Vous êtes sur la page 1sur 7

SPE 106305

Production Data Analysis for Commingled Multilayer Gas ReservoirsGraphical Aides


for Flow Regime Identification and History Matching
John P. Spivey, SPE, Phoenix Reservoir Engineering

Copyright 2007, Society of Petroleum Engineers

This paper was prepared for presentation at the 2007 SPE Hydraulic Fracturing Technology
Conference held in College Station, Texas, U.S.A., 2931 January 2007.

This paper was selected for presentation by an SPE Program Committee following review of qwt = q1 + q2 + q3 + q4
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper qw4 = q1 + q2 + q3 + q4
for commercial purposes without the written consent of the Society of Petroleum Engineers is Layer 4 q4
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 where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

qw3 = q1 + q2 + q3
q3
Abstract Layer 3
Most of the major unconventional gas plays in the U.S. have
multiple productive intervals. Wells are completed with
multiple fracture treatments, and are produced commingled.
q2
An earlier paper1 described a single-well, multilayer, Layer 2 qw2 = q1 + q2

analytical reservoir simulator* for estimating individual layer


properties in multilayer unconventional gas reservoirs by
automatic history matching production and spinner survey
Layer 1 q1 qw1 = q1
data simultaneously. The present paper describes several
interactive graphical aides that greatly facilitate the history
matching process when used in conjunction with the
analytical, multilayer reservoir simulator. These graphical
tools include an allocation graph and layer and cumulative
production log graphs. These graphs allow the engineer to
quickly review the quality of the match at any point during the Fig. 1Gas well in a multilayer unconventional gas reservoir with
commingled production.
history matching process.
Use of these graphical tools has improved the history
In single-layer unconventional gas reservoirs, production
matching workflow to the point that a complete history match
data analysis has been one of the most useful tools for
of a well with 20 frac stages, two years of daily production,
estimation of reservoir and fracture properties. Single-layer
and three spinner surveys can be performed in as little as four
production data analysis methods include type curve matching
to six hours, from construction of the initial data set through
using constant pressure type curves2-5, type curve matching
completion of the final match.
using constant rate type curves6-9, and automatic history
matching using an single-layer analytical reservoir
Introduction
simulator10-13.
Wells in low permeability and other unconventional gas
To estimate individual layer properties for wells with
reservoirs are often completed by commingling production
commingled completions, production log (spinner survey) data
from multiple intervals, as shown in Fig. 1. In many cases,
must be incorporated into the production data analysis
intervals are too far apart vertically to be produced by a single
workflow to separate the contributions of the individual layers.
hydraulic fracture, so multiple fracture treatments must be
One method that has been proposed is to allocate production
used. Estimating reservoir and fracture properties for
to individual layers by interpolation using the spinner survey
individual layers in such reservoirs is crucial to evaluating
data, then analyze the resulting reconstructed individual layer
fracture treatment effectiveness, optimizing future stimulation
rate histories using single-layer methods14-16. Unfortunately, if
treatments, and forecasting future performance.
there is a strong contrast in layer properties, the allocation
method will not accurately reproduce the individual layer flow
*
Patent pending. rates, thus leading to erroneous estimates of layer properties1.
2 SPE 106305

A much better way of estimating individual layer to honor the higher quality portions of the data while ignoring
properties in a commingled gas reservoir is to history match the poorer quality data.
production and spinner survey data simultaneously using the
fit-for-purpose, single-well analytical reservoir simulator
Allocation Graph Stage 5
described in Ref. 1. This approach has been successfully
100000
applied to a number of wells in the Pinedale Anticline Obs Tot
Obs PL
area17,18. The data set used in the Pinedale studies includes Obs PL RT
Sim
Sim RT
wells with as many as 22 frac stages and as many as five 10000

spinner surveys.
Because of the large number of parameters that must be

Rate

Mscf/D
1000
varied during the history match, successful application of the
analytical reservoir simulator requires a combination of
manual and automatic history matching. Several interactive 100

graphical tools have been found useful in the history matching


process, including an allocation graph and layer and
cumulative production log graphs. Use of these graphical 10
2000 2001 2002 2003

