Technical Paper
TS001-0812-3
Crude Oil Sampling
Customer Satisfaction
by Mark A. Jiskoot, Managing Director, Jiskoot Ltd
Custody Transfer Measurement Symposium (Saudi Aramco), Dharan, Saudi Arabia
Introduction
Loss control specialists wish to be sure of both
the quantity and quality of the oil received on
board ship and at point of discharge (receipt
port). Confidence in the measurement accuracy
reduces risk and improves the demand price of
the batch.The quality measurement accuracy is
a function of all the steps involved in sampling,
failure to maintain the representivity of the
sample at any point i.e. from pipeline to analysis
invalidates the sample by introducing random
errors.
The API 8.2 was originally published in 1983, the
IP 6.2 in 1986 and the ISO 3171 in 1988; in
reality the ISO work preceded the IP standard
though it was published later.
Jiskoot is unique in that we can provide the
design and all the components/sub-systems
required to maintain sample representivity
through to analysis. Jiskoot also provides site
installation, training and proving.
Installation of the best equipment available will
not ensure accuracy unless great care is taken
over the complete system design.There is a
consistent and growing body of evidence of
inadequate compromise that leads to avoidable
errors. System design can also optimise the
equipment selection to suit the operational
characteristics and physical placement of the
equipment.
This paper will outline some of the problems
and how they can be addressed, for both
shipment and receipt terminals.
Until this year, in practice, the IP 6.2 was the
most recent standard and included results from
the many trials that went on in the early and
mid 80s. Recently considerable testing has been
performed both in the US, EC and elsewhere on
samplers, leading to revisions to the standards,
which means that systems formerly complying
to the revised API may no longer meet the ISO
or IP standards.There are several areas of
debate that are of interest and these stem from
fundamental system design issues.
Many of us have outlined that there are four
steps required to achieve a representative
sample, there are in fact six:
1.
Installation position relative to the custody
transfer position
2.
Homogeneity of Pipeline Contents
3.
Extraction of a flow proportional and
representative sample
4.
Sample handling and mixing
5.
Laboratory analysis
6.
Prove all of the above and show ongoing
performance
Crude Oil Sampling Customer Satisfaction
Summary
Sampling System
Line Fill V
S.B.M
Ship
Shore
Unsampled or
LOAD Batch -V
Change receiver before end of load
UNLOAD Sample V before load arrives
Location of the sampling system relative to the
custody transfer
The location of the sampling system relative to the
custody transfer position can be critical to the
accuracy of the sampling operation.The volume of
the pipeline between the custody transfer position
and the sample position is often called the Line Fill
Volume (Linefill).This can be relevant if the Linefill
is substantial relative to the batch volume and also
depending on whether the properties of the Linefill
are expected to differ from the bulk batch
properties. In a discharge operation, substantial
Linefill (such as between an SBM and a land
mounted sampling system) can be used to dump
the worst parts of a batch in the hope that they
may be allocated to the next ship discharging.The
end of any cargo is likely to contain more sediment,
heavy fractions and water than the bulk of the
discharge.
LineFill Must be Handled Correctly
One of the major problems in this situation is that
the analysis results are normally required before
the vessel sails, and cannot wait for another ship to
displace the line so the sampler can sample what
has passed the custody transfer point.
Two approaches may be taken to resolve the line fill
issue:
frequently sampled at a rate proportional to that
which is determined by the next batch size
expected to load.This becomes a problem if the
planned next batch volume is changed for any
reason.
Of similar importance to the Linefill issue is the
Dead Volumewithin the actual sampling system,
this is the volume of sample between the point at
which the sample is isolated from the pipeline and
the sample receiver.
This becomes a design issue in some systems for
operational reasons.Where, for example, an in-line
sampler is installed in a location inconvenient for
operators to retrieve the samples, or where the
sample device is remote from the receivers. In
either case, consideration should be given to the
Dead Volume (of grabs that have been extracted)
that does not end up in the receiver. Some
designers will sample at a high rate at the end of
the batch to displace the Dead Volume to the
receiver; and likewise at the start of the next batch
displace the Dead Volume to slops before starting
to collect the real sample. Dead Volume problems
are made worse by the fact that a line that does
not consistently slope down towards the receiver
can form a U-trap which will retain water.
