SPE 146394
Why Y?
John M. Williams, The Petroleum Institute Abu Dhabi
Copyright 2011, Society of Petroleum Engineers
This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Denver, Colorado, USA, 30 October2 November 2011.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
This paper provides an analysis of a widely-used
pressure-volume-temperature (PVT) parameter,
demonstrating common errors made in its use and the lack
of detailed descriptions in common reservoir engineering
texts. Use of the Y-function is a valuable tool for
increasing the accuracy of bubble point measurements,
and it has other important uses, but its detailed application
is poorly understood, leading to unnecessary errors in one
of the most fundamental PVT properties.
Best practices have been developed and are presented,
to encourage proper application of the Y-function,
together with detailed recommendations for setting up
Excel spreadsheets to help identify the most precise
bubble point, and optimize Y-function values.
More extensive use of the Y-function to obtain bubble
points for down hole or surface oil samples is
recommended as it could improve quality control and
selection of more representative samples.
Introduction
The Y-function determination is a traditional part of oil
PVT studies, providing an equation to match the two-
phase region of the flash study, but its uses are often not
fully understood, both by PVT specialists and reservoir
engineers. The subject has been raised in talks by an SPE
Distinguished Lecturer, but there is a need for a reliable
reference to the technique giving a balanced analysis.
Proper use of the Y-function can lead to bubble point
measurements of the highest precision, essential for
reliable reservoir property correlation work and of great
use in the quality control of samples, but the mathematical
procedures must be tempered by a fundamental
understanding of the measurements themselves.
This paper presents a detailed approach to Y-function
optimization using basic and advanced Excel spreadsheet
techniques, which can be used to ensure bubble point
measurements are the most accurate physical property
measured on the reservoir fluid. The approach also
addresses the problem of calculating fluid
compressibilities without excessive smoothing.
The Y-function has been mentioned in talks by an SPE
Distinguished Lecturer7, but there does not seem to be a
reliable reference to the technique giving a balanced
analysis. Although numerous papers and books refer to the
use of the Y-function, there is rarely any discussion of the
experimental method used to collect the data, nor of the
effect of measurement accuracy on bubble point pressure
and Y-function values. Both these issues are discussed in
some detail here.
Experimental Procedure
The Y-function is obtained from pressure and volume
measurements during the flash liberation experiment in a
Pressure-Volume-Temperature (PVT) study of an oil
reservoir fluid. This study, usually performed at the
reservoir temperature, involves stepwise increases in
volume with equilibrium pressure being achieved at each
stage, typically by agitation the fluid being agitated Since
there is no change in the overall composition of the
sample, this experiment is also known as the constant
mass study (CMS) or constant composition expansion
(CCE). The pressure response is normally in two distinct
parts: while the fluid is a one-phase liquid, the pressure
falls rapidly with decreasing volume, corresponding to the
compressibility of the oil; once below the bubble point
pressure decline is much slower since the increasing
proportion of gas greatly increases the compressibility of
the system. In most cases, the volume-pressure plot gives a
clear change in slope at the bubble point, and this is the
basis of the experimental determination of bubble point.
Note that the experimental volume is often reported in
terms of relative volume, the volume divided by the
volume at the bubble point pressure. One-phase data are
used to determine oil compressibility, while the two-phase
data below bubble point pressure, and the bubble point
data are used to derive the Y-function.
Similar pressure-volume experiments are commonly
used to obtain bubble point pressure for a liquid sample,
which serves as an important quality control parameter.
Although these measurements may be made at reservoir
temperature for down hole samples, or separator
2 SPE 146394
temperature for separator samples, they are commonly
performed at ambient temperature, and no Y-function data
are reported. An example of a bubble point determination
is shown in Figure 1, where measurements have been
plotted by hand, and the bubble point pressure obtained by
finding the intersection of hand-drawn curves for the one-
phase and two-phase parts of the P-V response.
It should be noted that the bubble point pressure is
almost always an interpretation based on the measured
data, especially when the fluid is studied in a pressure cell
that has no viewing port of window (commonly referred to
as a blind cell). The same situation occurs when the
measurement is performed in a sampling tool or sample
bottle. Even with a windowed-cell, the cell geometry is
such that it is not possible to observe the first
infinitesimally small bubble of gas which corresponds to
the bubble point, but subsequent observation does provide
a lower limit on the measurement.
Figure 1. Bubble point determination on a down hole
sample.
