You are on page 1of 29

SPE-196009-MS

Time-Enhanced Material Balance Analysis

Ralph G McNeil, IHS Markit retired; Aleksandr Antonov, S. Hamed Tabatabaie, and Louis Mattar, IHS Markit

Copyright 2019, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Calgary, Alberta, Canada, 30 Sep - 2 October 2019.

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
The traditional Material Balance Analysis only utilizes cumulative production and static average reservoir
pressure. Time does not feature in the equations. However, there are significant advantages if time is
incorporated into the analysis. For example: a) identifying if all the wells belong to the same reservoir; b)
identifying the effect of external energy sources such as gas or water drive; c) incorporating the contribution
of communicating tight reservoirs; d) visualization of the results in pressure-time format.
The time-based analysis presented in this paper supplements the conventional methods. It helps reduce
the non-uniqueness of the solution. In contrast to the conventional Havlena-Odeh plotting variables, which
are complex and non-intuitive, the pressure-time plot and corresponding pressure-history match are much
easier for an engineer to comprehend and to evaluate the validity or uniqueness of the results.

Introduction
The classical Material Balance Equation (MBE) has been presented in different formats in the reservoir
engineering literature. A basic form is presented in Equation 1 (Havlena & Odeh 1963).

Equation 1

The variables in Equation 1 are defined in the Nomenclature. A few relevant comments are significant:

• Np, Rp, Wp and Δp are known (measured) quantities.

• N is the original-oil-in-place – unknown; required to be determined using MBE.

• m is the ratio of initial hydrocarbon volume of the gas cap to initial hydrocarbon volume of the oil
and is given by m = (GBgi)/(NBoi). G is the original-gas-in-place – (It is usually unknown and is
required to be determined using MBE); In certain fromulations of the MBE, G is used explicitly
instead of m.
• We is the water influx from an aquifer – definitely unknown; required to be determined using MBE.

• The remaining variables in Equation 1 are fluid and rock properties and are either constant or vary
with pressure, and are usually known either from lab measurements or correlations.
2 SPE-196009-MS

Havlena and Odeh (1963) re-formulated Equation 1 into Equation 2, which is written in terms of:

• F- net reservoir withdrawals (production).

• Eo - oil expansion.

• Eg - gas expansion.

• Efw - expansion of rock and connate water.

• We - water influx.

Equation 2

2 can be viewed as the equation of a straight line for which N, G and We are unknowns. In principle, 3 or
more data points are required, and if more than 3 points are available, the method of least squares error is
used to determine the best values of N and G. In general, the LHS (left-hand-side) of Equation 2 is plotted
against the bracketed term on the RHS (right-hand-side), and an aquifer model is assumed and adjusted
until the resulting graph is straight. The resulting intercept and slope are the original- oil-in-place (N) and
the original-gas-in-place (G).
While the plot of Equation 2 is straightforward, the practice of using this equation results in very
ambiguous answers. In this paper we will demonstrate some of the serious issues associated with the practice
of using the MBE, and how incorporating a time-focused analysis can help resolve some of them. Moreover,
we will give examples that demonstrate that the MBE of Equation 2 cannot yield a unique answer for N
or G without additional information that is external to the MBE (for example geological, size of gas cap,
size and strength of aquifer). In the absence of this external information, the answer is non-unique and can
be misleading if taken at face value.

Limitations of MBE
While the MBE of Equation 1 or Equation 2 is rigorous and mathematically correct, it has significant
potential for errors as discussed by several authors (Tehrani 1985, Gilicz 1991, Carlson 1997). Some errors
are due to the very nature of the equation:

• It is a "stiff equation", for the case of a single-phase oil system.


∘ The compressibility of single-phase oil is so small (~10-6 vol/vol/psi) that a very small error
in the measurement of the reservoir pressure can result in a significantly different volume of
oil-in-place.

• There is a huge contrast between gas and oil compressibility.


∘ Gas is 100 to 1000 times more compressible than oil, and a small change in gas volume
corresponds to a large equivalent change in oil volume.

• The compressibilities of oil and water are very similar in magnitude (~10-6 vol/vol/psi).
∘ Consequently, it is very difficult to differentiate between oil and water volumes-in-place on the
basis of pressure alone.