1.
Ignore the problem.
Pipeline Homogeneity
2.
Sample the Linefill into a separate receiver in an
attempt to generate enough sample to analyse;
this normally requires that the sampling system
operate at a very high rate over a relatively short
duration.The minimum volume required for
analysis will constrain this possibility.
This is a greatly misunderstood and misapplied area
of sampling, there tend to be great problems in this
area, both when loading and discharging cargo.
Loading situations can be better controlled because
the flowrates rangeability can be less and the water
content of the oil can be reduced by correct shore
procedures such as water drawing the tanks. In
discharge operations the flowrate is determined
only by the ship (and any backpressure caused by
shorelines) and this is subject to other operational
problems such as tank line ups, crude oil washing
and stripping.The flow range for loading rarely
tends to exceed 5:1 whereas for discharge is rarely
better than 30:1! It is our experience that some
type of mixing is required for almost all
installations, both load and discharge, be it by
choosing a sampling position after a series of
In a loading situation, the volume between the
sampling system and the custody transfer point is
Dead Volumes of Fast Loop Systems are Less!
elbows, bends, valves or using a mixer.
When considering the type or adequacy of pipeline
mixing the designer should study both the REAL
rangeability and the sample extractor design. It is a
major oversight to state that the average flowrate is
the design point for the system because the
minimum flowrate only happens for a short time or
for a small percentage of the batch.These points
are precisely where the worst problems occur.
Other than the issue of mechanical strength of the
sample extractor, it is important to remember that
the required dispersion quality is a function of the
extraction device, i.e. the size of the opening.
of the take-off is irrelevant provided wall effects are
avoided.The orientation of the take-off i.e. vertical
or horizontal is irrelevant to a fast-loop sampling
system as there in no potential for gravitational
separation to occur within the take-off probe if
correctly designed, though it is critical to in-line
extractors.
Extraction of a flow proportional and
representative sample
This involves several major considerations, the most
obvious is the issue of flow proportionality and
most designers have little problem achieving this
within their control systems.
In principle the dispersion quality of the oil is a
function of the turbulence induced by piping
elements or mixing device. If the extraction
mechanism was capable of taking a slice of the
whole cross section, then no mixing would be
required! Following this argument logically, the
One overlooked issue within fiscal metering
systems is particularly evident when sampling at
high rates; all flow computers work on a time based
cycle and unless the sample grab is interrupt-driven
the intergrab time/volume can be found to vary
substantially. i.e. if the computer cycle time is 1
second, frequently the system will only be able to
Small sample probe entries can only cope with small water droplets
Bluff Body Effects
Large sample probe entries can cope with large water globules
larger the size of the opening to the extractor, the
less sensitive it will be to dispersion quality.
In considering mixing devices the designer should
have several points on his list:
What is the dispersion quality required by the
extraction device?
Where should the extractor be placed in
regard to the mixing elements?
What is the pressure drop caused by the
mixing device, or the running costs?
What utilities are required?
How maintainable is the mixer?
What is the real rangeability of the device?
One point often overlooked in the installation of
extraction devices is a misinterpretation of the
standards; the take-off point and the orientation of
the extractor. If the pipeline is adequately mixed for
the extraction device, the cross-sectional position
initiate samples in increments of one second! 30
grabs/min is one every 2 seconds and 31 grabs/min
is 30 at 2 seconds plus an extra grab on the last
cycle!
Regarding the extraction of a representative
sample, a body of work is now in progress with
regards to the ability of sampling higher water
concentrations, between 10 and 100 % (the NEL
Hl-WATER project). All vendors would
acknowledge that extracting a sample from a
homogeneous fluid is easy, however, extracting one
from a poorly distributed and dispersed medium
such as crude can be problematical. It is best to
ensure that the sampling extractor draws its sample
from a stream that is as well dispersed as is
possible, this will reduce the sensitivity of the
sample to any issues of gravitational separation (i.e.
streamline patterns) as well as increasing the overall
probability that a good sample may be extracted. In
short, what this means, is that the sample
extraction mechanism should be suited to the
medium which it is sampling, fast capture action,
less disturbance to streamlines (by the use of a
pitot style entry) and large opening to the
extraction device will ensure better representivity
in pipelines with poorer mixture quality.
