to offline for maintenance. This larger piping. Another was that the exchanger
stabilise the increased production from reboiler was designed for 42 kg/cm2 simply was not big enough and a new
the preflash and atmospheric crude pressure steam (Figure 1) to replace one larger exchanger was needed. Many tests
columns. Total reboiler heat input using 16 kg/cm2 pressure steam. After were conducted to investigate these and
needed to increase to remove enough of commissioning, the exchanger had a various other hypotheses that could
the C4s to meet the light straight-run calculated service overall heat-transfer explain the poor performance.
(LSR) Rvp specification for gasoline pool coefficient (U) as low as 25% of design. Determination of root causes must
blending. However, the debutaniser Various field tests were done to evaluate begin with a review of the specific
reboiler heat input was not sufficient to potential root causes. The exchanger system and equipment design, and the
meet the light naphtha Rvp specifications was eventually taken out of service and application of fundamental engineering
after startup, particularly when processing inspected to ensure the actual design principles. As engineering tools become
light crude blends. The debutaniser’s conformed to the drawings and assess increasingly complex, sophisticated and
maximum throughput was limited to whether the shell or tube sides were easier to use, such as computational
90% of design when the reboilers were badly fouled. However, only a marginal fluid dynamics (CFD) models, engineers
clean. While in operation, fouling performance improvement was noted have a tendency to rely on these models
resulted in a 3.5% reduction in reboiler when the exchanger was put back into and search for complex causes and
duty per month (equivalent to a reduced service. Even when the reboiler was solutions. This should not be a surprise
maximum throughput of about 750 bpd clean, its U value was less than 25% of given the increasing focus on
every month). At times, this began to design clean performance. Although the engineering tools rather than on
limit the overall crude unit throughput. reboiler was never significantly fouled, understanding fundamental equipment
The revamped debutaniser column its performance was sensitive even to operating principles. Yet, understanding
was designed to operate with two small amounts of fouling compared to these principles remains the single most
reboilers. One used gas oil product. Its the other reboiler. important factor in the design of process
heat input was therefore set by product It is common to speculate on equipment and troubleshooting when
yield. This reboiler, which had not been potential causes when equipment does the equipment does not perform as
mechanically altered during the revamp, not meet expected performance. In this expected.