• The mathematical re-formulations of the MBE, while mathematically correct can result in
significantly different answers depending on what variable is being minimized when using the
method of least squares (for example, for the same dataset, minimizing the error in F/Eo will result
in a different value for N than minimizing the error in pR).
The limitations of the MBEs and possible ways to overcome them will be illustrated using case studies.
SPE-196009-MS 3

Case 1: Oil Reservoir with Gas Cap - G and N not known


Figure 1a shows the location of the oil and gas wells for the Judy Creek Viking A pool in Alberta.
Figure 1b shows the oil, gas and water production data.
Figure 1c shows the Havlena-Odeh plot.

Figure 1a—Location of oil (green) and gas (red) wells

Figure 1b—Production data

Figure 1c—Havlena-Odeh plot

It is obvious from the well map that this is a gas-cap/oil reservoir. There is no indication of a
communicating aquifer, either geologically or from observing the production and pressure data. Therefore,
the Havlena-Odeh plot should be a straight line of intercept=N and slope=G.
Even though the pressure data were considered to be of good quality (long shut-ins in high permeability
reservoir) the Havlena-Odeh plot shows considerable scatter, and many different straight lines can be drawn
through the data resulting in a wide range of N or G.
4 SPE-196009-MS

Inspection of the pressure data on a pressure-time plot, Figure 2a, reveals interesting facts that are
totally obscured by the (mathematically correct) Havlena-Odeh plot. When viewed in this way, it becomes
immediately clear that the 7 wells, with the 25 data points circled in red, do NOT belong to the same pool
as all the other wells. A new pool was defined, excluding these 7 wells. The resulting pressure-time plot
is shown in Figure 2b.

Figure 2a—Pressure-time plot of ALL gas and oil wells

Figure 2b—Pressure-time plot of Viking A pool wells, EXCLUDING 7 wells

It was hoped that the use of the properly grouped oil and gas wells would result in an easier-to-interpret
Havlena-Odeh plot. The revised plot is shown in Figure 3a. While it shows a more coherent data set, it is
still difficult to pick a correct straight line through the plotted points. A series of potential interpretations
are shown in Figure 3b, and the range of answers is unacceptably wide.

Figure 3a—Havlena-Odeh plot of Viking A pool, EXCLUDING 7 wells


SPE-196009-MS 5

Figure 3b—Possible Havlena-Odeh interpretations

A pressure-time history-match was incorporated to supplement the Havlena-Odeh Analysis in the hope of
reducing the range of answers. By this, we mean that a reasonable Havlena-Odeh straight line was selected,
and the resulting N and G were determined, Figure 4a. Then, using these N and G values and the actual
oil and gas production, the average reservoir pressure was calculated for every production interval, and
a continuous calculated average reservoir pressure was determined. This was compared to the measured
average pressures, Figure 4b.

Figure 4a—One possible Havlena-Odeh interpretation (N = 191.8 MMstb, G = 54 Bscf)


6 SPE-196009-MS

Figure 4b—Corresponding pressure history match (solid line is calculated from MBE; symbols are measured data)

It is evident from the history-match, Figure 4b, that this is not a very good match. It turns out that a
small change in the Havelena-Odeh line could significantly improve the history match, as seen in Figures
5a and 5b, below.

Figure 5a—Slightly different interpretation (N = 273.6 MMstb, G = 16 Bscf)

Figure 5b—Much better pressure history match

Obviously the Havlena-Odeh interpretation when coupled with a pressure-time history match is much
better than the Havlena-Odeh interpretation on its own. However, in spite of this significant improvement
SPE-196009-MS 7

in interpretation, and excellent pressure history-match, we observed that there were many Havlena-Odeh
matches (different combinations of intercept (N) and slope (G)) that resulted in equally acceptable pressure
history-matches. Three such matches are shown below, in Figures 6a, 6b, and 6c.

