Stepwise simulation of vacuum transfer

line hydraulics
A stepwise hydraulic calculation determines the pressure profile of a vacuum
transfer line by linking the hydraulic model to process simulation results

Harry Ha Fluor Canada

Matthew Reisdorf Fluor Enterprises
Abdulla Harji Fluor Canada

W
hen designing a vacuum pressure is ensured in the flash zone revenue. From a process design point
transfer line, a robust to allow the unit to reach target oil of view, the pressure drop in the
hydraulic model that vapourisation at an acceptable heater vacuum transfer line should be as
predicts velocity and a corresponding outlet temperature (HOT). A pressure small as possible to maximise the
pressure drop is crucial. A stepwise drop in the vacuum transfer line sets yield of vacuum gas oil. This usually
approach to hydraulic modelling of the heater outlet pressure (HOP), leads to a large transfer line and
vacuum transfer lines increases which in turn determines the HOT by increased heater passes, resulting in
accuracy and enhances the under- vapour-liquid equilibrium. significantly increased capital costs.
standing of two-phase fluid behaviour. The HOT is limited to an acceptable Therefore, it is essential to select the
Vacuum gas oil yield, reliability and value to avoid coke formation inside most cost-effective design, which
operability depend on correct design the heater coils. A high pressure drop meets both process design and
of the vacuum transfer line. in the vacuum transfer line increases mechanical requirements.
With the depletion of global the temperature difference from the The vacuum transfer line is a large,
conventional oil, refinery feedstocks heater outlet to the flash zone. Given a elevated line that routes the vacuum
are becoming heavier and often rely unit feed from the charge heater outlet
on unconventional heavy oil. to the vacuum tower flash zone.
Canadian oil sands-derived bitumen A typical objective Depending on capacity, the line’s
— a heavy, unconventional oil — is diameter can range from 48–84 inches
being increasingly processed in in the design of a inside diameter and its length is
Canadian upgraders to produce typically 40–70 ft. A typical piping
synthetic crude oil as part of the crude
vacuum unit is to layout for a transfer line includes
slate in US refineries. Processing maximise the yield either individual heater pass outlet
unconventional heavy oil like bitumen piping discharges into the main line
is challenging. Unconventional heavy of vacuum gas oil to routed to the vacuum tower, or half
oil is usually unstable and prone to the heater passes discharge into a
coking at high temperature. To improve a refinery’s manifold and the two manifolds
maintain reasonable run lengths, a discharge into the main line. The
temperature limit is applied to the
profitability piping design group should be
heater outlet, which indirectly puts a consulted to establish the preliminary
limit on the heater tube’s inside fixed HOT, an increased pressure drop transfer line routing, including
film temperature. in the vacuum transfer line decreases approximate lengths and allowance
the vacuum gas oil lift in the flash for thermal expansion. Since transfer
Design objectives zone, resulting in a lower yield of lines have a low allowable pressure
A typical objective in the design of a vacuum gas oil. Therefore, the vacuum drop, pressure loses due to fittings
vacuum unit is to maximise the yield transfer line’s hydraulics plays a should be minimised.
of vacuum gas oil to improve a crucial role in achieving the desired The number of parallel heater tube
refinery’s profitability. The vacuum product yields and operational passes is determined by the required
overhead system, column flash zone, reliability. cross-sectional area at the heater outlet
vacuum transfer line and the charge A study1 showed that one extra kPa to accommodate the large volume of
heater have to be optimised as a single added to the total pressure drop of a two-phase flow. At the heater outlet,
system to ensure that design objectives transfer line reduces the gas oil yield there are typically four to eight
are met during unit operation. by about 0.2 vol%. For a refinery with separate heater tube passes from one
Based on the steam and cracked gas a 100 000 bpd throughput, each kPa of or more cells. While cost-effective
loads, the vacuum overhead system is pressure drop in the vacuum transfer heater design favours using fewer
configured so that a low absolute line implies a significant loss of tube passes, the need to stay below