Sample Handling and Mixing
Frequently little consideration is given to the
sample handling and mixing process by the system
designer.The handling, mixing and analysis
requirements should be some of the first points to
be considered!
Many installations are completed by project
departments who then hand them over to
operations only to find that a whole new section of
engineering and logistics are undertaken because all
the steps were not adequately consideredIt would
be most unusual for any group to undertake more
operational responsibility without comprehending
the need, and if possible ensuring that the operation
fitted with their practices.
In designing a sampling system the type of receivers
are a key issue and how they are handled, mixed,
subsampled and samples stored is critical.The type
of receivers used depends greatly on the type of
operation. Facilities that transport a variety of
crudes will have different demands for interbatch
contamination than those shipping out a single and
consistent crude type.When sampling for example
for heavy metals, the system and receivers must be
extremely well cleaned, whereas if the system is for
loading, say Brent crude, every batch from the same
tank, then some cross contamination is unlikely to
cause a significant bias. Likewise the considerations
of RVP and density will vary by operation.
One item on which all designers should be clear is
that of the quality of the operations staff used for
each step of the process. In general the number of
critical handling steps undertaken in the field should
be minimised because these steps are largely
uncontrolled and unauditable.This is why in general
we recommend portable receivers; uncoupling a
portable receiver and removing it to a laboratory
takes a mixing and subsampling procedure and
places it in an environment with a much higher
degree of control (and comfort!).
Another issue raised frequently is loss of light ends,
particularly with higher vapour pressure oils.The
loss of light ends can of course affect density
measurement but this can be easily avoided by
ensuring that the receiver relief pressure is set
above the RVP of the fluid at the maximum
temperature that the receiver is likely to
experience. It should also be pre-purged with a light
inert gas blanket.
Where the product is a light hydrocarbon or
perhaps where the oil is spiked with gas, such as
happens in the North Sea then high pressure
receivers must be used.The constant pressure
cylinders used for light hydrocarbon are generally
not suited to crude/water duty as sufficient energy
cannot be imparted to the fluid to mix it for sub
sampling. Cylinders with balls or manually agitated
baffles have consistently been found inadequate for
dispersing water into crude pre-analysis. Jiskoot and
several vendors can provide proven receivers
suitable for this duty.
Laboratory analysis
The laboratory analysis step is of course the key to
the whole system and without adequate analysis
methods any efforts to provide a representative
sample, have no value. In general ASTM 4007
(Centrifuge) has been dropped in favour of
Coulometric Karl Fisher which has been found
substantially better, however care should be taken
over some sensitive issues within this method such
as reagent matching to cell and titration
procedures.
Prove all of the above!
The real test for a sampling system is to prove the
system in its entirety i.e. as a complete system, this
can only be done by water injection and there are
many who will provide this service.Third parties
should witness this operation and all the inspection
companies should be invited to attend.There is no
substitute for proving a sampling system, it should
be repeatable and repeated if primary
characteristics of the system are changed e.g. crude
types, operational rates, fundamental system
components etc.
The API also now allows a component testing
approach which is accredited with equal validity to
a system proving. It is our position that this does
not provide as good an overall test for several
reasons:
Component testing requires that the system
be divided into several key areas such as
pipeline profile, sample receivers etc.
Each one of these tests must have an
associated laboratory analysis which of course
increases the uncertainty of the overall
analysis.
Further to this point a pipeline profile can
CONFIRM a problem but it cannot verify the
suitability of the extractor design. Due to
practical considerations a profile probe may
only have openings designed,and sized a
certain way (normally 6mm). For a profile
probe to substitute for a sample extractor
take-off, the profile probe would require
openings of the same size, design and
operational characteristics as the sample
extractor to allow it to respond in the same
way i.e. a sampling system with a 33mm takeoff would require a profile rake with 33mm
openings!
