Importance of testing for vacuum ejectors
in refinery service
Efforts should be made to identify and avoid errors in the specification, engineering,
and manufacturing of vacuum system equipment before they manifest at start-up
V
acuum systems are critically
important to the performance
of refinery crude vacuum distillation units. Vacuum tower flash
zone pressure is a result of vacuum
system suction pressure plus pressure drop through the overhead
vapour line and column internals.
To maximise recovery of gasoil
from vacuum residue, the flash zone
should operate at the lowest possible
pressure without exceeding tower
capacity. Unfortunately, many vacuum towers operate above their
design or expected pressure, resulting in lower vacuum gasoil yields
and reduced profitability.
Many papers and articles in the
technical literature discuss the performance and troubleshooting of
vacuum ejector systems in refinery
service.1,2. This article focuses on
the importance of shop testing vacuum ejectors to ensure they meet
their design parameters of pressure
versus capacity and consequently
prevent significant economic losses
that can result from their underperformance. An additional benefit
is the ejector performance curves
derived from actual testing are paramount to any future troubleshooting,
optimisation, and revamping of vacuum systems in operating units.
Crude unit vacuum systems
Multi-stage steam jet vacuum ejector systems are almost universally
used to produce vacuum in refinery crude distillation units. They
are particularly well suited to the
large vapour volumetric rates present and high compression ratios
normally required. The systems are
arranged in two to four stages, with
each stage consisting of an ejector
www.digitalrefining.com
Vacuum resid yeild (LV% on crude)
EDWARD HARTMAN and TONY BARLETTA Process Consulting Services, Inc.
LAURENT SOLLIEC and PETER TREFZER GEA Wiegand GmbH
40%
39%
38%
37%
36%
35%
34%
33%
32%
Unit 1
31%
30%
20
22
24
26
28
30
32
34
Flash zone pressure, mmHg absolute
Unit 2
36
38
40
Figure 1 Flash zone pressure impact on vacuum residue yield
and discharge surface condenser.
Depending on unit capacity and
desired flexibility, each stage may
have ejectors or surface condensers in parallel. Steam works as the
motive fluid providing the energy
for compression. Load to each stage
consists of non-condensable gas,
condensable hydrocarbons vapour,
and water vapour. Surface condensers minimise the quantity of
condensable vapours and cool the
non-condensable gas mixture flowing to the next ejector stage and leaving the system. This reduction in
load results in smaller ejector size in
the intermediate and final stages, as
well as lower overall energy usage.
Discussed in further detail under
vacuum ejector fundamentals, ejector suction pressure is a function
of its load and performance curve,
as long as its discharge pressure
is below its maximum discharge
pressure (MDP). In a multi-stage
crude vacuum unit system, overall suction pressure is a function of
vapour overhead load from the vac-
uum tower and first-stage ejector
performance curve. This holds true
provided the first-stage condenser,
as well as ejectors, condensers, and
interconnecting piping in subsequent
stages do not cause the first-stage
MDP to be exceeded.
In some cases, high vacuum system suction pressure, and consequently high tower pressure, is
caused by process loads pushing
the ejectors out on their curves.
However,
Process
Consulting
Services (PCS) has encountered several instances of ejector design errors
leading to tower pressures up to 10
mmHgA above design right from
start-up. When vacuum system fails
to deliver from day one, system
performance can further degrade
rapidly with normal exchanger fouling and the initial 10 mmHgA miss
can increase to 20 mmHgA or more
above design suction pressure.
The impact of flash zone pressure
on refinery profitability is magnified in units processing heavy
crudes. Figure 1 shows vacuum res-
PTQ Q2 2022 01
Motive nozzle
Diffuser throat
Diffuser
Discharge
Motive
steam
Pd
m=mm + ms
Pm, m m
Suction chamber
Ps, m s
Suction load
P = Absolute pressure
m = Mass flow rate
Figure 2 Steam jet vacuum ejector diagram
Suction pressure (mmHg absolute)
idue yields for two different vacuum columns over a range of flash
zone pressures. Unit 1 runs a blend
of heavy Canadian crudes, while
Unit 2 processes a blend of heavy
Venezuelan and light US crudes.