www.digitalrefining.com/article/1000617 REVAMPS 2009 11

 the critical velocity necessitates an gas phase alone. Therefore, many two-phase flow hydraulics.2 Reviews
adequate number of tube passes. transfer lines, designed to run under and model evaluations have been
A limiting factor in minimising the sonic velocity, actually operate at conducted extensively by researchers
vacuum transfer line’s size is the bulk critical velocity, especially near the and developers of hydraulic
HOT. For a state-of-the-art and deep- heater outlet and at the column calculation tools. However, no
cut vacuum unit, the maximum entrance. The potential impacts of universal model has been widely
recommended oil temperature in a critical velocity are not trivial. The accepted. Depending on the criteria
heater is usually 365–415°C to avoid vibration of shock waves could result and the system studied, different users
excessive cracking and coking within in failure of the vacuum transfer line. favour one model over others. This
the heater coils. To reduce cracking Critical velocity can also lead to article will focus on how two-phase
and coking, the HOT must be set low excessive entrainment of liquid flow hydraulics can be simulated,
enough to ensure the film temperature droplets to the wash zone, even with a provided the model is chosen.
does not exceed the maximum well-designed column feed device.1 Following a comparison of
recommended oil temperature. Heater designers attempt to limit the numerous model performances, the
The pressure in the vacuum transfer velocity in the heater coil to reduce Dukler-Taitel model2 with liquid hold-
line keeps decreasing from the HOP vibration problems, and often target a up from the HTFS method3 was
to the pressure of the column flash velocity limit of 80% of critical velocity selected to calculate two-phase flow
zone, which is normally set at 20–30 within the heater coil itself. With hydraulics with the aid of Fluor’s
mmHg absolute for a deep cut. A modern deep-cut columns, avoiding proprietary hydraulic software. The
typical pressure drop in a vacuum critical velocity throughout the Dukler-Taitel model was derived from
transfer line is about 100–150 mmHg vacuum transfer line often becomes dynamic similarity analysis with
to achieve a deep cut point. impractical. To the designer, the task experimental data, which is dependent
Considering the low pressure at on liquid hold-up but
the flash zone, the pressure drop independent of flow regime. By
of the vacuum transfer line is assuming constant slip for two-
quite significant. Corresponding phase flow, the pressure drop
to this pressure change, the through a conduit can be
temperature also changes Static pressure = 3.5 psia Static pressure = 4 psia calculated with adequate
Velocity head = 0.8 psi Velocity head = 0.3 psi
isenthalpically from the HOT to Total = 4.3 psia Total = 4.3 psia accuracy when employing a
the flash zone temperature, as better correlation of liquid hold-
the feed goes through adiabatic up. For details of the method,
flashes in the transfer line. refer to the original paper.2 The
Depending on the total pressure liquid hold-up, defined in the
drop, a temperature difference HTFS 1992 design report,3 was
of 10–20°C can be expected Figure 1 Example of an expander in vacuum transfer line developed from HTFS graphical
between the heater outlet and correlations by optimising an
the flash zone. As a result, the then becomes one of balancing the analytical function against the original
vapourisation, density of fluid, risks of vibration and entrainment graphical curves. The correlation was
volumetric flow and the transport with the reward of a high yield and tested with the HTFS data bank and
properties all change simultaneously low cost. was claimed to give better results than
along the transfer line. Therefore, a the previous HTFS correlations.3
stepwise, equilibrium simulation is Hydraulic models for transfer For compressible fluids, the
necessary to reflect the continuous line calculations volumetric flow of gas changes
changes in a vacuum transfer line. The fluid velocity within the vacuum through a pipe as a result of static
An important consideration in transfer line1 is normally high — pressure variations along the line. The
transfer line design is the two-phase either close to or at critical velocity. It acceleration pressure drop is
critical velocity, which raises concerns is generally believed that a associated with the expansion of the
about vibration in the line, especially homogeneous-phase dispersed flow gas phase as pressure is reduced. It
acoustic-induced vibrations. Field regime is present at design and becomes significant in two-phase
measurements of the vacuum transfer normal operations, while a separated- systems with high mass velocity and
line and calculations using theoretical phase annular flow regime is observed low pressures, both of which prevail
hydraulic models confirm the during turndown cases. in a vacuum transfer line. Therefore,
existence of critical velocity and its For a better understanding of the the acceleration effect cannot be
influence on the pressure profile flow regime and hydraulics within a ignored, and this effect is also
inside a vacuum transfer line. vacuum transfer line, a two-phase calculated in Fluor’s proprietary
This phenomenon is difficult to flow model should be applied. Over software. Acceleration effects are most
predict for the two-phase flow system. the years, many models have been pronounced in tees and expanders,
Two-phase critical velocity is much developed to account for different as the static pressure changes
lower than the sonic velocity of the fluid systems and flow regimes for significantly across these segments.