Other factors can also affect the determination of the
bubble point. In the case of very volatile oils, the oil
compressibility is high and changes significantly with
pressure such that it may be difficult or impossible to
identify the bubble point from a discontinuity in the
volume-pressure response, and a visual observation of free
gas becomes essential. Highly viscous crudes may take
exceptionally long times for gas bubbles to separate from
the oil, which means the system may not reach full
equilibrium during the experiment, and visual observation
may be impossible. If a sample is not thoroughly
homogeneous, for example a down hole sample that has
separated into oil and gas phases while the sampler is
being retrieved, and has not been mixed together, the
pressure volume curve usually does not show a clear
break, since different portions of the fluid will have
different bubble points, and the overall response is that of
a combination of different fluids. If possible, such a
measurement should be repeated after increasing the
sample pressure are further agitation and/or equilibrium
time.
Experimental technique and the experience of
laboratory personnel can also affect measurement
accuracy. Sometimes measurements are made while trying
to achieve equilibrium pressures at round numbers, but
this can lead to multiple adjustments to cell volume, each
of which disturbs the equilibrium, and may lead to data
either being recorded before full equilibrium has been
achieved, or to pressures being approximated to the target
value. Data at rounded values is thus likely to be of lower
accuracy, especially in the two-phase region where
equilibrium takes longer. Best practice involves using
target pressure values at regular intervals, but once near
the target, bringing the system to proper equilibrium and
then recording the exact pressure. Once below the bubble
point, it is advisable to use target volumes, starting by
repeating the volume increases just before the bubble point
for at least two steps, and then doubling volume increases
until the maximum cell volume is achieved.
In some cases supersaturation has been observed where
the system pressure drops below the actual pressure,
before rising again. This may be of very short duration,
and is often recognized as an indication that the pressure
has just passed below the bubble point pressure. However,
sometimes the situation lasts longer, especially with
measurement systems with no, or little agitation, and it
may be possible to make measurements in the metastable
state. This is actually a continuation of the one-phase
pressure response. Once a higher pressure is detected as
the system becomes two phase, it is clear the previous
measurement was under supersaturated condition.
Y-function Definition
Firstly it is worth pointing out that the term Y-function
is not exclusive to petroleum engineering, and searches
indicate other fields such as nuclear physics where the
term is used.
In petroleum engineering the term is defined by the
following empirical equation:
Y = (Pb-P)/[(Pabs) x (Vr-1)] (1)
where Pb represents the bubble point pressure, P
represents the pressure for an equilibrium measurement in
the two-phase region, Pabs is the same pressure in
absolute units, and Vr is the relative volume for the same
equilibrium measurement, namely the actual volume
divided by the volume corresponding to the bubble point.
Since (Pb-P) is a difference, both pressures can be in either
gauge or absolute units, providing they are consistent. The
SPE 146394
term (Vr-1) may also be written (V/VPb-1). It is
immediately clear that the Y-function formula becomes
very sensitive for equilibrium measurements close to the
bubble point where the numerator and the denominator
approach zero.
Figure 2 shows data from reference 2, in which the
reported bubble point has been excluded, on the basis this
it is an interpretation based on the other data. Three
different plots of Y-function values have been made using
three different estimates for the true bubble point. These
clearly demonstrate the sensitivity of the Y-function to the
bubble point pressure, and indicate that the actual bubble
point pressure of 2620 psig was apparently not the most
precise value.
Figure 2. Y-function dependence on selection of Pb
Analysis of Y-function Uses
1. Quality Control of P-V Data
One of the most basic applications of the Y-
function, like any plot of data, is as a quality control
tool. Any unusual deviations from a smooth response
should be investigated. They may simply be due to
transcription errors in entering measured values in
the spreadsheet, but can also derive from errors in
recording experimental measurements, or possibly
errors in experimental techniques. In the latter case, it
is better to eliminate the questionable datum, unless
the mistake can be corrected with certainty, for
example if an obvious digit has been missed from the
recorded value. As explained earlier, the Y-function
is extremely sensitive to measurement precision near
the bubble point pressure, so is less useful for quality
control of data in this location. Laboratory classes in
which students use pressure-volume data to develop
3
Y-function relations1, often involve transcription
errors which are readily identified.
2. Smoothing of Volume Data
The use of the Y-function for smoothing volume
data has been proposed6, using a rearranged version
of the best fit line Y = aP + b. However, a review of
the different accuracies in experimental data, show
that pressure measurements are typically reported to
1 psi in several thousand (e.g. 0.02%) while volume
precisions are typically 0.005 cm3 in 100 cm3 (0.005
%). it is thus probably more realistic to modify
pressures, as changes will probably be within the
normal precision of the measurements. Note
however, that no smoothing of data should be
attempted until the most reliable bubble point datum
has been obtained as described below.