Figure 6a—Match 1: N = 300 MMstb, G = 7 Bscf

Figure 6b—Match 2: N = 200 MMstb, G = 45 Bscf


8 SPE-196009-MS

Figure 6c—Match 3: N = 100 MMstb, G = 82 Bscf

It is evident from the above that the Material Balance Equation does NOT yield a unique solution for N
and G, even when the data quality is good. This problem is not a result of the Havlena-Odeh formulation,
but is intrinsic to the nature of the Material Balance Equation. The reality is that there are two sources of
energy, namely, expansion of the oil leg and expansion of the gas cap, and accordingly a small gas cap (G)
and a large oil leg (N) can have the same expansion energy as a small gas cap (G) and a large oil leg (N).
Therefore, to be able to use the MBE, either G or N must be specified using independent information, such
as: a) absence of gas cap (G=0), b) ratio of net gas pay to net oil pay from petrophysical logs, c) geological
mapping of the gas cap, d) geological mapping of the oil leg, or e) geologically determined ratio (m) of the
gas pore volume to the oil pore volume. Given any of this information, that is external to the MBE, will
allow the MBE to yield unique values of N and G.
For the data of this Judy Creek reservoir, we investigated the whole range of N and G values that would
result in a good history-match of the pressures similar to those of Figures 6a, 6b, and 6c. The result is shown
in Figure 7a. The value of G represents the free gas (in the gas cap), but it should be remembered that the
oil in the oil leg also contains solution gas. It is noted that as the size of the gas cap (G) decreased, the size
of the oil leg (N) increased and the gas dissolved in the oil (N*Rsi) increased. It was interesting to note that
in every case Figure 7b, the total gas-in-place (gas-cap-gas + solution-gas) was a constant (118 Bscf). This
phenomenon has been confirmed with "synthetic" gas-cap/oil-leg data sets.
SPE-196009-MS 9

Figure 7a—Complete range of N and G values that give a good pressure history match similar to Figure 6a, b, c, above

Figure 7b—Tabulation of complete range of N and G values

If the bubble point pressure is less than the initial reservoir pressure, then a gas cap cannot exist and
G=0, in which case N can be uniquely determined. In the case of the Judy Creek Viking A reservoir, there
is no doubt that a gas cap does exist (see well locations in Figure 1a); there are nearly as many gas wells
as oil wells.
There are numerous situations where the bubble point pressure equals the initial reservoir pressure,
indicating the possible presence of a gas cap, but no independent estimate of its size. This is the situation in
this example – there definitely is a gas cap present, but there is no independent confirmation of its size. In
such a situation, it is sometimes possible to conduct a Gas-Material-Balance (p/z) calculation on the early
pressure data of the gas producers and obtain an estimate of the size of the gas cap – before a significant
amount of solution gas is exsolved from the oil or any oil is produced from the oil leg. This is shown in
Figure 8. In this example, 80 Bscf is a reasonable estimate of the initial gas-cap size (G). Referring to Figure
7b, this value of G (80 Bscf) yields a corresponding value for maximum OOIP (N) of 100 MMstb. Note
that the curvature of the data points, away from the p/z straight line in Figure 8, represents the effect of un-
produced solution gas in the total system from the oil leg.
10 SPE-196009-MS

Figure 8—Gas Material Balance only on gas wells producing from the gas cap

Learnings from Case 1


This case illustrates that.
1. A pressure-time plot helps identify wells that belong to the same pool and wells that do not. A pressure-
time plot must always be a first step in Material Balance Analysis.
2. Interpretation of the Havlena-Odeh plot is greatly aided when the analysis is supplemented with a
history-match on a pressure-time plot.
3. In spite of having relatively good and plentiful pressure data, the MBE results are non-unique, even
when the pressure-time history-matched plot is used.
4. The total gas in the entire reservoir (gas-cap-gas + solution-gas) is constant, for all the combinations
of N and G values.
5. The only way to determine N uniquely is if G (or its equivalent) is specified from an independent
source.
6. It is possible to estimate G, if the gas wells are analyzed and the initial slope of the p/z analysis used.

Case 2: Gas Reservoir with Water Drive


The Clarke Lake Slave Point A is a large gas field in north eastern BC, Canada. During the pressure-
time portion of this study, we separated the Slave Point wells into more than 3 distinct groups, where the
water drive was strongest in the eastern wells and weakest in the western wells. The group discussed herein
demonstrated the strongest water drive effect. It contains 37 gas wells. At the time of this study, these wells
had produced 910 Bscf of gas 130 MMstb of water. A summary of the production is shown in Figure 9.
SPE-196009-MS 11

Figure 9—Production Data

When the data are plotted in Gas Material Balance format (p/Z versus cumulative gas produced) the result
is shown in Figure 10. By all standards, the scatter in the pressure data is considered reasonable. This results
in a range of interpretations in OGIP from 2.2 to 3.2 Tscf.