As pressure increases from 20 to
40 mmHgA, Unit 1 residue yield
increases by about 3 LV% on crude
and Unit 2 residue yield increases
by about 4 LV% on crude. For a
crude capacity of 100 MBpd, this
corresponds to 3 and 4 MBpd of
incremental vacuum residue, respectively. Considering a vacuum gasoil to residue downgrade penalty
of $15/Bbl, the higher flash zone
pressure results in profit losses of
$16MM/yr and $21MM/yr.
Troubleshooting and diagnosis of
crude vacuum systems underperformance can be difficult and time
consuming. The process of determining ejector loads in an operating unit
is full of uncertainties. It requires
good field data, laboratory analysis, and flow measurements that are
often unavailable. Furthermore, it
relies heavily on process simulation
models that are only as good as the
simulation inputs. For units that
have been in operation for more than
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2000 3000 4000
5000 6000 7000
8000
a few months, additional uncertainties can further complicate efforts
to narrow the root cause(s) of high
vacuum column operating pressure.
These include, but are not limited
to, condenser fouling, ejector motive
steam nozzle erosion, and tower
internals damage that increases system load.
Therefore, every effort should be
made to identify and avoid errors
in the specification, engineering,
and manufacturing of vacuum system equipment before they manifest
themselves at start-up. In vacuum
surface condensers, undersized shell
nozzles, over-optimistic heat transfer coefficients, and vapour internal
bypass are common problems. In
vacuum ejectors, motive steam nozzle size and position are common
culprits. In addition to generating
accurate performance curves, shop
testing vacuum ejectors provides a
unique and valuable opportunity to
prevent such problems.
Steam jet vacuum ejector
fundamentals
Steam jet vacuum ejectors are essentially compressors with no moving parts. Figure 2 is a diagram of a
9000 10000 11000 12000 13000 14000 15000
Equivalent water vapour load @ 70˚F (lb/hr)
Figure 3 Suction pressure vs load performance curve
02 PTQ Q2 2022
steam jet vacuum ejector that establishes nomenclature and symbology
used throughout this article.
The basic operating principle of
a steam jet ejector is momentum
transfer. The motive nozzle converts
steam pressure energy into velocity energy, resulting in a supersonic
velocity jet of steam that entrains
the vapour and gas mixture from
the suction chamber. The resulting
mixture of motive steam and suction load enters the diffuser where
velocity energy is converted back
to pressure. At the diffuser throat,
a pressure discontinuity or shock
wave occurs which is responsible for
an important property of steam jet
ejectors – suction pressure is independent of discharge pressure up
to a certain limit, and is only influenced by the amount of suction load.
Figure 3 presents the performance
curve of a large first-stage ejector,
showing the relationship between
suction pressure and load.
As mentioned above, suction pressure is independent of discharge
pressure only up to a certain point.
When discharge pressure becomes
too high, the shock wave in the
ejector throat can become unstable,
leading to ‘broken’ ejector operation
that produces high and sometimes
fluctuating suction pressure. When
an ejector is operating in a broken
state, the suction pressure is unpredictable and depends on both suction load and discharge pressure.
The discharge pressure that breaks
the ejector is known as maximum
discharge pressure, or MDP. MDP
is commonly represented as a single number for a given ejector, but
it actually varies with suction load.
Figure 4 shows a combined plot of
suction pressure and MDP curves
versus suction load for an actual
first-stage ejector. Note that the axes
on this figure are reversed from the
typical pressure versus load capacity
curve, with X-axis indicating pressure and Y-axis representing load.
This arrangement makes it easier
to combine the individual capacity
curves of a multi-stage system in the
same plot.
Although ejectors can handle a
wide variety of suction gasses with
varying temperature and molecular
weight, ejector curves and ejector
www.digitalrefining.com
Equivalent water vapour load @ 300˚F (lb/hr)
15000
14000
13000
12000
11000
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
make up those ratios and with the
proper correction factors.
Test stand setup and limitations
Suction
0
10
20
30
40
MDP
50
60
70
80
90
100
110
120
Pressure (mmHg absolute)
Figure 4 Suction pressure and MDP vs load performance curves
testing are based on the principle of
equivalent load. Any combination
of suction vapours can be converted
to equivalent water vapour (EWV)
or dry air equivalent (DAE) load
by simple corrections for molecular weight and temperature. Using
equivalent load makes it easy to convert and plot any given suction load
onto the equivalent load curve to
check performance. Equivalent load
can also be represented at different
reference temperatures. This process is described in detail in industry standards.3,4 All loads discussed
herewith are on an EWV basis.
