SPE-183813-MS

Method of Water Influx Identification and Prediction for a Fractured-Vuggy

Carbonate Reservoir

Li Yong, Research Institute of Petroleum Exploration and Development PetroChina; Jia Chunxia, Shanghai Normal
University; Peng Hui and Li Baozhu, Research Institute of Petroleum Exploration and Development PetroChina;
Liu Zhiliang, Tarim Oilfield Company PetroChina; Wang Qi, Research Institute of Petroleum Exploration and
Development PetroChina

Copyright 2017, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Middle East Oil & Gas Show and Conference held in Manama, Kingdom of Bahrain, 6-9 March 2017.

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
Naturally fractured-vuggy carbonate reservoirs in China have some distinctive characteristics, which
reservoir is a discontinuum and isolated developed. And these reservoir usually have strong aquifer support.
So one of the most important things for these reservoirs is to avoid too early water breakthrough, because
water breakthrough explicitly reduce oil productivity and recovery. Therefore, how to identify aquifer
influx in advance are important in order to guide well production rate optimization and improve reservoir
performance.
This paper proposes four diagnostic curves to identify aquifer influx phase for producers with aquifer
support, which are mainly based on well daily production and pressure data. And whole production period
of a producer can be classified into 3 phases based on the curves: no aquifer influx phase, early aquifer
influx phase, and middle-late aquifer influx phase. The methods are validated by reservoir simulation and
well performance data. Then aquifer influx phases of all wells in a fractured-caved carbonate reservoir are
classified, and a correlation for well oil cumulative production prediction is established.
This new method has been applied to K fractured-vuggy carbonate reservoir in China. The mid-depth
of K reservoir is 6000m, and current oil recovery is only around 7% of OOIP with primary depletion and
more than 100 producers. Around one third of producers have been water flooded, most of which water
cut increased from 0% to more than 60% within only one month. The four proposed diagnostic curves for
aquifer influx identification are tubing pressure drop vs. cumulative oil production curve, flowing material
balance curve, Blasingame type curve, and dimensionless expansion production ratio curve. And how to
use diagnostic curves to identify aquifer influx are detailed illustrated based on the reservoir simulation
data and actual well production data. Then production phases of all wells are identified and classified.
Furthermore, based on performance of water flooded wells, correlation between oil cumulative production
and dimensionless expansion production ratio are established, which are used for well performance
prediction.
This paper offers a new method and a case study of aquifer influx identification and prediction for aquifer
drive carbonate reservoirs. It also provides a methodology and reference case for other similar oil fields.
2 SPE-183813-MS

Introduction
Aquifer or water influx is an important issue which effects the performance of oil and gas reservoirs.
Ultimate recovery from oil and gas reservoirs is usually affected by the aquifer influx either from the
perimeter of the reservoir or below and from both (Biu, V. T., Biu, E., Buduka, S., etc., 2011). The researches
usually focused on the water breakthrough time in oil reservoirs. And these literatures mainly studied on
water coning breakthrough time calculation formula which are based on simplified conceptual models.
Methods of modeling water influx include the Hurst modified steady-state method, and various unsteady
state methods such as those of Van Everdingen-Hurst, and Carter-Tracy, etc. (Schilthius, R. J., 1936; Van
Everdingen, A.F. and HURST, W., 1953; Van Everdingen, A.F., TIMMERMAN, E.H. and MCMAHON,
J.J., 1953; HURST, W., 1958; Carter, R.D., and Tracy, G.W., 1960; Fetkovich, M.J., 1971) More recently,
the four methods introduced by Leung for finite and infinite aquifers provide relatively simple models to use
with substantial improvements in accuracy and efficiency over the previous methods (LEUNG, W.F., 1986;
Lies, H. K., 2000). Due to the inherent uncertainties in the aquifer characteristics, all the models require
historical reservoir performance data to evaluate aquifer property parameters. The fact that the reservoir-
aquifer boundary pressure is not constant with time gives rise to computational challenges (Omeke, J. E.,
Nwachukwu, A., Awo, R. O., et. al, 2011). While recently engineers are still interested in these kinds of
methods (Shimada, M., & Yildiz, T., 2009). Some predictions on water breakthrough time were made by
using material balance analysis and water invasion indicative curves. Recently, rate transient analysis (RTA)
gets rapid development in the studies of reserves evaluation and reservoir characterization (Kuo et al, 1983;
Lee and Tung, 1990; Blasingame et al, 1991; Mattar and McNeil, 1998; Li et al, 2016). Using RTA in water
invasion prediction is still in its infancy (Agarwal et al, 1998; Denney, 2005; Iik et al, 2010; Li et al, 2009,
2010). And Li Y. et. al. provides a new method with 3 diagnostic curves to identify aquifer influx status for
gas wells, which are mainly based on well production and pressure data. And the whole production period
of gas wells can be classified into 3 periods based on the diagnostic curves: no aquifer influx period, early
aquifer influx period, middle-late aquifer influx period (Yong, L., Baozhu, L., Jing, X., et. al, 2015). These
methods are useful for gas wells. And this paper is try to establish the related methods for the oil producers
in reservoirs with aquifer support.