Figure 10—Gas Material Balance (p/Z) plot, showing range of interpretations

On the plot of Figure 10, a large number of data points appear to be clumped together, as shown by the
circle. The natural tendency is to view these points as multiple confirmations of the true reservoir pressure
(with measurement scatter), and regarding the preceding points as not fully built-up, thus encouraging the
3.2 Tscf interpretation. However, the more revealing way of understanding these data is to look at them
on a time-basis, as shown on Figure 11. The clump of pressure data points is now stretched over several
years, and shows that, even though production is continuing (albeit at a much smaller rate), the pressure is
increasing systematically for about two years and then decreasing slowly or stabilizing thereafter.
12 SPE-196009-MS

Figure 11—Clump of pressures is clearly stretched out when plotted over time

It is always good practice to reverse-engineer the Material Balance Equation and create a full-life pressure
plot to history-match the measured data. Using the two interpretations (2.2 Tscf and 3.2 Tscf), the average
pressure was calculated for all time, using the production data and the MBE. For these two OGIP (2.2
and 3.2), the history-match is shown in Figure 12. The 2.2 Tscf line matches the early years, but clearly
NOT the late years. On the other hand, the 3.2 Tscf line matches the late data, but does NOT honour the
intermediate years.

Figure 12—History-match of 2.2 Tscf and 3.2 Tscf Reservoirs

Very often, when doing a gas material balance analysis, water production is ignored for two reasons:
a) usually the reported water rates are not reliable, and b) because gas is so much more (1000 times)
compressible than water, and the water contributes very little energy to the system. However, it is noted
that the Clarke Lake Slave Point A reservoir has produced a significant amount of water (130 MMstb),
which indicates the presence of a large body of water in communication with the gas. Moreover, the slow
SPE-196009-MS 13

rise in pressure during the 1980s and 1990s, in spite of continuing production, is indicative of an active
aquifer (water drive). The pressure data has been re-interpreted, using the MBE with an active aquifer
(Fetkovitch formulation, Moghadam et al, 2011). The water drive analysis consists of accounting for the
aquifer influx and plotting p/Z** instead of p/Z. Note that using the p/Z** analysis shifts the non-aligned
p/Z data according to the strength of the aquifer, until they form a straight line. This is demonstrated in
Figure 13.

Figure 13—Comparison of p/Z and p/Z** (aquifer), showing data shift to align all points

The p/Z** analysis using an active aquifer, and the corresponding pressure-history-match are shown in
Figure 14 and Figure 15. This analysis, which incorporates an aquifer, results in a much smaller OGIP (1.7
Tscf, +/- 0.2 due to data scatter), and it gives a much more consistent history match of the measured data,
throughout the whole life of the reservoir (see Figure 15).
14 SPE-196009-MS

Figure 14—The p/Z** analysis using an active aquifer

Figure 15—History-match of 1.7 Tscf gas reservoir with an active aquifer

To confirm our understanding of the situation, and to illustrate the effect of an active aquifer, a synthetic
dataset was created and labelled Case 2a.

Case 2a: Simulated Gas Reservoir with Water Drive


A gas reservoir was modeled with an active water drive (Fetkovitch aquifer model). The production history
is shown in Figure 16. The reported pressure data were obtained from the simulator, but were chosen at
different times in such a way as to mimic a real-life situation – some long shut-ins, some short shut-ins.
These are given in Table 1.
SPE-196009-MS 15

Figure 16—Production history of well in simulated gas reservoir with active aquifer

Table 1—Simulated reservoir pressures to mimic real-life data frequency and shut-in durations

A p/Z material balance plot is shown in Figure 17. As in the previous example (Case 2), two possible
interpretations are shown, reflecting which data are considered to be more reliable and representative. Using
just the p/Z plot, we might easily consider the last few points to be data scatter. However, when these pressure
points are plotted on a time scale, along with the production data, a different picture emerges, see Figure
18. The time-pressure plot shows that the pressure increase is occurring over a period of several years, it
is systematic (not data scatter) and it is significant.
16 SPE-196009-MS

Figure 17—Two potential p/Z analyses giving significantly different answers

Figure 18—Pressure data plotted on a Time plot, along with production history

The pressure data were re-interpreted assuming a water drive, using the p/Z** analysis (Moghadam
2011). As a starting point, a reasonable estimate of the size of the gas reservoir can be obtained by analyzing
the early time data on a p/Z plot, before any significant water influx has occurred from the aquifer (Figure
19a). This by itself is a much better interpretation than that previously shown in Figure 17 (10 or 18 Bscf).
The rigorous analysis using p/Z** (accounting for aquifer support) is given in Figure 19b, and gives an
OGIP of 6.5 Bscf.
SPE-196009-MS 17