Entrainment ratio:
.
ω = mm
.
ms
Ejectors are scalable devices in
terms of both physical size and
non-dimensional
parameters.
Similitude laws allow ejectors to
be scaled by applying appropriate
ratios to all critical geometric parameters, including throat area, motive
steam nozzle throat area, and motive
steam nozzle distance to throat,
among others. A family of ejectors
sharing identical geometric ratios
also shares all combinations of Cr,
Er, and ω. The entrainment ratio
ω is not exactly constant because
it is affected by friction through
the throat – an ejector with a larger
throat always performs better than
a similar scaled down ejector with
a smaller throat. Therefore, for a
defined ejector geometry, testing at
any set of fixed Cr, Er, and ω is valid,
regardless of the absolute values that
Ejector similitude and scaling
Ejectors are commonly defined by
the following three non-dimensional
parameters:
Compression ratio: Cr = Pd
PS
Expansion ratio:
Er = Pm
Ps
Liquid ring vacuum
pump or atmosphere
TI
Motive
steam
Air
Superheater
(optional)
Separator
Condensate
drain
Load
steam
TI
Direct contact
condenser
PI
HEI
nozzle
Separator
Condensate drain
Figure 5 Test stand configuration
03 PTQ Q2 2022
Cooling
water supply
PI
PI
Cooling water
return
Previously
referenced
industry standards cover many testing
related topics in detail. The standards are written to cover testing of
ejectors for a wide range of services.
Ejectors designed for refinery vacuum systems normally bring additional challenges due to size and
utilities consumption. Their testing
should include a range of suction
loads, with suction pressure and
MDP recorded at each load point.
Although formal testing guidelines
allow testing only three points, adding additional test points is trivial
after going through the effort of setting up the ejector on the test stand.
Depending on unit process design
and modes of operation, it may also
be advisable to check unusual loads
corresponding to expected process
swings. Ejectors that may operate
during periods of very low load
should be tested all the way down to
zero load to ensure suction pressure
stability. In addition, the zero-load
suction and discharge pressures are
essential at start-up of the vacuum
column for comparison with in-situ
measurements.
Refinery ejectors are typically tested
with steam as load rather than air.
Figure 5 shows a representative test
stand setup. This test setup allows
for control of motive steam pressure, suction load, and ejector discharge pressure. Motive steam is let
down to the test pressure through
a pressure control valve. If the vacuum system design conditions specify superheated motive steam, then
superheat may be added by an electric superheater or motive steam pressure can be adjusted to account for
the lack of superheat.4 Suction load
is metered by adjusting steam pressure upstream of a Heat Exchange
Institute (HEI) nozzle under critical flow. Ejector discharge is routed
to a direct contact condenser where
motive and load steam are condensed by mixing with cooling water.
Condenser pressure is maintained
by a downstream vacuum producing equipment (liquid ring vacuum
pump) and it can be adjusted by an
air intake control valve.
www.digitalrefining.com
Equation 1 can be used for motive
steam flow calculations, as well as
suction load flow calculations using
steam through an HEI nozzle under
critical flow. The unit conversion
factor U (see Table 1) combines the
nozzle flow coefficient and discharge
function at supercritical flow:
Values of unit conversion factor U
Unit conversion factor U
867.9
0.5804
ṁ
lb/hr
kg/hr
Manufacturer’s test stands are limited in physical size, steam boiler
capacity, steam pressure, and cooling water heat removal capability.
Given size limitations, most large
first-stage ejectors and many larger
second-stage ejectors do not physically fit in the existing test facilities.
Additionally, large ejectors that consume thousands of pounds per hour
of motive steam or ejectors designed
for high pressure motive steam
may exceed test stand capabilities.
The ability to add superheat to the
motive steam can also pose a challenge. Therefore, test methods based
on model ejectors, scaled process
conditions or a combination thereof
may be required. General guidance
is to use the simplest test possible,
while resorting to rigorous scaling
factors and accurate empirical correction factors when needed.
A. Direct test
Testing is most straightforward
for small ejectors that fit on the test
stand and require motive and load
steam that are within steam generation capacity both in terms of flow
rate and pressure. In these cases,
motive steam is set at design, and
the measured suction pressures, discharge pressures, and equivalent
loads can be plotted directly on the
final ejector curves without any correction factors. Due to its simplicity,
direct test is the preferred method
whenever possible.