3. Validate and Increase the Precision of the Bubble
Point
The formula for the Y-function is extremely
sensitive to errors in selection of the bubble point
pressure, as indicated previously in Figure 2. Since
no physical measurement is usually made at the true
bubble point, the Y-function curve is an excellent
tool for optimizing the bubble point selected. Most of
the reference texts consulted2,3,5, do not report this
feature of the Y-function and infer that it may not be
used as regularly as it should. Most references
overlook to point out that changing the bubble point
pressure normally requires correction of the volume
at the bubble point, since both are integral to the Y-
function formula. One reference that does raise the
issues of volume errors7, suggests using a separate
X-function for smoothing volume data.
The data used in the generation of Figure 2 are re-
plotted in Figure 3 after optimization of the bubble
point to 2622 psig, rather than the 2620 psig value
reported in the original text2, clearly showing the
value of the technique. Figure 3 also shows data
plotted assuming errors of +1 or -1 psi in the reported
values, and this clearly illustrates that this level of
error is adequate to explain the remaining Y-function
deviations from the best fit straight line.
With some data, it may be possible to use this
form of interpretation method to obtain values for
bubble point pressure with surprising precision,
namely fractions of a pound per square inch. Of
course, this precision should not be confused with the
accuracy of pressure measurements, typically quoted
as a fraction of a percent of the full scale reading of
the pressure gauge, meaning that measurements
made in the lower region of a pressure gauge may
have relatively low accuracy. This can be improved
by proper calibration of the pressure gauge against a
reference pressure system. In fact, close inspection of
the Y-value data in Figure 3 shows systematic
positive deviations from the best fit line at low
pressures and negative deviations at higher pressures
4 SPE 146394
(apart from the pressures near the bubble point). This
could well be due to gauge accuracy variations.
Figure 3. Y-function after Pb Optimization, and
Effect of Errors
Y-function plots for sample bubble points have
never been seen by the author in laboratory reports,
but may be used by some laboratories. More
extensive use of the Y-function to obtain bubble
points for down hole or surface oil samples could
improve quality control and selection of more
representative samples. More precise measurements
would enable higher confidence in sample quality,
especially if the same pressure gauge is used for all
the samples, so that errors in accuracy would be
identical. This is in fact common practice in many
laboratories.
More precise bubble point measurements of
reservoir fluids may also help quantify compositional
differences in fluids, especially when one phase
samples are available as these avoid errors in
recombination derived from gas oil ratio
measurements.
4. Modelling of Reservoir Behavior (the Y-Method)
The fact that the Y-function usually provides a
good fit of two-phase data has traditionally been used
as a simple model for volume changes in the
reservoir3,4,8. This approach is often referred to as the
Y-method, and can be used as a tool to evaluate or
predict pressure and volume changes in a reservoir
with production. For reservoir applications, a form of
the Y-function similar to the following is used:
Y = (Pb - PR )Boi/[PR(Bt - Boi)] (2)
where Pb is again the bubble point, PR is the reservoir
pressure, Boi is the initial formation volume factor,
and Bt is the formation volume factor at reservoir
pressure PR.
There are obviously major constraints with this
method, essentially linked to the volumetric changes
occurring in the reservoir being consistent with the
way the flash liberation study is performed in the
laboratory. This means that additional corrections
need to be included if a gas cap is present, if water
production occurs, or there is a significant water
drive. The method is certainly not recommended if
significant differential behavior is occurring in the
reservoir. Since modern reservoir modelling tools are
much better at combining all the likely contributions
to reservoir pressure-volume changes, it appears that
the Y-method is now of limited value. Whereas the
method is covered in a former edition of the
Petroleum Engineering Handbook8, there is no
reference to it in the index of the latest edition.
Excel Spreadsheet Techniques
Excel spreadsheets are very widely used to record and
interpret laboratory measurements, and this is certainly
true of PVT experiments and Y-function data. Careful use
of a number of techniques can facilitate data handling, and
ensure best quality bubble point and Y-function values are
obtained. The following basic techniques are
recommended, though spreadsheet capabilities are
regularly extended by software developers, and more
powerful techniques such as the use of macros may be
suitable.
- Set up a spreadsheet with a view to using it as a
model for subsequent use, so not include variable
values in formulae, and anticipate that the number
of measured data will vary between experiments.
- Build a table of pressure and volume (or relative
volume data) and add a blank line in the table
with a label add new lines above this line.
Make sure that data sets plotted in graphs include
the blank lines, and then data in additional lines
inserted above will also be plotted. Otherwise any
data added will not be included in the graph
without
- Allocated separate cells for parameters such as
bubble point pressure and volume at the bubble
point pressure, and link all calculations of Y-
function to these cells. This makes it easy to
optimize the bubble point pressure by viewing the
effect of changes in the Y-function graph. If the
spreadsheet includes a best fit curve for the one-
phase response, this can be used to calculate
bubble point volume automatically from bubble
point pressure.