Figure 19a—p/Z plot through initial data points gives estimate of OGIP (7.8 Bscf)

Figure 19b—p/Z** plot straightens ALL data points and gives estimate of OGIP (6.5 Bscf)

The history-match using the MBE and 6.5 Bscf with aquifer support is shown in Figure 20. It is a perfect
match because the data were derived from a simulator.
18 SPE-196009-MS

Figure 20—History Match of Gas-Aquifer Material Balance Equation with Data

Learnings from Case 2 and 2a


These cases illustrate that.
1. A pressure-time plot helps identify wells that have an external energy source, such as an active aquifer.
A tell-tale sign of an external energy source (water drive or connected reservoir) is the trend of
consistently increasing static pressures when there are long periods of no production.
2. If the external energy source is not identified, the resulting Material Balance Analysis can be
significantly wrong (Case 2 - Clarke Lake: 1.7 Tscf compared to 3.2 Tscf; Case 2a - Simulated case:
6.5 Bscf compared to 18 Bscf).
3. The interpretation of a gas reservoir with water-drive is reasonably unique (given data scatter).
a. This is unlike a gas-cap/oil-leg reservoir (Case 1) in which the oil-leg size depends on the
assumed gas-cap size.
b. This is similar to the finding in Case 1, where the MBE yielded a unique Total-original- gas-
in-place (gas-cap + solution-gas).
4. Note that it is possible to have external energy sources other than water drive (for example see
Connected Reservoirs – Case 4), in which case, a good pressure history match might also be possible
by using a connected gas reservoir model instead of an aquifer.
5. The generalized MBE (Equation 1) is equally applicable to oil and gas systems. However, for gas
reservoirs, it is much more convenient to formulate it into the p/Z or p/Z** format (Moghadam et al
2011). Note that p/Z is an alternative representation of Bg, at constant temperature.

Case 3: Oil Reservoir with Water Drive


This case represents a single well oil reservoir, underlain by a large aquifer. The well has produced 11
MMstb of oil, 4 Bscf of gas and 42 MMstb of water. Initial reservoir pressure was 4350 psi. There were some
measured reservoir pressures at yearly intervals. The oil had a bubble point pressure of 2000 psi. Therefore,
the reservoir was undersaturated (no gas cap). The volumetric calculation of OOIP was unreliable because
the well production had exceeded that estimate. This exercise attempts to use Material Balance Analysis to
determine the OOIP. The oil rate and static reservoir pressures are shown in Figure 21.
SPE-196009-MS 19

Figure 21—Production and pressure data

Using the MBE, several interpretations gave an acceptable match of the pressures, as shown in Figure
22 below. To start with, the OOIP was assumed to be approximately 60 MMstb (this was based on the
cumulative oil produced and an assumed recovery factor of approximately 20%). A Havlena-Odeh analysis
and a pressure history match were performed by varying the aquifer size and aquifer strength, until a good
fit of the data was obtained. This process was repeated using half the OOIP (30 MMstb) and then half again
(15 MMstb), and also doubling the OOIP (120 MMstb) and doubling it again (240 MMstb).
20 SPE-196009-MS

Figure 22—Acceptable Matches with OOIP ranging from 15 MMstb to 450 MMstb
SPE-196009-MS 21

It is evident from the above analyses that the OOIP could range from 15 MMstb to 240 MMstb. (In fact,
the OOIP of 450 MMstb was the maximum that could yield an acceptable match). This range of answers (a
factor of more than 30 times) is clearly not a satisfactory outcome of using the MBE.

Learnings from Case 3


This case illustrates that.
1. When the aquifer is much bigger than the oil reservoir, the range of possible OOIPs is unacceptable
(a factor of 30+).
2. The reason for this behavior is that both the oil reservoir and the huge aquifer are contributing to the
reservoir energy. The contribution from each can be traded for the other.
3. When the aquifer is 100 times bigger than the oil reservoir, a 50% reduction in the oil size is an
insignificant fraction of the total (oil + aquifer) volume, and is well within the data scatter.