Background of K reservoir
The fractured-vuggy or fractured-caved carbonate reservoir in China is completely different to the carbonate
reservoirs in the other countries. And most fractured-vuggy carbonate reservoirs in China is located in
Tarim Basin. And most reservoirs are in Ordovician, which have deep buried depth. The K fractured-
vuggy carbonate reservoir is one of the typical paleokarst carbonate reservoir in Tarim Basin. K reservoir
is developed and impacted by multi-tectonic movement, multiple karst superimposed transformation, and
multiple reservoir forming process. There are multi-scaled vugs, fractures and caves developed in K
reservoir, and the whole reservoir connectivity is completely poor, which makes lots of isolated flow units
with separated water-oil contact. There is no communication among different isolated flow units. As figure 1
shows, the buried depth of K reservoir is around 6000m, and several different isolated flow units developed
in the profile which is proved by well drilling and production performance. It can be seen that the oil-water
contacts are different for different flow units, furthermore, the gas oil ratio are also different. Most units in
K reservoir have weak to strong aquifer support (Figure 1). And because of the developed high permeability
channel- fractures and caves in K reservoir with the aquifer support, water breakthrough happened and water
cut sharp increased to more than 60% for lots of wells with less than a year production. Figure 2 shows one
of the typical well production history in K reservoir, which only has less than one month dry oil production
and water cut increased to 100% within less than 10 days. So early water breakthrough and rapid water cut
increasing have been one of the most problem in K reservoir, and average annual production decline rate
is more than 25%. So one of the most important things for these kind of reservoirs is to avoid early water
SPE-183813-MS 3

breakthrough, because water breakthrough explicitly reduce oil productivity and recovery. And because of
the isolated characteristic and poor connectivity among different flow units, the aquifer influx period and
water breakthrough could be identified and predicted for all wells in K reservoir. After identify aquifer
influx in advance, well production rate could be optimized and reservoir performance could be improved.

Figure 1—Conceptual geological profile of typical reservoir flow units in K reservoir

Figure 2—Typical well production history of K reservoir

Four Diagnostic Curves of Water Influx Identification

Water Influx Understanding from Reservoir Simulation Results
In order to understand the water or aquifer influx characteristics and build corresponding identification
curves, reservoir simulation is firstly used. Reservoir simulation is forward modeling and simulation, so
we can know the details and reasons of the simulation results, which is representative to the subsurface
reservoir. Different reservoir simulation models are built, which includes no aquifer support models, edge
aquifer models with different aquifer size, and bottom aquifer models with different aquifer size.
4 SPE-183813-MS

Based on our understanding and study from the well performance in K reservoirs, 4 curves are used for
aquifer influx identification. These curves are tubing pressure drop vs. cumulative oil production curve,
flowing material balance curve, Blasingame type curve, and dimensionless expansion production ratio
curve. The tubing pressure drop is the difference between initial tubing pressure and current tubing pressure,
which is dp=pt_initial-pt_current. The equation of flowing material balance is as follows:

(1)

Where q is the oil production rate, m3/d; Δp is flowing pressure difference, MPa; Bo, Boi is oil formation
factor, dimensionless; Np is cumulative production, m3; ct is the total compressibility, MPa-1; N is the original
oil in place, m3. And the flowing material balance curve is the relationship between oil normalized rate

and normalized oil cumulative production . For the reservoir with primary production and

without any aquifer support, the flowing material balance should be a straight line. The Blasingame type
curve is usually used for original oil in place calculation, and the type curves is the relationship between
dimensionless normalized rate and dimensionless time. And the well normalized rate and material

balance time tc is calculated based on well production data. Through matching well data and
the type curves, reservoir parameters could be calculated, which include original oil in place, permeability,
skin factor, etc. For the dimensionless expansion production ratio (Npr) curve, the ratio is shown in equation
(2):

(2)

Value of dimensionless expansion production ratio could stands for the aquifer size. For convenience,
the flowing bottomhole pressure drop in equation (2) could use tubing pressure drop for approximation.
Figure 3 shows two models without any aquifer support. One model is a box model without any structure
difference for the same depth. The second model is a reservoir with anticlinal structure.

Figure 3—Simulation models without aquifer support

Figure 4 shows the corresponding 4 curves based on the pressure data and production rate from simulation
results of the second model. And there is no characteristic difference of the 4 curves between the two
models. It can be seen that the first two curves (tubing pressure drop vs. cumulative oil production curve
and flowing material balance curve) all show straight line for the reservoir without any aquifer support.
SPE-183813-MS 5

And the simulated well data obey one of the type curves for Blasingame type curve method. While for the
dimensionless expansion production ratio curve, dimensionless expansion production ratio firstly rapidly
increase from 0 to 2000, and then its increasing trend slow down as shown in the blue rectangular.

Figure 4—Characteristics of the 4 curves for reservoir without aquifer support

Similarly, corresponding simulation models with different size of edge or bottom aquifer support are built
and simulated. Figure 5 shows the simulation models with aquifer support.

Figure 5—Simulation models with aquifer support

Figure 6 shows the corresponding 4 curves based on the pressure data and production rate from simulation
results of the second model. It can be seen that the characteristic of the four curves are completely different
from that of reservoir without any aquifer support. For the tubing pressure drop vs. cumulative oil production
6 SPE-183813-MS

curve, it shows three line trends (For actual well production data, it could be three different trend curves):
first line with short period (red solid line), and the second (thin dashed red line) and third line (heavy dashed
red line) show relative long period. The first line shows the production characteristic of the oil bearing area.
After pressure drops to some extent, aquifer enters the reservoir, and reservoir has aquifer support with
pressure dropping slower than the first period. So tubing pressure drops slower than that of the first period.
While for the third line which represent for the third period, aquifer enters most of the reservoir and well is
going to water breakthrough, or well has been water breakthrough and continues producing water. Because
of water breakthrough and the two phase of fluid flow in the reservoir, flow resistance increase, which means
the production pressure difference increase. So tubing pressure drops faster than that of the second period.
It can be seen that flowing material balance curve shows the similar characteristic of the tubing pressure
drop vs. cumulative oil production curve, which also shows three different trends. While for the Blasingame
type curve, the first period production data also obey the type curves. Then after aquifer enters the reservoir,
production data deviate from type curves to the upside. And just before or after water breakthrough, the
production data rapidly drops to the downside. For the dimensionless expansion production ratio curve, it
seem that it just increases all the time. Actually, the first period (blue line) shows rapid increase then is
drops for a short period. Then it increased for the second period (black solid line). And the slope of the third
period (black dashed line) decreases compared to the second period (black solid line). So dimensionless
expansion production ratio curve also shows three periods although it seems not as clear as the other curves.