tools has improved the history matching workflow to the point Date
where a complete history match of a well with 15 to 20 frac Fig. 2Allocation graph for Stage 5, before matching
stages, two years of daily production, and two or three spinner production logs.
surveys can be performed in as little as four to six hours, from
construction of the initial data set through completion of the Allocation Graph
final match. The allocation graph, shown in Fig. 2, is the primary tool for
Each graph displays data for either one layer at a time or monitoring the progress of the history match during Phase 1 of
one production log at a time. The interactive feature allows the recommended workflow. The allocation graph is a
the engineer to click either of a pair of buttons to navigate semilog rate-time graph with five curves: the observed total
forward or backward through the layers or production logs. well rate, the observed and calculated layer rates for the
Displaying only one layer or production log at a time reduces current layer, and the observed and calculated wellbore rates
the clutter on the graphs. At the same time, the interactive at the top of the current layer. Since the observed layer and
navigation feature allows the engineer to quickly review the wellbore rates are obtained from the production logs, each of
quality of the match for all layers and all production logs. these two curves must be plotted as a series of discrete points.
The next section describes the recommend workflow for The observed total well rate and the calculated layer and
the history matching process. Subsequent sections describe wellbore rates are plotted as continuous curves.
the interactive graphical aides and their application in greater The observed and calculated layer rate curves display the
detail. quality of the match for a given layer (the current layer),
showing how well the simulated layer rate matches the
Recommended Workflow corresponding layer rate from the production log. The
The best history matching workflow we have found so far is as observed and calculated wellbore rate curves display the
follows: quality of the match of the wellbore rate (sum of the rates
Phase 1. Match two or three layers at a time, starting from from the current and all lower layers).
the bottom of the well and working upwards. During this
Allocation Graph
phase of the match, the production data are ignored. The Stage 5

fracture and layer properties for the first two or three layers 100000
Obs Tot
Obs PL
are varied to match the corresponding production log Obs PL RT
Sim
responses, ignoring contributions from the remaining layers. 10000
Sim RT

After a satisfactory match of the first two or three layers is


obtained, the fracture and layer properties for the next two or
Rate

Mscf/D

three layers are varied to match the production log responses 1000

for those layers. As each set of layers is matched, the


calculated layer and wellbore rate curves should match the 100

observed production log responses for the current set of layers


as well as all lower layers.
Phase 2. Fine-tune the match by varying the fracture and 10
2000 2001 2002 2003

layer properties for those layers with the highest flow rates to
Date
match both production logs and cumulative well production.
Fig. 3Allocation graph for Stage 5, after matching
If the production and spinner survey data are good quality and production logs for Stages 1 through 6.
the production log matches from Phase 1 are good, automatic
history matching can be used to estimate layer properties for
several layers simultaneously. If the data quality are poor,
manual history matching may be necessary to force the match
SPE 106305 3

Fig. 3 shows the allocation graph for Stage 5, after impossible to distinguish the observed and calculated
production logs for Stages 1 through 6 have been matched. cumulative production curves.
Note the excellent agreement between the calculated layer and
Well Cum Prod vs. Date
wellbore rate curves and the corresponding production log
6
rates. Obs
Sim

5
Allocation Graph Stage 19

100000
Obs Tot 4
Obs PL

Cum Prod
Obs PL RT
Sim

Bcf
Sim RT 3
10000

2
Rate

Mscf/D

1000
1

0
2000 2001 2002 2003
100

Date

10
Fig. 6Cumulative production graph, after fine-tuning match to
2000 2001 2002 2003
honor both production and production log data.
Date

Fig. 4Allocation graph for Stage 19, after matching Production Log Graphs
production logs for all stages. Two graphical presentations of the spinner survey data are
used in the history matching process: 1) a line graph showing
Fig. 4 shows the allocation graph for Stage 19 after all the observed and calculated wellbore flow rate as a function of
production logs have been matched. A good match on the depth, and 2) a bar graph displaying the observed and
allocation graph does not necessarily mean a good overall calculated individual layer flow rates. Fig. 7 and Fig. 8 show
match; the observed and calculated well cumulative a wellbore flow rate graph and layer flow rate graph,
production curves must also agree. respectively.
Well Cum Prod vs. Date
Cumulative production log graph
6
Obs
Sim
The wellbore flow rate production log graph may also be
5
called the cumulative production log graph to emphasize the
fact that the wellbore flow rate at any given depth is the sum
4 of the layer flow rates for all layers below that depth. On the
Cum Prod