A proving exercise therefore provides a result with
better overall credibility. However we always
recommend that the sample handling, mixing and
Water%
ISO
API by Tank
API by Meter
0.5
0.050
0.130
0.090
1.0
0.050
0.150
0.110
1.5
0.075
0.160
0.120
2.0
0.100
0.170
0.130
3.0
0.150
0.190
0.150
5.0
0.250
0.230
0.190
analysis be tested as a separate issue before proving
because it is frequently an area of failure.
Proving sampling systems probably provides the
worst contention between the APls final 8.2
(released November 1994) and the current IP and
ISO standards.The API position is clear, the API
determined acceptability criteria for the 1983
standard before widespread knowledge of the
performance of systems or analysis methods were
well established.The acceptability criteria of 0.05%
for a fiscal system is clearly a closer tolerance than
the accuracy of the laboratory analysis as stated
within the standards and herein lies a major
problem.
There are two databases of information, one in the
US, the other in the EC and there is substantial
variance in results. Results from repeated tests
from a variety of typical European systems have
shown consistent repeatability within the original
API/IP and ISO criteria.These results are
unmatched within the US database and the
question must be why?
There are several reasons that explain the
variances, the first must relate to the analysis
methods.The body of results from the UK relate to
specific crude types, frequently handled by an
extremely consistent and experienced staff; based
at export terminals.This alone is bound to reduce
the repeatability/reproducibility errors. By contrast
the US database tends to be from a large variety of
sites with different crude types; all import
configurations.
The second consideration is over the types of
sampling systems proven. In general those in the US
database are generally in-line systems whereas
those in the EC are fast-loop based. No
consideration should be given to receiver types as
in general these are the same. From the body of
the paper, a fast-loop system with a large entry is
far less sensitive to pipeline water dispersion and
distribution.
One aspect on which there is no dispute is the
requirement for repeatability, all parties agree that
the sampling systems should provide repeatable
results i.e. prove within repeatable tolerances.
Once the system is designed and proven to be
acceptable, it is necessary to be able to show the
shipper that the system operated correctly during
the transfer.
There was a long tradition of equipment failure in
equipment designed in the 70s and early 80s which
has largely been overcome by current technology,
however to prevent dispute some form of
performance checking should be provided, this
should generate CONTINUOUS proof of
performance.
The fact that the total demand volume of sample
from a system (typically 15 litres or 5 USG) is
matched by the actual volume in the receiver at the
end of the transfer is no proof of correct sampling.
It is necessary that the performance of all the
component parts are within acceptable linearity for
the complete duration of the operation and can be
so proven.
This means the flow measurement, sample control
system and the sample volume per grab. Acceptable
performance is taken to mean that the volumes
achieved and the flow measured are within +/- 10%
of demand i.e. performance factors are in the range
of 0.9 to 1.1. It would however be a mistake not to
point out that if the system provides a 0.9 figure on
one batch, a 1.1 on the next and a 0.9 on the next
that this is good performance - it shows
inconsistency.
Linearity is more important than absolute accuracy.
Performance factors for the sampling system, both
in regards to the flowmeter performance and to
the sample extractor performance, and in
combination, should be made for both the complete
batch and for short intervals within the batch.
Ideally hard copy printout from the system showing
actual recorded volumes should be made and
signed off by interested parties.
Choosing a Philosophy - Hierarchy of Sampling
When considering the primary choices for sampling
systems, there is evidence to the old idiom that
you get what you pay for performance.The
standards are extremely clear in the fundamental
statement that tank sampling of non-homogenous
media is unacceptable, there is a further hierarchy
that can be noted:
Ships Tank
The worst place to attain a sample is in shipstanks,
this is because the shape of the tank internally is
poorly known and the dynamics of sampling
properly are unattainable .
overcome. All shipboard samplers or any manifold
or loading arm located samplers will be subject to
the question of whether a collection receiver
should be used for each sampler and if one receiver
is to collect samples from 2 - 4 arms then how to
assure the representivity given that the flowmeter
performance and volume per grab of the samplers
may vary per line.
Shore Tank
Land Based Sampling systems
This is better than the ships tank because the shape
is better known, however tests have proven that
sampling a shore tank properly is also unattainable.