- It is also possible to include a parameter which
indicates the number of data which represent the
one-phase part of the flash liberation study.
Entries can be numbered and a suitable IF
statement used to separate the pressure or volume
data (or both) into separate columns facilitating
plotting, and curve fitting. This functionality
helps when a new set of data has a different
number of one-phase points than in the previous
data set, and is very useful when there is one
datum (or more data) very near the bubble point,
and the plots from different data sets need to be
compared.
SPE 146394
- Obtain a best fit curve for the one phase data
keeping the order of the polynomial as low as
possible. This enables the calculation of bubble
point volume from bubble point pressure, as
mentioned above.
- It is very useful to obtain a best fit curve for the
same one-phase data when plotted with volume
on the y-axis. Differentiating the resulting best fit
line equation, provides dV/dP values which can
be used to derive oil compressibility values using:
Co = -1/V.dV/dP (3)
- If smoothing data by using a best fit line from
Excel, select the Trendline Label, choose
Format Trendline label in the right click points for down hole or surface oil samples is
menu, and set the format to scientific with at least
four decimal places. This is especially important
if the best fit equation contains both positive and
negative terms.
- Since there appears to be no way to access best fit
line formulae from Excel graphs within the
spreadsheet cells themselves, either the formulae
need to be transcribed manually, or a more
advanced function must be used to derive the
parameters of the best fit within one or more cells
in the spreadsheet. For example:
o SLOPE Returns the slope of the
linear regression line through data
points in known y's and known x's.
o INTERCEPT Calculates the point at
which a line will intersect the y-axis
by using existing x-values and y-
values.
This option eliminates the problem with loss of
precision in best fit line parameters mentioned
above.
- Whether the original experimental volume data
were given in the form or relative volumes or cell
volumes, it is advisable to recalculate relative
volume data based on the bubble point volume
optimized by use of the Y-function. T
- Once the spreadsheet has been used successfully,
clear the input data and re-save as a model. Any
parameters or changes which must be entered by
hand can be explained by adding a note entry in
the appropriate cell. To avoid the model being
overwritten, set the file to read only (right click
on the icon and choose properties and select the
read only box.
Conclusions
Use of the Y-function is a valuable tool for increasing
the accuracy of bubble point measurements, and it has
other important uses, but its detailed application is poorly
understood, leading to unnecessary errors in one of the
most fundamental PVT properties.
5
This work includes the following:
Review of the uses of the Y-function and their
limitations
Analysis of the effect of experimental
uncertainties
Review of experimental measurements with some
best practices proposed.
Recommendations for optimization of the Y-
function, with demonstration of the benefits.
Detailed recommendations for the development
of Excel-based techniques for improved bubble
point determination and Y-function optimization.
More extensive use of the Y-function to obtain bubble
recommended as it could improve quality control and
selection of more representative samples. It may also help
with identification of compositional variations in
reservoirs.
Acknowledgments
The author would like to thank the Petroleum Institute
for permission to publish this work. The author would also
like to thank Mr. Jan Srensen, formerly of Saga
Petroleum, who first asked the author the question that
forms the paper title, and Mr. Mimoune Kissami, also at
the PI, for useful discussions on the subject of the Y-
function.
References
1. Williams, J.M.: Dedicated HSE Education for
Undergraduate Petroleum Engineers, SPE paper
98691, presented at the SPE International Conference
on Health, Safety, and Environment in Oil and Gas
Exploration and Production held in Abu Dhabi,
U.A.E., 24 April 2006.
2. The Properties of Petroleum Fluids, William D.
McCain Jr., 2nd Edition, PennWell Publishing
Company, 1990. ISBN: 0-87814-335-1
3. Volumetric and Phase Behavior of Oil Field
Hydrocarbon Systems, M.B. Standing, SPE, Dallas,
1977
4. Applied Petroleum Reservoir Engineering, Craft, B.C.
and Hawkins, M.F., Prentice Hall Inc., New Jersey,
1959
5. Phase Behavior, Whitson C.H. and R. Brul, M.R.,
S Monograph volume 20, Sciety of troleum
Engineers, 2000, ISBN: 1555630871, 240 pages
6. Reservoir Engineering Handbook, Ahmed, T.,
Elsevier Inc., 2006, ISBN 13: 978-0-7506-7972-5
7. Reservoir Fluid Properties - State of the Art and
Outlook for Future Development, Marhoun, M., SPE
101449 (Presentation), 4th July 2002
8. Petroleum Engineering Handbook, Editor Bradley
H.B., 1987, SPE, Texas, USA, p. 40-6.