cumulative production log graph, the curve representing the


Bcf

3
measured production log data is constructed by graphing the
2
wellbore rate from the production log as a function of depth.
The observed curve on the cumulative production log graph is
1 independent of the discretization of the reservoir into
individual layers.
0
2000 2001 2002 2003 The curve representing the simulated production log data
Date is constructed by drawing a line graph connecting points at the
bottom and top of each simulation layer, representing the
Fig. 5Cumulative production graph, after matching all
production logs, but before matching production data. wellbore flow rate at each point. Between adjacent layers, the
curve will be a vertical line indicating that no fluid enters the
Fig. 5 compares the observed and calculated cumulative wellbore between the layers. Within a given layer, the curve
production curves at the end of Phase 1 of the history is a straight line segment sloping upward to the right,
matching workflow, after matching the production logs for all reflecting the assumption that flow into the wellbore is
layers, but before including the production data in the match. uniform across the layer.
In this case, the agreement between the observed and
calculated cumulative production is quite good considering Incremental production log graph
that no attempt has been made to honor the cumulative The bar graph showing individual layer flow rates is also
production data. called the incremental production log graph. For the
Fig. 6 shows the final match after fine tuning the match to incremental production log graph, the calculated layer rates
honor both production and production log data during Phase 2 are taken directly from the simulator. To calculate the
of the recommended workflow. For this particular case, observed layer rates from the actual production log, the
automatic history matching was used to vary properties of the discretization of the reservoir into individual layers must be
five layers with the highest flow rates to obtain the final known. Ideally, the observed layer rate for each layer would
match. The final match is good enough that it is almost be obtained by subtracting the wellbore flow rate at the bottom
of layer from that at the top of the layer. In practice, the
4 SPE 106305

measured flow at the top of one layer may be different from significantly higher or lower than the observed rates (e.g.,
that at the bottom of the next higher layer. To ensure that the Layers 12, 16, and 17).
sum of the layer rates will equal the total well production rate,
12/1/2001
the observed layer rate for the bottom layer is taken as the rate
from the production log at the bottom of the second layer. The 8000

layer rates for subsequent layers are calculated by subtracting Obs


the measured production log rate at the bottom of the layer 8500 Sim

from that at the bottom of the next higher layer. In effect, any
9000
fluid entering the welbore between layers is assumed to come
from the lower layer.
9500

12/1/2001

Depth, ft
10000
8000

Obs 10500
8500 Sim

11000
9000

11500
9500
Depth, ft

12000
10000 0 1000 2000 3000 4000 5000
Rate, Mscf/D
10500
Fig. 9Cumulative production log graph after matching
11000 production logs for Layers 1 through 6.

12/1/2001
11500

20

19
Sim
12000
Obs
0 1000 2000 3000 4000 5000 6000 18

Rate, Mscf/D 17

16

15
Fig. 7Wellbore flow rate, or cumulative, production log graph
14
before matching production logs.
13

12
12/1/2001
Layer

11

20 10

19
Sim 9
Obs 8
18

17 7

16 6

15 5

14 4

13 3

12 2

1
Layer

11

10 0 200 400 600 800 1000 1200


9 Flow rate, Mscf/D
8

7 Fig. 10Incremental production log graph after matching


6 production logs for Layers 1 through 6.
5

4
Fig. 9 and Fig. 10 show the cumulative production log
3

2
graph and the incremental production log graph, respectively,
1 after matching production logs for Layers 1 through 6 during
0 200 400 600 800 1000 1200 1400 Phase 1 of the recommended workflow.
Flow rate, Mscf/D

Fig. 8Layer flow rate, or incremental, production log graph


before matching production logs.

Fig. 7 shows the cumulative production log graph, and Fig.


8 the incremental production log graph, before beginning
Phase 1 of the matching process. In this example, several of
the calculated layer rates are already very close to the
observed rates (e.g., Layers 4 and 7), while others are
SPE 106305 5

with the observed well rate, but two layers, Layers 3 and 14,
12/1/2001
have offsetting mismatches. All the other layer flow rates are
8000 in agreement with the observed values from the production
Obs log. Fig. 14 shows the incremental production log for the
8500 Sim same case. Although the two graphs display essentially the
same information, the wellbore graph for the calculated
9000
cumulative production log curve is shifted to the left between
Layers 3 and 14, reflecting the offsetting mismatches for those
9500
two layers.
Depth, ft