Case 4: Connected Gas Reservoir


The classic MBE treats a reservoir as a sealed entity (except for aquifer support). However, given the
complex geological nature of reservoirs, there is a large number of reservoirs that consist of a primary
reservoir (good permeability) in communication with lower quality reservoirs. This communication can
happen at the wellbore, in the case of multilayers, or in the reservoir when there is crossflow between layers,
or around the edges of the reservoir in the case of composite systems (dolomitized or fractured systems).
Payne (1996) presented a simplified analysis to deal with such situations of "connected reservoirs". The
interaction can be between many inter-communicating reservoirs. However, a simple and very common
manifestation is a group of wells producing from one reservoir with good permeability, and this reservoir is
in communication with a low permeability system, as can be seen in the schematics of Figures 23a and 23b.

Figure 23a—Communicating Layers


22 SPE-196009-MS

Figure 23b—Composite System

These connected reservoirs are usually modelled in a manner similar to the way aquifers are – the energy
of the producing reservoir is supplemented by influx from the communicating reservoir. The extent of
supplementary influx is controlled by three factors:
I. The pressure difference between the producing reservoir and the connected reservoir.
II. The size of the connected reservoir.
III. And the transfer coefficient between the connected and the producing reservoirs.
For a gas reservoir the mathematics have been formulated in a manner similar to the p/Z formulation, with
Z being replaced by Z** to account for the additional gas being transferred from the connected reservoir
to the main producing reservoir, Moghadam 2011.
The following interpretations result.
I. A "volumetric" (not connected) gas reservoir, is represented by a straight line (Figure 24a) when p/
Z is plotted versus cumulative gas production (Gp).
II. For a "connected" reservoir, the p/Z line is curved (Figure 24b), representing the external influx of
gas from the connecting reservoir. If a straight line is drawn through the early data, the OGIP of
the producing reservoir can be estimated (at that time, there has been little contribution from the
connecting reservoir).
III. For a "connected" reservoir, the p/Z** line is straight (Figure 24c), accounting for the external influx
of gas from the connecting reservoir.
SPE-196009-MS 23

Figure 24a—"Volumetric reservoir – straight p/Z line

Figure 24b—"Connected" reservoir – curved p/Z line

Figure 24c—"Connected" reservoir – straight p/Z** line

The case presented below is that of a gas reservoir, Sundown Cadotte A pool, in North Eastern British
Columbia, Canada. There are five producing wells, and their production and static pressure history is shown
in Figure 25. These wells have produced dry gas with minimal water (2 stb/MMscf) for 30 years.
24 SPE-196009-MS

Figure 25—Production and Pressure Data for Cadotte A pool

The data is plotted on a p/Z plot in Figure 26. The curvature of the data is classically associated with
a water drive. If this were true, it would have to be a weak water drive because the reservoir pressure has
depleted significantly. If it were a water drive reservoir, the straight line through the initial data would be
representative of the size of the gas reservoir (negligible water encroachment at that time). This is shown in
Figure 26. The estimate of the OGIP is 47 Bscf. However, these wells have produced 69 Bscf, much more
than the estimated OGIP. Additionally, the only water produced is water of condensation. It is concluded
that the curvature of the p/Z plot is not due to aquifer support.
To interpret the curvature of the data, a "connected reservoir" model was used. The producing reservoir
was 47 Bscf. And the communicating (low permeability) reservoir was 24 Bscf. The p/Z** plot is shown
in Figure 27 and the corresponding pressure-history-match in Figure 28. The excellent history-match of the
pressures and the total gas produced (69 Bscf) compared to the total OGIP (47+24=71 Bscf) is consistent
with the low (approximately 100 psi) reservoir pressure at the end of production.

Figure 26—p/Z straight line through early data shows the size of the gas reservoir to be approximately 47 Bscf
SPE-196009-MS 25

Figure 27—p/Z** straight line (all data) using "connected gas reservoir" model; Producing
Reservoir = 47 Bscf, Supporting Reservoir = 24 Bscf, Transfer Coefficient = 8.4 Mscfd/psi

Figure 28—Pressure-history-match using "connected gas reservoir" model

Learnings from Case 4


This case illustrates that.
1. When the p/Z data is curving away from the initial straight line, the reason could be either aquifer
support or a connected reservoir.
2. The p/Z straight line through the early data is reasonably close to (usually a little smaller than) the
OGIP in the producing reservoir.
3. If the quantity of gas produced is more than 50% of the indicated OGIP (from the initial straight line),
it is not likely aquifer support. If the produced gas exceeds 100% of the initial straight line, then it
must be a "connected reservoir".
4. The combination of the size of the connecting reservoir and the transfer coefficient may not be unique.
To avoid overstating reserves, it is recommended that the size of the connecting reservoir be kept to
a minimum consistent with the measured pressure data.
26 SPE-196009-MS

5. When doing a p/Z** analysis it is invaluable to look not only at the p/Z** plot but also at the pressure-
history-match plot. This plot is the ultimate verification of the interpretation.