Figure 6—Characteristics of the 4 curves for reservoir with aquifer support of 400 time of HCPV

Furthermore, performance of reservoir with different aquifer size support are also simulated. Figure 7
shows the comparison among reservoirs with different aquifer size (Size0: 10 times of HCPV; Size1: 19
SPE-183813-MS 7

times of HCPV; Size2: 26 times of HCPV; Size3: 100 times of HCPV; Size4: 250 times of HCPV; Size5: 400
times of HCPV) for flowing material balance curve and dimensionless expansion production ratio curve. It
can be seen that the bigger aquifer size, the bigger change of the second and third period curves happens
compared to that of the first period. For flowing material balance curve, there is little difference among
curves of size3, size4 and size5, which means flowing material balance curve is not sensitive to aquifer
size when aquifer size is bigger than 100 times of HCPV. Actually it similar to the first and beginning of
the second period of dimensionless expansion production ratio curve. As the enlarged view shows, there
is also little difference among curves of size3, size4 and size5. And there is also little difference among
dimensionless expansion production ratio curve of size3, size4 and size5 for the first 4 years, while the
difference becomes bigger after 4 years production. Furthermore, there is no aquifer size larger than 100
times of HCPV under current understanding.

Figure 7—Curve characteristics of two curves for reservoir with different aquifer size

Water Influx Identification Curves

Based on the simulation results of different reservoir models and the characteristic analysis of the four
curves, it can be seen that the whole production history of a well drilled in a reservoir with aquifer support
could be divided into three phases: no aquifer influx period, early aquifer influx period and middle-late
aquifer influx period.
For no aquifer influx period, the well drainage area does not touch the aquifer, and no aquifer enters the
reservoir. So at this period, the four curves are similar to the curve characteristics of simulation results of the
reservoir without aquifer support (figure 8). Both tubing pressure drop vs. cumulative oil production curve
and flowing material balance curve show a straight line. The production data obey one of the Blasingame
type curves, and dimensionless expansion production ratio curve is approximately shown as a horizontal
line.
For early aquifer influx period, the well drainage area reach the aquifer, and aquifer enters the reservoir
for pressure support. During this period, the drop speed of bottomhole flowing pressure or tubing pressure
is usually slower than that of no aquifer influx period because of aquifer support. So during this period, the
four curves are much different from the curve characteristics of simulation results of the reservoir without
aquifer support. All the four curves show similar trend, which all deviate from curve trend of no aquifer
influx period to the upside as shown in figure 8 (For Blasingame curve, it deviates from the line of boundary
dominated period to the upside).
For middle-late aquifer influx period, aquifer enters a lot of areas of the reservoir. And the well maybe
waterflooded or not. So this period include a short period before water breakthrough and the whole period
after well water breakthrough. During this period, the fluid flow resistivity become larger than before
because of two phase flow around the bottomhole, and production pressure difference become larger than
before. So the drop speed of bottomhole flowing pressure or tubing pressure is usually faster than that of
8 SPE-183813-MS

early aquifer influx period. During this period, all the four curves also show similar trend, which all deviate
from curve trend of early aquifer influx period to the downside (For dimensionless expansion production
ratio curve, it deviates from fast increasing trend to slower increasing trend or maybe decreasing trend).

Figure 8—Four curves for water influx identification

These four identification curves have the similar function of each other. They could be used separately.
While if these four methods collaborated and constrained with each other, more accurate results could be
got. When engineers use the four curves together, engineers can go back to check your analysis after you
found results of some curve are different from that of the others. Different answer may give engineers some
inspiration or insight to the well performance, which help the engineers to reach the final solution. And
because bottomhole flowing pressure is used in Blasingame type curve and flowing material balance curve
while tubing pressure is used in tubing pressure drop vs. cumulative oil production curve and dimensionless
expansion production ratio curve, so the analysis results of Blasingame type curve and flowing material
balance curve could be much reliable than that of tubing pressure drop vs. cumulative oil production curve
and dimensionless expansion production ratio curve.