B. Scale test
For ejectors that physically fit on
the test stand but that require more
motive steam flow or higher motive
steam pressure than available, the
actual ejector can be tested by scaling the process conditions. Scaled
test results are then translated to the
final ejector curves by using Cr, Er,
www.digitalrefining.com
P
psia
bar
ρ
lb/ft3
Kg/m3
Note: P and ρ are for the motive steam upstream of the nozzle.
Table 1
m.steam,critical =U * Dn2 * √P * ρ
[1]
Testing methods
Dn
in
mm
and ω non-dimensional parameters.
C. Superheated or saturated steam
test
Ejector performance is based on
mass flow. For typical refinery levels
of superheat (100-300°F), the extra
energy due to superheat generally
has a negligible impact on ejector
performance. However, superheat
is important because of its effect
on motive steam density, which
changes the mass flow of steam
through the critical flow motive
nozzle. Steam with more superheat has lower density, so, for the
same pressure, mass flow of superheated steam through a fixed orifice
is less than it would be for saturated
steam. In terms of testing an ejector
designed for superheated steam,
options are:
If the test stand can supply superheated
steam For direct and model testing,
superheat the steam to achieve the
exact motive steam design pressure
and temperature. For scale testing,
superheat the steam such that both
expansion ratio Er and steam mass
flow rate scale correctly. The ratio
of superheated steam density to
saturated steam density should be
the same in both cases.
If the test stand cannot supply superheated steam For model testing, the
model’s motive steam nozzle throat
area can be reduced to account
for the higher density of the saturated steam used on the test stand.
For direct or scale testing, motive
steam pressure can be adjusted to
get the correct motive mass flow.
Superheated steam has a lower density than saturated steam of the same
pressure, so density can be matched
by lowering test steam pressure
until motive flow rate at the lower
pressure matches the superheated
motive flow rate at design pressure.
This calculation is straightforward
with steam tables and Equation 1.
Adjusting steam pressure to
account for superheat impacts the
expansion ratio Er, such that the test
points Cr, Er, and ω are no longer
equal to those for the ejector curve
points. Correcting the test point for
the difference in Er requires adjusting the entrainment ratio ω by an
empirical factor which, similarly to
the efficiency factor, is also based on
manufacturer’s experience.
D. Model test
Ejectors are scalable devices, so performance of a geometrically scaled
ejector can be accurately ratioed up
to predict performance of the fullsize ejector. For large ejectors, it is
common practice for the manufacturer to use a scaled model of the
actual jet to optimise steam nozzle
position and complete the factory
performance tests.
Ejector efficiency is significantly
impacted by wall friction inside the
jet. As ejector size increases, ejector
throat area increases more quickly
than wall surface area, so that larger
ejectors are more efficient because
less energy is lost to wall friction.
Therefore, performance of a fullsize ejector always exceeds that of a
smaller scaled version of the same
ejector. In terms of ejector testing
and certification, this efficiency difference can be accounted for by an
empirical efficiency factor to produce
accurate ejector capacity curves, or
the curves can be used as tested and
the increased efficiency thought of as
‘bonus’ or ‘safety margin’. The technique explained in testing method
B (scaling of the process conditions)
can also be applied to a model ejector if required by the test stand
capabilities.
Crude vacuum unit ejectors testing
A revamp designed by PCS of a
major US Gulf Coast crude vacuum
unit illustrates the application of the
PTQ Q2 2022 04
1st stage model test results and translation to full-size ejector
Model test results
Pm,test
Ps,test
Pd,test
psia
mmHgA
mmHgA
194.4
11.1
90.0
m·s,test
lb/hr
89
m·s
lb/hr
9,634
Full-size ejector results
Pm
Ps
Pd
psia
mmHgA mmHgA
194.4
11.1
90.0
Table 2
3rd stage ejector scaled test results and translation to design conditions
m·s,test
lb/hr
325
Scaled test results
Pm,test
Ps,test
Pd,test
m·s
psia
mmHgA
mmHgA
lb/hr
28.7
27.8
123.8
1,408
Design test results
Pm
Ps
psia
mmHgA
128.8
111.0
Pd
mmHgA
495.0
Table 3
testing methods. Unit charge consists
of atmospheric residue from a blend
of heavy Venezuelan and light US
crudes. In order to maximise refinery
profitability, one of the main revamp
objectives is to improve heavy vacuum gasoil (HVGO) cutpoint to over
1050°F, thus minimising vacuum
residue yield. Therefore, required
vacuum unit design conditions are
stringent: heater outlet temperature in the 780°F range, stripping
steam rate in excess of 8.0 lb/bl of
residue and 30 mmHgA vacuum
tower flash zone pressure. The fourstage existing vacuum system was
revamped with all new ejectors, as
well as new first- and second-stage
intercondensers.