Case 5: Single Phase Oil Reservoir (no gas cap, no aquifer)


The cases discussed so far have highlighted the complexities and non-uniqueness of the MBE. The following
case illustrates that in certain situations such as an undersaturated oil reservoir with no aquifer or connected
reservoirs, the MBE does yield a unique answer.
The Sinclair Paddy L pool, is located in Alberta, Canada. During the pressure-time portion of this work,
we excluded 2 wells from that pool designation, based on the pressure trends observed on the pressure-
time plot. Now, the pool grouping consists of 4 oil producers and a water injector, and its production, water
injection and pressure history are shown in Figure 29. The oil is 38 API, with an initial pressure of 1640
psi and a bubble point pressure of 1540 psi. The initial solution gas ratio was 400 scf/stb but the producing
GOR averaged 800 scf/stb, which is consistent with this reservoir being only slightly undersaturated.

Figure 29—Production/Injection and Pressure History of Paddy L Pool

The Generalized Havlena-Odeh plot is shown in Figure 30. Because of the nature of the plotting variables
in this plot, the noise is exaggerated, and it shows that a range of answers is possible. However, combining
this analysis with a pressure-time history-match or a specialized Havlena-Odeh plot of (F-We) vs (Et),
results in a totally unique interpretation of 16.5 MMstb, as shown in Figures 31a and 31b. Since for this
simple reservoir system, there is only 1 variable (N) to be varied, a unique match is expected and, indeed,
is obtained.
SPE-196009-MS 27

Figure 30—Generalized Havlena-Odeh plot showing range of answers

Figure 31a—Havlena-Odeh Plot of Paddy L oil reservoir (16.5 MMstb)

Figure 31b—Pressure-History-Match of Paddy L oil reservoir (16.5 MMstb)

Learnings from Case 5


This case illustrates that.
1. For single phase reservoirs, with no aquifer or connected reservoirs, the MBE does result in unique
answers of OOIP (N) or OGIP (G).
28 SPE-196009-MS

2. The Generalized Havlena-Odeh plot often shows multiple possible interpretations, which are readily
resolved by a specialized Havlena-Odeh plot, and ultimately by a time-based pressure- history-match.

Discussion
The traditional Material Balance Equations have not been altered. Conventional material balance analysis
consists of some esoteric combination of variables that yields a straight line. The method recommended
in this paper shifts the focus from the concept of straight lines to the concept of history-matching the
reported pressures as a function of time. This plot is NOT a straight line and reflects the variability of
production operations. The required answer for hydrocarbon-in-place is not reflected in a straight line but
in the goodness of fit of the history-match.
When the "pressure-time" plot is made the focus of the analysis, it not only significantly enhances the
interpretation, but it brings with it several advantages: a) it often helps illustrate that some of the wells in
the well grouping do NOT belong to the same reservoir and should be excluded; b) it highlights the data
quality and validity.
In conventional material balance of an oil reservoir the size of the gas cap (m or G) is a variable embedded
within the plotting parameters, and being out-of-sight, is often out-of-mind. In contrast, the time-based
analysis presented in this paper, puts the choice of "m or G" front-and-center, and the consequences of that
choice are displayed in the history-match.
While there is no new theoretical development in this paper, the contribution of this work is the application
of this Material Balance Analysis in a practical way, which greatly enhances its utility, and presents to
engineers, in a direct and tangible manner, the implications of various interpretations.
We gave examples of wells that had been originally grouped together, but that do NOT, in fact, belong in
the same reservoir. We presented examples of data which, when using the conventional pressure- cumulative
plots, would appear as data scatter. However, using the time-enhanced material balance analysis showed
that the "apparent" data scatter was, in fact, a manifestation of pressure support from an external energy
source (tight gas or water drive).
Not only can practising engineers determine hydrocarbon-in-place, but they can, using the pressure-time
plot, validate the data, and easily evaluate the relevance or uniqueness of various assumptions.