12 REVAMPS 2009 www.digitalrefining.com/article/1000617

Figure 1 illustrates an example of the amount of vapour, the highest velocity flows into the flash zone through an
expander in a vacuum transfer line, and, thus, the highest pressure drop. inlet device, all of the velocity energy
where the fluid is travelling near Therefore, complete gas-liquid of the fluid is lost to friction, but the
critical velocity. equilibrium is assumed within the static pressure in the pipe at the exit is
According to Bernoulli’s law, both a transfer line in this instance. essentially the same as that in the
fluid’s pressure and velocity column flash zone. The exit loss is due
contribute to the total energy Stepwise calculation of transfer to the dissipation of the discharged jet.
contained in the moving fluid. At high line hydraulics There is no pressure drop at the exit
velocity at the expander inlet, the Ideally, an integrated numerical except for an insignificant pressure
fluid’s velocity head is a significant equation should be used to reflect the loss due to the inlet device (3.0-3.8
component in this sum. As the fluid differential changes in properties mmHg).1 There have been cases when
moves across the gradual expander, along the vacuum transfer line. extra pressure loss was given to the
the amount of energy lost to friction is However, such a solution is difficult to inlet device in calculations of vacuum
relatively low. The velocity is find for such complex systems. As a transfer line hydraulics. This extra
significantly reduced and much of the compromise, the vacuum transfer line pressure drop changes the pressure
velocity head is converted to static is split into several segments so that profile and thus the transfer line’s
pressure. The net result is that the the variation in transport properties is design, which will be addressed in
static pressure rises across the insignificant for each segment. detail in a case study.
expander. This effect has been Consequently, properties are approxi- When the vacuum transfer line is
documented in two-phase flow across mately constant across the segment, associated with vacuum tower design,
a sharp expansion,4 and it would be and the hydraulic models described normally the temperature at the heater
expected to be even more pronounced above can be applied to each segment. outlet and the pressure at the column
across the more gradual expansions in flash zone are predetermined. A
pipe fittings. backward calculation is required for
For a simple check of the vacuum
For a simple check of transfer line hydraulics, step-by-step,
transfer line’s hydraulic model, a
designer should review each point
the vacuum transfer from flash zone to heater outlet to
heater outlet to meet the constraints.
where the velocity drops suddenly — line’s hydraulic model, To catch fluid property changes as the
commonly expanders and tees where calculation moves from one segment
two process streams combine. As the a designer should to another, a process simulation is
fluid passes through the expander, the done by either Pro-II or HYSYS
static pressure should rise at these review each point software to flash the stream
points. If the static pressure does not adiabatically over the expected
rise at these points, the calculation
where the velocity pressure range of the vacuum transfer
warrants further scrutiny to confirm if drops suddenly line. The changes in properties are
the fluid has hit the critical velocity. then tabulated over the full pressure
As the fluid moves through the range of the vacuum transfer line.
transfer line and the pressure drops, The total pressure drop is simply the When the hydraulic calculation goes
some portion of the liquid vapourises, sum of all segments. to the next segment, the properties are
adding to the gas flow. There is a This method is similar to the use of updated from the simulated property
debate on whether or not the gas- a numerical integration method, such table according to the calculated
liquid phase is in equilibrium within as the Simpson method, to solve the current inlet static pressure of that
the transfer line. Some process differential equations. Applying the segment. The calculation is iterated
designers claim that the vacuum Dukler-Taitel model to each segment until the changes of properties for
transfer line should be modelled by and summarising total hydraulic each segment are insignificant
assuming non-equilibrium between changes result in an accurate compared to the previous iteration;
vapour and liquid because of the high simulation of vacuum transfer line the calculation is then converged.
transfer line velocity,5 while other hydraulics. As a rule of thumb, the Another important issue with
designers prefer to assume significant pressure change of each individual respect to the vacuum transfer line is
or complete equilibrium throughout segment should be no more than 10% the pressure discontinuity that
the transfer line. of the inlet static pressure of that occurs within it when the internal
Non-equilibrium models give better segment. Separate segments are velocity reaches the critical velocity
predictions of wash rate to provide suggested for all pipe fittings (elbows, of the fluid. For an ideal gas
sufficient wetting of the wash bed.5 expanders, inlet device, and so on). system, the maximum velocity that
However, for the purposes of transfer Most importantly, expanders should can be achieved is limited by the
line hydraulics, the most conservative be placed in a segment of their own, maximum velocity of a pressure wave
assumption is to consider the fluid in as the static pressure is expected to travelling in the pipe, which is
equilibrium throughout the transfer rise significantly through the equivalent to the sonic velocity of the
line, as this produces the highest expander. When the two-phase fluid gas. As the gas flows through the