1st stage ejectors – model test with
adjustment for steam superheat
The first-stage ejectors cannot be
directly tested due to their sheer size,
requiring a model test. The model
ejector is scaled down based on the
similitude laws, resulting in a scale
factor of 100 to 1 ratio. As previously mentioned, the model ejector
is less efficient than the actual ejector due to wall friction effects. This
efficiency difference is accounted for
on the translation of each test point
to the full-sized ejector curves by an
empirical factor. Although the friction correction factor is correlated to
Reynolds number, in practice GEA
relies on an extensive database of
test results, compiled over decades,
to determine this empirical factor.
Design motive steam conditions
include superheat, which cannot be delivered on the test stand.
05 PTQ Q2 2022
Therefore, the model ejector is fitted with a motive nozzle throat area
sized to account for the higher density of the saturated steam used on
the test stand, such that the correct
motive steam mass rate is delivered. Test results from one of the
measuring points and its respective
translation to the full-sized ejector is
exemplified in Table 2.
3rd stage ejectors – scale test with
adjustment for steam superheat
The third-stage ejectors physically
fit in the test stand but their motive
steam rate exceeds test stand capacity. Therefore, a scale test is performed with a scale factor of 4 to
1 ratio. Again, the design motive
steam conditions include superheat, which cannot be delivered on
the test stand. In order to achieve
the same motive steam mass rate
through the motive nozzle, a superheat temperature correction factor
calculated from steam properties is
applied to the motive pressure in
conjunction with the scale factor.
As previously mentioned, adjusting steam pressure to account for
superheat results in a different
expansion ratio Er between design
and test conditions, while the compression ratio Cr remains constant.
Therefore, the entrainment ratio ω
is adjusted by an empirical factor
based on GEA’s experience. Test
results from one of the third- stage
test measuring points and its respective translation to the design conditions is exemplified in Table 3.
All ejector elements in this system
were tested by one of the methods
illustrated above. This acceptance
testing allowed GEA to demonstrate
the expected design performance of
each ejector, as well as to generate
accurate suction and MDP performance curves. Overall vacuum system performance and its impact on
post-revamp unit operations will be
discussed in a future article.
Conclusion
Acceptance or shop testing of vacuum ejectors is used to generate an
accurate set of ejector performance
curves. More importantly, it provides an opportunity to prevent ejector underperformance by identifying
and correcting design or manufacturing errors. Given the significant
profit loss that can result from a
‘broken’ ejector in a crude vacuum
distillation unit, shop testing of new
ejectors should be mandatory. This is
not a topic to cut corners on cost.
References
1 Lieberman N, Cardoso R, Troubleshoot
operation of a steam ejector vacuum system,
Hydrocarbon Processing, Feb 2016, 59-64.
2 Cantley G, et al, Maximise VGO yield, PTQ
Revamps & Operations, 2005, 22-25.
3 Heat Exchange Institute, Inc., Standards for
Steam Jet Vacuum Systems.
4 German Standard DIN 28 430, Rules for the
Measurement of Steam Jet Vacuum Pumps
and Steam Jet Compressors.
Edward Hartman is a process engineer with
Process Consulting Services Inc, in Houston,
TX, USA. He has 35 years of experience in
studies and design packages of refinery units,
specialising in distillation equipment design.
Tony Barletta is Vice President with Process
Consulting Services. Inc. He has over 34 years
of experience with refinery revamps and
process design for heavy oil units.
Laurent Solliec is an R&D Manager at GEA
Wiegand GmbH, Germany. He specialises
in customer oriented solutions and ejector
technology, and heads the test bench facility.
He holds a PhD in fluid dynamics engineering.
Peter Trefzer is Product Manager for vacuum
systems and ejectors at GEA Wiegand. He
specialises in multi-stage ejector applications
and holds a degree in mechanical engineering.
LINKS
More articles from: Process Consulting
Services
More articles from the following
categories:
Crude and Vacuum Units
www.digitalrefining.com