Conclusion
1. When conducting a material balance analysis, the first thing is to group all the wells that belong to the
same reservoir, and exclude all others. Well grouping is done by comparing static pressures on a time
basis and best done when pressure data is plotted along with production and injection data versus time.
2. The MBE is rigorous, but due to the very nature of the equation, number of unknowns and the quality
of data, it has significant potential for errors and does NOT yield a unique answer.
3. Interpretation of the Havlena-Odeh plot is greatly aided when the analysis is supplemented with a
history-match on a pressure-time plot. However, sometimes, in spite of having relatively good and
plentiful pressure data, the MBE results can be non-unique, even when the pressure-time history-
matched plot is used.
4. A pressure-time plot helps identify wells that have an external energy source, such as an active aquifer.
A tell-tale sign of an external energy source (water drive or connected reservoir) is the continuously
increasing static pressures when there are long periods of no production. If the external energy source
is not identified, the resulting Material Balance Analysis can be significantly wrong.
5. In every situation, a pressure-time history-match plot is the ultimate validation of any analysis, and
returns the analyst from the mathematical world to the real world of measurements.
SPE-196009-MS 29

Nomenclature
Bg = Gas formation volume factor, res bbl/scf
Bgi = Gas formation volume factor at pi, res bbl/scf
Big = Gas formation volume factor of injected gas, res bbl/scf
Bo = Oil formation volume factor, dimensionless
Boi = Oil formation volume factor at pi, dimensionless
Bt = Total formation volume factor, dimensionless
Bti = Total formation volume factor at pi, dimensionless
Bw = Water formation volume factor, dimensionless
cf = Formation (rock) compressibility, volume/volume/psi
cw = Water compressibility, volume/volume/psi
Efw = c f + c w S w 1− s w Δp
Eg = Bg − Bgi
Eo = Bo − Boi + Bg(Rsi − Rs)
Et = E o + B oi B gi m E g + B oi ( 1+m ) E fw
F = Np[Bo + Bg(Rp − Rs)] + WpBw − WiBiw − GiBig
G = Initial gas in place in reservoir, scf
Gi = Cumulative gas injected, scf
m = (GBgi)/(NBoi)
N = Initial oil in place in reservoir, stb
Np = Cumulative oil produced, stb
p = Pressure, psi
Δp = Pressure drawdown, psi
Pi = Initial pressure, psi
Rp = Cumulative gas-oil ratio, scf/stb
Rs = Solution gas-oil ratio, scf/stb
Rsi = Initial solution gas-oil ratio, scf/stb
Sw = Water saturation, dimensionless
We = Cumulative water influx (encroachment), res bbl
Wi = Cumulative water injected, res bbl
Wp = Cumulative water produced, res bbl
Z = Gas deviation factor, dimensionless

References
Carlson, M. R. (1997, November). Tips, Tricks and Traps of Material Balance Calculations. Petroleum Society of Canada.
doi: 10.2118/97-11-04
Gilicz, A. (1991, January). The Material Balance as a Straight Line - an Aid to Tehrani's Method. Society of Petroleum
Engineers, 21891-MS SPE General - 1991
Havlena, D., & Odeh, A. S. (1963, August). The Material Balance as an Equation of a Straight Line. Society of Petroleum
Engineers. doi: 10.2118/559-PA
Havlena, D., & Odeh, A. S. (1964, July). The Material Balance as an Equation of a Straight Line-Part II, Field Cases.
Society of Petroleum Engineers. doi: 10.2118/869-PA
Payne, D. A. (1996, November). Material-Balance Calculations in Tight-Gas Reservoirs: The Pitfalls of p/Z Plots and a
More Accurate Technique. Society of Petroleum Engineers. doi: 10.2118/36702-PA
Moghadam, S., Jeje, O., & Mattar, L., (2011, January). Advanced Gas Material Balance in Simplified Format. Society of
Petroleum Engineers. doi: 10.2118/139428-PA
Tehrani, D. H. (1985, September). An Analysis of a Volumetric Balance Equation for Calculation of Oil in Place and
Water Influx. Society of Petroleum Engineers. doi: 10.2118/12894-PA

You might also like