10000

12/1/2001
10500
8000

11000 Obs
8500 Sim
11500
9000

12000
0 1000 2000 3000 4000 5000 9500
Rate, Mscf/D

Depth, ft
10000
Fig. 11Cumulative production log graph after matching
production logs for all layers.
10500

12/1/2001 11000

20 11500
19
Sim
Obs
18
12000
17
0 1000 2000 3000 4000 5000
16
Rate, Mscf/D
15

14

13 Fig. 13Cumulative production log graph where Layers 3 and 14


12 have offsetting mismatches.
Layer

11

10

9 12/1/2001
8

7 20

19
Sim
6
Obs
5 18

4 17

3 16

2 15

1 14

13
0 200 400 600 800 1000 1200
12
Flow rate, Mscf/D
Layer

11

Fig. 12Incremental production log graph after matching 10

production logs for all layers. 9

7
Fig. 11 and Fig. 12 show the cumulative and incremental 6

production log graphs, respectively, at the end of Phase 1 of 5

the workflow, after matching all production logs. 4

2
Mismatched layer rates 1

Because the layer production log graph shows incremental 0 200 400 600 800 1000 1200
rates for each layer, mismatch in the production rate for one Flow rate, Mscf/D
layer will not affect any of the other layers. On the other Fig. 14Incremental production log graph where Layers 3 and 14
hand, a mismatch in flow rate for one layer will cause the have offsetting mismatches.
calculated wellbore production log curve to be offset from the
observed production log rate for layers above the mismatched Flow behind pipe
layer. Occasionally, a poor cement bond may allow fluid to exit
For example, Fig. 13 shows the cumulative production log the wellbore at one set of perforations, flow upward through a
graph for a case where the calculated total well rate agrees channel in or behind the cement, and reenter the wellbore at
6 SPE 106305

another set of perforations higher up the wellbore. When this Conclusions


phenomenon occurs, the measured production log will show a The following conclusions may be drawn from this study:
decrease in flow rate at the interval taking fluid, as shown in 1. The recommended workflow for history matching wells
Fig. 15 and Fig. 16. with commingled production is as follows: Phase 1.
Crossflow into a depleted zone might also cause a decrease Match the production log responses for two or three layers
in the measured production log response. However, for a at a time, beginning with the bottom layer and working
depleted layer to take a significant fraction of the wellbore upwards. During Phase 1, no attempt is made to honor the
fluid would require an average layer pressure much lower than total well production data. Phase 2. Fine-tune the match
the wellbore pressure along with an unusually high layer by adjusting the properties for those layers with the highest
permeability. flow rates to honor both spinner survey data and
cumulative production data simultaneously. If the data are
12/1/2001
high enough quality, automatic history matching can be
8000 used for the fine tuning process.
Obs 2. The allocation graph provides an excellent tool for
8500 Sim monitoring the progress during Phase 1 of the history
match workflow.
9000
3. The layer, or incremental, production log graph provides a
quick visual comparison of the quality of the match for the
9500
individual layers.
Depth, ft

10000
4. The wellbore, or cumulative, production log graph
provides a quick visual comparison of the quality of the
10500
overall match for all layers. A mismatch in one layer will
shift the calculated production log curve left or right of the
11000 measured production log.
5. In a low-permeability reservoir, a significant drop in
11500 wellbore flow rate is more likely to be caused by flow
behind pipe that by crossflow into a depleted interval.
12000
0 1000 2000 3000 4000 5000
6. Using the graphical history matching aides presented in
Rate, Mscf/D
this paper in conjunction with the recommended workflow
and the multilayer analytical simulator presented in a
Fig. 15Cumulative production log graph where there is a previous paper, a history match of a well with 20 frac
channel behind casing allowing fluid to bypass the spinner stages, two years of daily production data, and three
between Layers 13 and 14. production logs may be completed in as little as four to six
hours from construction of the data set through completion
12/1/2001 of the final match.
20

19
Sim References
18
Obs 1. Spivey , J.P.: Estimating Layer Properties for Wells in
17 Multilayer Low-Permeability Gas Reservoirs by Automatic
16 History Matching Production and Production Log Data, paper
15 SPE 100509 presented at the 2006 SPE Gas Technology
14 Symposium, Calgary, Alberta, Canada, 1517 May.
13
2. Fetkovich, M.J.: Decline Curve Analysis Using Type Curves,
12
JPT (June 1980) 10651077.
Layer