www.eptq.com PTQ REVAMPS 3

 Outlet
Steam
in

Steam

hl

Condensate
out
Inlet

Condensate
Figure 2 H-Shell exchanger

Reboiler basics Figure 3 Reboiler liquid head

Potential reboiler problems can generally
be separated into three areas: steam-side, heat exchanger models assume the shell- Reboiler design
process-side and exchanger design. side fluid flow is distributed so that the The new steam reboiler was a TEMA H-
Generally, there are only a couple of total exchanger surface area is utilised for shell type designed for a low pressure
steam-side problems, the most common heat transfer. In reality, the specific drop with no vertical baffles (Figure 2).
of which is flooding the exchanger with exchanger design will determine whether H-shell exchangers have two inlet and
condensate, which reduces the the fluid entering the exchanger is two outlet nozzles, with a horizontal
condensing surface area. Potential uniformly distributed or not. In the case baffle separating the inlet and outlet
process-side problems include a high under consideration, the new steam nozzles. A partition baffle in the middle
system pressure drop, excessively low reboiler design UC was 1355 kcal/h of the exchanger essentially splits the
and high circulation rates, and slug flow m2 °C, but actually achieved only 300 shell side into two separate sections,
in the reboiler return piping. kcal/h m2 °C. each taking flow from its own inlet
Thermosiphon reboilers circulate tower Clean exchanger coefficients do not nozzle.
bottoms fluid through the shell side of include fouling resistance. The H-shells can be designed with or
the exchanger based on system exchanger service heat-transfer without vertical baffles. The maximum
hydraulics. The liquid level in the coefficient includes fouling resistance allowable exchanger pressure drop
bottom of the column and the globe on the inside and outside of the tubes. depends on the overall system design.
valve at the inlet of the shell side (if Total resistance to heat transfer is the When the column liquid level above the
present) is used to control circulation. sum of the shell-side film coefficient, centreline of the exchanger is low, the
Too low a circulation rate increases the tube-side film coefficient and the maximum allowable pressure must be
percentage of vapourisation, which fouling resistance. For a crude unit low, as the available head is limited.
reduces the heat-transfer coefficient, and debutaniser steam reboiler, fouling However, the maximum pressure drop is
too high circulation reduces resistance is generally 25–35% of the sometimes intentionally specified very
vapourisation, which can lead to slug total resistance, with shell-side fouling low because of conservatism or design
flow in the return piping. As for potential more common. It is rare to have any guidelines. The consequences are a low
exchanger design errors, many have significant fouling inside the tube when heat-transfer coefficient, large surface
caused low duty. using steam. area and increased likelihood of poor
Exchanger design problems can In practice, inside and outside fouling shell-side flow distribution. The H-shell
reduce the service heat-transfer resistances are lumped together in an exchanger pressure drop must be
coefficient by affecting inside and overall fouling resistance commonly balanced against the resultant heat-
outside film coefficients. Thermal referred to as a fouling factor. The dirty transfer coefficient.
conductivity of the tube is small (or service) heat-transfer coefficient can The reboiler maximum allowable
compared to the film coefficients. be calculated as shown in Eq 2: pressure drop depends on the overall
Therefore, this term can be ignored. system design. Process-side hydraulics
Thus, the “clean” heat-transfer 1 = 1 + R must be carefully evaluated at the design
coefficient (UC) is calculated from UD UC F Eq 2 stage so the exchanger design is not
Equation 1 (Eq 1): compromised. Shell-side fluid circulation
where: depends on the available liquid level,
1 = 1 + 1 UD = heat-transfer coefficient, dirty exchanger pressure drop, density
Uc hi ho Eq 1 UC = heat-transfer coefficient, clean difference between the liquid in and
RF = overall fouling factor mixed phase outlet, and the piping
where: system pressure drop. Often, revamps
UC = heat-transfer coefficient (clean) Fouling begins as soon as an are constrained by the existing vessel
hi = inside film coefficient exchanger is put into service. Crude skirt height, liquid level and reboiler
ho = outside film coefficient unit debutanisers can foul badly on the return nozzle location (Figure 3). Ideally,
shell side from corrosion products and shell-side circulation rates should result
Computer models calculate film water in the feed from the crude unit in reboiler outlet conditions with 25–
coefficients based on the specific overhead receiver. At one point in the 35% vapour in the mixed phase.
exchanger design, fluid properties, model case study, the reboiler UD was only 110 Balancing the exchanger pressure
assumptions and equations. For example, kcal/h m2 °C. drop and heat-transfer coefficient is

www.eptq.com PTQ REVAMPS 5

 problems are not the cause of poor
performance. Moreover, the Natref
FC
system was designed with a steam
desuperheater, and field testing
confirmed the desuperheater was
42 kg/cm2 steam performing per design. (Figure 4).
Tube-side operation depends on the
specific control system and condensate
drain system. In this example, steam
BFW flow was controlled to the reboiler based
TC on required duty. As duty goes up, the
steam rate increases, with the flow-
control valve opening to allow more
flow. As the valve opens, the pressure
downstream of the control valve
increases. Once the valve is fully open
and the downstream pressure is at
PC maximum, no more heat can be added.
As pressure downstream of the flow-
LP
steam
control valve increases, the condensing
LC temperature increases, raising the
exchanger LMTD. The tube-side
condensing pressure can change from a
minimum of approximately 15 kg/cm2
condensate header pressure to a
LC maximum of 39 kg/cm2 steam pressure.
Pressure downstream of the flow-control
valve is a good indicator of exchanger
HP
condensate performance, because the higher the
pressure needed for a given duty, the
lower the heat-transfer coefficient. Even
with the flow-control valve wide open,
Figure 4 Reboiler control system the steam flow rate was low and so was
the exchanger duty.
provide some margin. Often, not Condensate flooding is the other
surprisingly, the result is an exchanger common problem. Condensate flows
Skid bars Inlet nozzle
on bottom on bottom that is designed larger than necessary from the exchanger into the condensate
without baffles. All this increases the drum. As long as the condensate level is
likelihood of shell-side flow irregularities in the external drum, the exchanger is
and poor performance. H-shell reboilers not flooded with condensate. But once
or condensers designed without baffles the external drum is full, the condensate
are prone to lower-than-expected heat- level is inside the exchanger, reducing
transfer coefficients. the surface area. Field tests showed the
Plan view Following the initial startup of the condensate level was in the external
(top) unit, the maximum observed duty was drum and not flooding the exchanger.
much lower than the design. The steam This was also confirmed through
flow-control valve was nearly wide open, neutron baskscatter tests that were
pushing the condensing temperature to conducted on the reboiler.
maximum. The original exchanger data The exchanger bundle was removed
sheet clean heat-transfer coefficient (UC) from service and visually inspected to
Impingement was calculated at 1355 kcal/h m2 °C, with determine if fouling was the problem.
baffle
a design service heat-transfer coefficient Visual inspection and poor performance
UD of 627 kcal/h m2 °C. The highest after cleaning showed that fouling was
average monthly heat-transfer not the root cause of poor performance.
coefficient, UD, achieved was only 391
Skid bars kcal/h m2 °C, with the exchanger Exchanger design
operating at or below 250 kcal/h m2 °C Poor shell-side performance was
Figure 5 Exchanger bundle much of the time. therefore postulated. The exchanger
was a six-metre long U-tube design
critical. Calculating the shell-side Identifying the problem with two horizontal baffles and no
pressure drop accurately requires a When something does not perform vertical baffles. The exchanger design
model that predicts the phase change as according to design, the most likely was evaluated using Heat Transfer
the shell-side fluid flows upward causes should be investigated first. Eq 1 Research Institute’s (HTRI) proprietary
through the reboiler and vapourises. shows that exchanger performance IST model. Using reasonable fouling
Since this calculation is subject to a depends on both shell- and tube-side factors, the calculated performance was
number of inputs and uncertainty heat-transfer coefficients. Since steam much better than actual duty.
inherent in the correlations, it is not reboilers generally have a high inside Review of the exchanger design
unusual for the process engineer to heat-transfer coefficient — except when showed mechanical features that caused
specify a conservatively low maximum there is a large amount of superheat or poor shell-side fluid flow distribution
allowable pressure drop. Moreover, the condensate floods the tubes, effectively through the bundle. Poor flow patterns
exchanger design engineer will also reducing surface area — tube-side can cause shell-side fluid to bypass