www.digitalrefining.com/article/1000617 REVAMPS 2009 13

 vacuum transfer line, its pressure where ρ is the density of gas; p and v Hydraulic calculations are
decreases and its velocity increases. are the pressure and specific volume conducted from the column flash
If the pressure drop through the of gas. zone, segment by segment, marked
pipe is sufficiently large, the gas Assuming ideal gas behaviour, by letters all the way to the heater
velocity exiting the pipe reaches sonic Equation 2 can be simplified as: outlet. The simulated hydraulic profile
velocity. If the pipe outlet pressure is of the vacuum transfer line is
further decreased or the pipe inlet illustrated in Figure 3, with location
pressure is further increased, the VC,G = kRT (3) points referring to the layout
Mw
excess pressure drop occurs beyond presented in Figure 2. The calculated
the pipe exit. This pressure drop is results match the specified process
dissipated in the shock waves and constraints: the operating pressure in
turbulence of the exiting gas. When where k = CP/CV, the ratio of specific the flash zone and the heater outlet
the ideal gas reaches its sonic velocity, heats and Mw = mol wt of gas. temperature. Transport properties
it has reached the maximum mass In this study, we compared the and the vapourisation of fluids within
flow rate that the gas can achieve. previous correlations with the the vacuum transfer line are simulated
For non-ideal gases and two-phase homogeneous model developed by by Pro-II using the Improved-
systems, the maximum mass flow Buthod7 and the theoretical approach Grayson-Streed (IGS) thermodynamic
through a piece of pipe is the critical derived by Kohoutek et al.8 We found package.
flow. An intuitive way of estimating that the calculated critical velocities The study of Laird et al1 showed that
critical velocity for real gases is to from these three methods are the thermodynamic packages have
use the method developed for an comparable and consistent. Therefore, substantial effects on predictions of
ideal gas, but with non-ideal gas the method of Hewitt and Semeria vapourisation at the flash zone
properties and it usually works well. is proposed here because of its condition of a vacuum tower. The IGS
However, two-phase gas liquid simplicity. package gives a vapourisation rate
between those given by the Grayson-
Streed (GS) and BK10 methods and
M was chosen to simulate the system
J
of interest.
K L N
I For a vacuum transfer line with total
O P Q R S pressure drops of 100–150 mmHg, the
GH simulated vapourisation changes can
A
BC D E F be as high as 15 wt% along the
vacuum transfer line. Using the
automated stepwise approach of this
Figure 2 Vacuum transfer line layout work, the property changes are
captured and updated for each
mixtures reach critical flow at a Stepwise method vs conventional segment based on the online
velocity much less than sonic velocity. method calculated static pressure of that
This phenomenon is better depicted Critical flow is a concern in the design segment. The calculated pressure drop
by the separated-phase model. Hewitt or rating of vacuum unit fired heaters, of each segment is less than 10% of the
and Semeria proposed a separated- transfer lines and relief header systems. inlet static pressure of that segment,
phase model to derive the critical Incorporated within Fluor’s proprietary which justifies the assumption that
velocity of two-phase flow (VC,M) from hydraulic software, an automation properties are constant throughout
the critical velocity of gas:6 program has been developed that each individual segment. For
updates the transport properties in comparison, the hydraulic profile
VC,G [x + (1 – x)C] hydraulic models from the stream calculated by the conventional method
VC,M =
:x + (1 – x)D2 C2 (1) properties simulated by Pro-II or (used by many transfer line and heater
HYSYS. With the aid of these programs, designers) is also shown in Figure 3.
where x is the vapour weight fraction, a case was studied for a typical vacuum With the conventional calculation, an
C is the ratio of gas density to liquid transfer line used in modern vacuum extra pressure drop (45 mmHg) is
density, D is the ratio of critical distillation units. A sketch of the assigned to the inlet device as an
velocity in gas to critical velocity in vacuum transfer line’s layout, with a allowance. Contrasted to the large
liquid and VC,G is the critical velocity branch line connecting the mega-line to pressure drop of the inlet device taken
of gas. one of the heater passes at the heater by the conventional method, a small
The critical velocity of gas can be outlet, is shown in Figure 2. The case and reasonable pressure drop (3.8
expressed as (Perry’s Handbook): presented, not specific to any plant, mmHg) is used in this work, which is
demonstrates the methodology and consistent with the values reported by
procedures used by designers to Laird et al.1
VC,G = ρ e uv o
1 up
(2) address key technical issues. As Figure 3 indicates, the simulated
s