Location of the sampling system in a shore based
location, on-shore or at the jetty head provides the
ability to design correctly and operate the samplers,
taking the sample from a single properly mixed
pipeline provides the optimum solution.
Samplers
Considering the relative virtues of shipboard and
other samplers, the designer is faced with the
Recovery Comparison between Shipboard
Line Samplers
Cargos
Crude Types
Total Cargo Volume
Shipboard Results
In Line Sampling System
Water Content Variance
and In-
40
AH, AL, AM, XL
39,569,097 BBLs
112,327 BBLs Water
125,530 BBLs Water
0.033%
requirement to do the best possible job in
providing an accurate sample. Given the state of
many land based systems installed in the 1970s and
early 80s a shipboard sampler can at times
represent an attractive improvement.
It will never be possible for shipboard sampling
technology to achieve the accuracys attainable by
their land based counterparts because of the
constraints imposed by the available equipment,
supplies and sampling system location.
Results of direct comparison of a proven land based
sampling system against shipboard samplers show a
significant undersampling of water by the shipboard
samplers used during the same unloading which
amounted to an average loss of 0.033% to the party
using the shipboards!
Shipboard Samplers in the Manifold
Location of a sampler at the manifold provides no
assurance of homogeneity of the cross section, it
also provides an extremely poor profile for the
measurement of flow with the required accuracy. All
pneumatic Shipboard Samplers are totally
inadequate to the duty because of the rangeability
of the pneumatic control systems themselves and
the susceptability of the control systems to the
vagaries of what is, at best, a wet ships air supply.
Electronic Shipboards do not suffer as much from
the flow measurement problems but the pipeline
homogeneity remains a problem which cannot be
What does Jiskoot offer to optimise performance?
Jiskoot as a turn-key systems manufacturer designs
and supplies fully engineered systems.We can
Flowmeter
Sample
probe
Flow
Jetmix
Jet function
Jet Mixing - High Rangeability
supply sampling solutions from Shipboard through
to full Fiscal and proven solutions, below are
described some of our solutions.
Pipeline Homogeneity
The Jiskoot Jet-Mix is designed for
pipelines with a high degree of flow
rangeability. Systems with a
rangeability of below 4:1 can be
served with static mixers but the
range capability of a Jet-Mix
provides infinite turn down by
design, the lower the flowrate the
more mixing is introduced.This is a
system that can be retrofitted to
existing pipelines without
Inturbine
shutdown.The dispersion quality of
the Jet-Mix is sized to suit the extractor design.
This is the most widely used mixing device for large
diameter fiscal applications on a world-wide basis.
Batches
231
Metered
Custody Volume Flowmeter Performance
Volume (BBLs)
Factor
(BBLs)
41,145,560
41,787,636
0.985
Extraction of a Flow Proportional and
Representative Sample
Flow measurement
Jiskoot performed a large number of tests on all
varieties of flowmeters in the mid 80s including,
turbines, strap on and insertion ultrasonic
flowmeters.This led Jiskoot to design a large rotor
insertion turbine InturbineTM.The Inturbine is
the simplest, most reliable and easily maintainable
Jiskoot 210 In-Line
Sampler
meter available.The size of the rotor, 140mm, and
the synergy of the design to that of an outboard
engine used for motor boats is deliberate.
Long term performance results in Egypt yielded an
average performance factor of 0.98 where they
replaced insertion ultrasonic meters.
Densitometer
Water Monitor
Fast loop
Take-off
This represents ALL the figures, no failures or short
volumes were experienced during the trial period.
Representative Samplers
The bulk of import receipt terminal sampling
systems installed on a world-wide basis are based
on the Jiskoot 210 in-line sample probe. Jiskoot
manufactures a full range of sampler types from
product samplers through Shipboard to In-line and
Fast Loop.
Recently European sampling technology has moved
towards so called Fast Loop systems for several
reasons, not least improved installations, accuracy
and maintainability.They also allow for the
integration of other small bore on line devices
within the same loop such as densitometers and
on-line water monitors.