6 PTQ REVAMPS www.eptq.com

portions of tube surface. The inlet nozzle
had an impingement plate, which
restricted flow into the tube bundle.
Furthermore, Figures 5 and 6 show the
exchanger was designed with skid bars
running its complete length. The reduced
inlet area and skid bar design were
causing a significant portion of the feed
to actually bypass much of the exchanger
surface area (Figure 6). The low calculated
coefficient was a result of a significant
bypass exacerbated by a low bundle
pressure drop design (no vertical baffles).

Solution
Many solutions were considered
including a new larger exchanger shell Figure 6 Original bundle
and bundle, and new reboiler return
piping, both of which were expensive
options. Furthermore, these options did Double Full-diameter
not address the root cause and may not segmental baffle
baffle
have worked. Instead, a new exchanger
bundle was designed and installed in
the same shell.
A new bundle was designed with the
intent to improve the flow distribution
to fully utilise the tube surface area
(Figures 7 and 8) and maximise the tube
outside coefficient ho. Vertical double Double
segmental baffles were installed to Full-diameter segmental
baffle baffle
improve the flow distribution. The baffle
Plan view
design was optimised to ensure shell- (top)
side fluid was forced through the entire
bundle. Thermosiphon hydraulics were
checked to ensure the higher pressure Figure 7 Bundle modification
drop would not reduce circulation and
cause problems with high percentage
vapourisation. The skid bars were
redesigned and impingement plates
eliminated. In addition, full-diameter
vertical baffles were installed between
the inlet nozzles to help with flow
distribution. After the new bundle was
installed, the service heat-transfer
coefficient improved from 250 kcal/h m2
°C or less to 950 kcal/h m2 °C (Figure 9).
The debutaniser was now able to process
light crudes while meeting LSR product
RVP specification.

The authors would like to acknowledge the

support of the Natref personnel who were
involved in this project, particularly the Figure 8 New bundle
CDU operations division.

30
Abe du Pont is the manager for process Installation GJ/hr C11068 duty 1400
Heat transfer coefficent U,

engineering at Natref. Du Pont obtained 25 of new kcal/h m2 CU 1200

a BSc Eng (Chem) at the University of tube bundle
Kcal/h. m2. °C

Pretoria in 1984. 20 1000

Duty, GJ/hr

Mark Fernsby is a process engineer 800

at the Natref oil refinery in Sasolburg, 15
South Africa. Fernsby has BSc and MSc 600
10
degrees in chemical engineering from the 400
University of Cape Town. 5
200
Tony Barletta is a chemical engineer with
Process Consulting Services in Houston, 0 0
Texas, USA. 0606 006 006 006 006 06 006 006 006 006 006 006
20
20 2 2 2 2 20 2 2 2 2 2 2
Email: [email protected] u g u g e p ep ep e p ep p p p p p p
A A S S S S S Se Se Se Se Se Se
28 31 2 4 6 9 11 13 15 18 21 23 26
Steve White is a chemical engineer with
Process Consulting Services in Houston,
Texas, USA. Email: [email protected] Figure 9 Improvement in service heat-transfer coefficient after new bundle installation

www.eptq.com PTQ REVAMPS 7