Water Influx Identification for K Fractured-vuggy Carbonate Reservoir

So based on the proposed four identification curves and production data of all wells in K fractured-vuggy
carbonate reservoir, the aquifer influx status of all wells could be classified and identified. Figure 9 shows
the identification results of a typical well in K reservoir. The well is currently in the early aquifer influx
period, and it has been produced for a certain period in the early aquifer influx period. And based on tubing
pressure drop vs. cumulative oil production curve and dimensionless expansion production ratio curve, it
seems this well has entered the middle-late aquifer influx period, which contradict to that of Blasingame
SPE-183813-MS 9

type curve and flowing material balance curve. And the final results are based on the results of Blasingame
type curve and flowing material balance curve. But the well is really going to enter middle-late aquifer
period very soon, which means that this well will be water breakthrough soon.

Figure 9—Four curves for water influx identification

The well whole production history could be classified based on the proposed four curves. If a well is still
producing in no aquifer influx period, there is no worry about water breakthrough. So the well could produce
at a relative high rate. While if a producer is in early aquifer period and has produced a certain amount of
production in this period, or a producer is in the middle-later period and is still not water breakthrough,
the producer should try to avoid water breakthrough as soon as possible, which should try to decrease its
production rate. Figure 10 shows the identification results, which mainly focus on the wells in early aquifer
influx period. These wells are the water breakthrough risk wells. So the production rate of these well should
be maintained or decrease but not increase any more for any reason.
10 SPE-183813-MS

Figure 10—Well location map of water breakthrough risk wells based on aquifer influx identification

Well Performance Prediction for Water Influx Wells

From the above, aquifer influx status could be identified. But how could we predict the performance
of producers with aquifer support. Several relationships are tested and the Npr@strart of early aquifer
influx period Vs. Cumulative oil production@strart of early aquifer influx, Npr@water breakthrough Vs.
Cumulative oil production@ water breakthrough and Initial Npr Vs. Cumulative oil production@strart
of early aquifer influx finally. Figure 11 shows the relationships. So if we can predict the Npr then the
cumulative oil production could be predicted. From equation (2) we know Npr is related with cumulative oil
production. It seems it is hard to predict Npr. If Npr could be predicted, cumulative oil production could also be
predicted from equation (2). But the relationship between Initial Npr Vs. Cumulative oil production@strart of
early aquifer influx period could be used for cumulative oil production prediction although the relationship
seems not as accurate as the other two. Based on this relationship, the initial Npr could be calculated, and
then the cumulative oil production at start of early aquifer influx period could be predicted.
SPE-183813-MS 11

Figure 11—Relationship between Npr and cumulative oil production

Conclusions
Whole production period of a producer in a fractured-vuggy carbonate reservoir with aquifer support can
be classified into three phases based on the curves: no aquifer influx phase, early aquifer influx phase, and
middle-late aquifer influx phase.
Four diagnostic curves are proposed to identify aquifer influx phase for producers with aquifer support,
which are tubing pressure drop vs. cumulative oil production curve, flowing material balance curve,
Blasingame type curve, and dimensionless expansion production ratio curve. Through reservoir simulation,
the curves characteristics are analyzed in detail. And all these curves have the function of identify the three
phases of the producers.
Production phases of all wells in K fractured-vuggy carbonate reservoir are identified and classified,
and water breakthrough risk wells are identified. Correlation between oil cumulative production and
dimensionless expansion production ratio are established, which are used for well performance prediction.