14 REVAMPS 2009 www.digitalrefining.com/article/1000617

total pressure drop of this work is
consistent with the results of the
4.0
conventional method. However, the
hydraulic profile of each segment A Stepwise method
3.5
differs substantially between the two
methods. Figure 4 shows the 3.0
calculated velocities of the fluid G
D Conventional

Static pressure, psia

mixture compared to the critical 2.5
method
velocities along the vacuum transfer
2.0
line using the proposed stepwise
method. The current work clearly K
1.5
indicates pressure discontinuities at O
five locations where the pressure 1.0
profile of the conventional method S
0.5
showed no sign of calculated velocity
ever reaching critical velocity. How 0.0
can two calculations come to such Location point
different conclusions? The differences
in this example are due to several
factors, which highlight some of the Figure 3 Hydraulic profile of a simulated vacuum transfer line
key challenges of this calculation:
• The inlet device typically has a demonstrates the correct response of running at critical velocity — typically
very low pressure drop. The current fluid velocity. The effect of pressure at the inlet to the column and near the
work uses 3.8 mmHg in its calculation, change on velocity is clearly shown at heater outlet. A deployment of the
which is a reasonable estimate for a location points C, E, H and L in Figure stepwise simulation method is crucial
modern column inlet device.1 In the 4 as the fluid passes through the for providing an accurate hydraulic
calculation using the conventional expanders. As the pressure rises across profile to support design, risk analysis
method, a much higher pressure drop an expander, a corresponding decrease and review.
(45 mmHg) is assigned to the inlet in velocity is predicted using the
device, which leads to a different stepwise modelling approach. Conclusion
pressure profile at the inlet to the Theoretically, a vacuum transfer line A stepwise hydraulic calculation has
column. Consequently, the vacuum should be designed to remain under been presented that determines the
transfer line is undersized, resulting in critical velocity. However, as a result pressure profile of a vacuum transfer
a potential choke flow at the column of uncertainty in calculating the line by linking the hydraulic model to
inlet in operation and significant critical velocity and the complexity of process simulation results. The
entrainment inside the column flash two-phase flow hydraulics, it is not velocity profile along the transfer line
zone. In the design of a vacuum unusual to see a vacuum transfer line is also reported in relation to the
transfer line, the pressure drop of the
inlet device should be confirmed by
the vendor and no extra pressure drop
should be assigned
• The acceleration effect across an 300
A
B
D
expander is not taken into account in K O P
the conventional calculation. Figure 3 250
shows a monotonic decrease in Critical velocity
pressure throughout the transfer line C S
200
for the conventional calculation, while G
Fluid velocity, ft/s