The accuracy gains are principally from the much
larger loop offtake size which reduces the system
sensitivity to water in oil dispersion and from
reduced dead volume between the sample
extractor and the receiver.The take-off can be
located convenient to the installation without
regard to the location of the main sample extractor
and sample receiver housing.The use of a pump in
the loop ensures that the water/oil mix which
passes through the cell sampler is well dispersed.
The question of Isokineticism is often quoted, by
definition this should mean that the velocity of the
fluid within the opening to the fast loop of a
sampling system matches that of the bulk average
velocity in the pipeline. In practice the API originally
quoted an acceptable range of 50% - 200% and the
IP after testing larger takeoffs a range of 10% 300%. It is impractical to match the loop velocity to
the pipeline velocity at all rates, it has become
obvious that with careful design fully Isokinetic flow
is not relevant to Fast Loop systems with large
bypass pipesize and a correctly designed take-off.
There is also confusion as to the velocity
requirements within the pipe loop itself, as opposed
to the entry.The requirement through the loop
must for practical reasons assure minimal pressure
drops but maintain the oil without allowing the
water to separate; there is an engineering balance
to be achieved here.
A further benefit to the fast loop approach is that
the part which is likely to require the most service,
becomes much easier and safer to service.
Replacement of the key seals in a fast loop system
does not require isolation and removal of a part
inserted into the main pipeline, it is achieved by
shutting the fast loop valves.Typical service time is
reduced by 80%.
Size
Area
CoJetix
250 x 150mm
37,500mm2
33.5mm
881mm2
22 x 8mm
176mm2
There has been a tendency by designers to try to
exclude the use of filters from fast loops. Filters and
strainers are frequently considered as the same
thing. Strainers installed in fast loops for the
purpose of removing large particulates are quite
acceptable.They will neither bias the sample against
sediment or water as the loop flowrate flushes the
filter volume some 1200 times per minute!
Because Jiskoot has developed both Jet-Mix and
Fast Loop Cell samplers, it was natural to
combine the two elements into what is known as a
Co-Jet-Mix system.The Co-Jet-Mix allows
optimal sizing of the system.The inlet of the JetMix loop is supplied with a large scoop takeoff,
this reduces the amount of energy dispersion
required and the sensitivity of the system to
pipeline profile.
The system fast loop is taken across the Jet-Mix
pump where the oil and water are well dispersed.
The argument is sometimes raised that the system
is taking a sample of oil that has already been
sampled because the suction is downstream of the
jet.This is correct but the design works in the same
form as a capacitor which is beneficial.The
maximum error that this could cause to the
sampling system under the worse possible
circumstances would be the volume of the Jet-Mix
loop divided by the volume of the BATCH!
Sample handling and mixing
Jiskoot has a range of sample receivers suited to
both low and high vapour pressure crudes, the
simplest is a Cubitainer, developed in conjunction
with the Trans Alpine pipeline (TAL), which is a
disposable receiver designed not to vent, through
to our PR-53 all stainless receiver which is
frequently copied, to our PIM type receiver which is
a dual piston sample receiver for high vapour
pressure crudes.
Laboratory mixers are available to mix each of our
receiver styles and these can vary from fairly simple
pumped loops through to an integrated laboratory
mixing system with PC reporting, running under MS
Windows.The degree of sophistication varies
according to the needs of the user, if a large
number of samples are handled then an integrated
system is useful.The Lab PC system weighs the
sample receivers, suggests mixing times, logs the
mixing duty and integrates data from Karl Fisher,
Densitometer, temperature and pressure into a
report that is uploaded to a local area network and
integrated into a SCADA supervisory reporting
system.
Sampler Control Systems
Jiskoot supplies a wide range of sampler controllers
from simple divider counters through to
sophisticated micro-processor control systems
providing full reporting, logging and interfacing to a
SCADA system such as those supplied to terminals
at MOT and Arab Petroleum Pipelines (SUMED).
Jiskoot also supplies the only certified zone 0
shipboard sampling system, with full data logging for
those that require this duty.
Full Scada systems that provide a graphical user
interface to multiple sampling systems as in the
case of Sumed also provide for a link to their
plantwide computer system and integrate the lab
results directly into bills of lading.