Reference
Biu, V. T., Biu, E., Buduka, S., & Prince, W. (2011, January 1). Pressure Redistribution and Aquifer Influx Modelling
Using Statistical Models, Direct Declining, Sum of Time Step, and Log-Determinant Approach. Society of Petroleum
Engineers. 10.2118/150799-MS.
Schilthius, R. J. Active Oil and Reservoir Energy[J]. Transactions of the AIME, 1936, 118(1):33-52.
HURST, W.: "Water Influx into a Reservoir and its Application to the Equation of Volumetric Balance", Trans., AIME
(1943) 151, 57.
Van Everdingen, A.F. and HURST, W.: "The Application of the Laplace Transformation to Flow Problems in Reservoirs",
Trans., AIME (1949) 186, 305.
12 SPE-183813-MS

Van Everdingen, A.F., TIMMERMAN, E.H. and MCMAHON, J.J.: "Application of the Material Balance Equation to a
Partial Water-Drive Reservoir", Trans., AIME (1953) 198, 51.
HURST, W.: "The Simplification of the Material Balance Formulas by the Laplace Transformation", Trans., AIME (1958)
213, 292.
Carter, R.D., and Tracy, G.W. An Improved Method for Calculatiing Water Influx. J. Pet. Tech., 58-60, Dec. 1960.
Fetkovich, M. J. (1971, July 1). A Simplified Approach to Water Influx Calculations-Finite Aquifer Systems. Society of
Petroleum Engineers. 10.2118/2603-PA.
LEUNG, W.F.: "A New Pseudosteady-State Model for Dual-Porosity/Dual-Permeability Aquifers and Two Interconnected
Single-Porosity Aquifers", SPE Res. Eng. (September 1986) 511.
Lies, H. K. (2000, January 1). Aquifer Influx Modelling for Gas Reservoirs. Petroleum Society of Canada.
10.2118/2000-029.
Omeke, J. E., Nwachukwu, A., Awo, R. O., Boniface, O., & Uche, I. N. (2011, January 1). A New Approach to Aquifer
Influx Calculation for Finite Aquifer System. Society of Petroleum Engineers. 10.2118/150733-MS.
Shimada, M., & Yildiz, T. (2009, January 1). Predicting Water Influx from Common Aquifers. Society of Petroleum
Engineers. 10.2118/120897-MS.
Kuo M.C.T, Occidental E, P Co. A simplified method for water coning predictions. Proceedings of the SPE Annual
Technical Conference and Exhibition, 5-8 October 1983 held in San Francisco, California. SPE 12067.
Lee S.H., Tung, W.B. General coning correlations based on mechanistic studies. Proceedings of the SPE Annual Technical
Conference and Exhibition, 23-26 September 1990 held in New Orleans, Louisiana. SPE 20742.
Blasingame, T. A. McCray, T.L., Lee, W.J. Decline Curve Analysis for Variable Pressure Drop/Variable Flowrate Systems.
Proceedings of the SPE Gas Technology Symposium, 22-24 January 1991 held in Houston, Texas. SPE 21513.
Mattar L., McNeil, R.. The Flowing Gas Material Balance. JCPT, 1998, 37(2):37-42.
Li Y, Li B.Z., Xia J., et al New Method of Aquifer Influx Status Classification for single well in Gas Reservoir with
Aquifer Support. Natural Gas Geoscience, 2015, 31(6): 94-99.
Agarwal, R.G., Gardner, D.C, Kleinsteiber, S.W et al. Analyzing Well Production Data Using Combined Type Curve and
Decline Curve Concepts. Journal of Petroluem Technology, 1998, 50(10): 1-2.
Denney D. Practical Diagnostics Using Production Data and Flowing Pressures. Journal of Petroleum Technology, 2005,
57(3): 1-3.
Iik D., Anderson D.M., Stotts G.W.J., et al Production Data Analysis—Challenges, Pitfalls, Diagnostics. SPE Reservoir
Evaluation & Engineering, 2010, 13(3): 1-15.
Li, Y, Li, B Z, Hu, Y L, et al. Water production analysis and reservoir simulation research in the Jilake gas condensate
reservoir. Petroleum Exploration and Development. 2010. 37(1): 89-93.
Yong, L., Baozhu, L., Jing, X., Jing, Z., Daigang, W., Zhongqian, Z., & Xiangjiao, X. (2015, November 24). New Method
of Water Influx Identification and Ranking for a Super-Giant Aquifer Drive Gas Reservoir. Society of Petroleum
Engineers. 10.2118/178059-MS.