the stepwise calculation correctly

150
shows a pressure rise across the
expanders E Calculated velocity L
H
• The pressure drop calculated across 100
pipe and fittings differs. The
conventional calculation shows a near- 50
constant velocity across each segment,
as the calculated pressure drops 0
slightly at each segment except for the Location point
inlet device. The stepwise calculation
method reveals the actual pressure
changes of each segment and Figure 4 Calculated stream velocity vs critical velocity through a vacuum transfer line

www.digitalrefining.com/article/1000617 REVAMPS 2009 15

 critical velocity of the fluid, which McGraw-Hill, New York, 1986. Matthew Reisdorf is a Process Manager at
clearly identifies the choke points 3 HTFS Two-phase pressure drop design Fluor Enterprises Inc, Houston, Texas. His
along the line where the critical report, 1 & 2, Design Report 28, July 1992. experience is in downstream refining, including
4 Lottes P A, Expansion losses in two-phase vacuum distillation units. He has a BS degree
velocity is reached. The method offers
flow, Nucl. Sci. Engr., 9, 1961, 26–31. in chemical engineering from Rice University,
a more accurate way to specify the 5 Barletta T, Golden S W, Deep-cut vacuum Houston.
pressure profile and line sizing of any unit design, PTQ, Q4 2005, 91–97. Email: Matthew.J.Reisdorf@fluor.com
two-phase flow system and provides a 6 Hewitt G F, Semeria R, Aspects of two- Abdulla Harji is Executive Director, Process
useful tool for engineering design of phase gas-liquid flow, Heat Exchangers: Design Technology, at Fluor Canada Ltd, Calgary,
the vacuum transfer line. Correct and Theory Sourcebook, edited by Afgan and Alberta, Canada. He has more than 35 years
vacuum transfer line design has been Schundler, McGraw-Hill, 1974, 289. of experience relating to engineering and
7 Buthod P, How to estimate pressure in operations support of refining, bitumen/
shown to enhance profitability,
heaters, Oil and Gas Journal, 1 July 1957. heavy oil upgrading, gas processing and
operability and safety of vacuum 8 Kohoutek J, Zachoval J, Odstrcil M, Stehlik petrochemicals facilities. He has a BSc degree
distillation process units. P, Solving practical industrial problems in two- in chemical engineering from Loughborough
phase multicomponent mixture flow-critical University, UK.
velocity, Heat Transfer Engineering, 22, 2001, Email: Abdulla.Harji@fluor.com
32–40.
References
1 Laird D, Hauser R, Schnepper C, Vacuum
tower design techniques for optimum
Links
Harry Ha is a Process Engineer with Fluor
performance and reliability, NPRA Annual Canada Ltd, Calgary, Alberta, Canada. He More articles from: Fluor
Meeting, 23–25 Mar 2003, San Antonio, has authored more than 20 technical papers More articles from the following
Texas. on fluid dynamics, transport properties and categories:
2 Dukler A E, Taitel Y, Flow pattern transitions characterisation of petroleum for refinery
in gas-liquid systems. Measurements and processes. He has a doctorate in chemical Process Modelling & Simulation
modeling, Advances in Multiphase Flow, 2, engineering from the University of Alberta. Revamps, Shutdowns and Turnarounds
edited by Zuber N, Hewitt G F, Delhaye J M, Email: Harry.Ha@fluor.com

16 REVAMPS 2009 www.digitalrefining.com/article/1000617