11

10
3. Carter, R.D.: Type Curves for Finite Radial and Linear Gas-
9 Flow Systems: Constant Terminal Pressure Case, SPEJ
8 (October 1985) 719728.
7 4. Fraim, M.L., and Wattenbarger, R.A.: Gas Reservoir Decline-
6
Curve Analysis Using Type Curves With Real Gas
5
Pseudopressure and Normalized Time, SPEFE (December
4

3
1987) 671682.
2
5. Palacio, J.C., and Blasingame, T.A.: Decline-Curve Analysis
1 Using Type CurvesAnalysis of Gas Well Production Data,
-500 0 500 1000 1500
paper SPE 25909 presented at the 1993 SPE Joint Rocky
Flow rate, Mscf/D Mountain Regional and Low Permeability Reservoirs
Symposium, Denver, Colorado, April 2628.
Fig. 16Incremental production log graph where there is a 6. Gringarten, A.C., Ramey, H.J., and Raghavan, R.: Applied
channel behind casing allowing fluid to bypass the spinner
between Layers 13 and 14.
Pressure Analysis for Fractured Wells, JPT (July 1975) 887
892.
7. Agarwal, R.G., Carter, R.D., and Pollock, C.B.: Evaluation and
Performance Prediction of Low-Permeability Gas Wells
SPE 106305 7

Stimulated by Massive Hydraulic Fracturing, JPT (March


1979) 362372.
8. Crafton, J.W.: Oil and Gas Well Evaluation Using the
Reciprocal Productivity Index Method, paper SPE 37409
presented at the 1997 SPE Production Operations Symposium,
Oklahoma City, Oklahoma, 911 March.
9. Agarwal, R.G., Gardner, D.C., Kleinsteiber, S.W., and Fussell,
D.D.: Analyzing Well Production Data Using Combined-Type-
Curve and Decline-Curve Analysis Concepts, SPEREE
(October 1999) 478486.
10. Gatens, J.M. et al: Analysis of Eastern Devonian Gas Shales
Production Data, paper SPE 17059 presented at the 1987 SPE
Eastern Regional Meeting, Pittsburgh, Pennsylvania, 2123
October.
11. Watson, A.T., Lane, H.S., and Gatens, J.M., III: History
Matching With Cumulative Production Data, JPT (January
1990) 96100.
12. Watson, A.T. et al: An Analytical Model for History Matching
Naturally Fractured Reservoir Production Data, SPERE
(August 1990) 384388.
13. Spivey, J.P., and Frantz, J.H. Jr.: "History Matching Production
Data Using Analytical Solutions for Linearly Varying
Bottomhole Pressure ," paper SPE 29167 presented at the 1994
Eastern Regional Conference and Exhibition, Charleston, West
Virginia, 810 November.
14. Poe, B.D. Jr.: Evaluation of Reservoir and Hydraulic Fracture
Properties in Multilayer Commingled Reservoirs Using
Commingled Reservoir Production Data and Production
Logging Information, US Patent Application No. 09/952,656,
Pub. No. US 2002/0043370 A1.
15. Manrique, J.F., Poe, B.D. Jr., and England, K.: Production
Optimization and Practical Reservoir Management of Coalbed
Methane Reservoirs, paper SPE 67315 presented at the 2001
SPE Production Operations Symposium, Oklahoma City, 26-28
March.
16. Larkin, S.D., et al.: Analysis of Completion and Stimulation
Techniques in a South Texas Field Utilizing Comprehensive
Reservoir Evaluation, paper SPE 93996 presented at the 2005
SPE Production and Operations Symposium, Oklahoma City,
17-19 April.
17. Huckabee, P.T. et al.: Field Results: Effect of Proppant
Strength and Sieve Distribution Upon Well Productivity, paper
SPE 96559 presented at the 2005 SPE Annual Technical
Conference and Exhibition, Dallas, Texas, 912 October.
18. Schubarth, S.K., Spivey, J.P, and Huckabee, P.T.: Using
Reservoir Modeling to Evaluate Stimulation Effectiveness in
Multi-Layered Tight Gas Reservoirs: A Case History in the
Pinedale Anticline Area, paper SPE 100574 presented at the
2006 SPE Gas Technology Symposium, Calgary, Alberta,
Canada, 1517 May.

Vous aimerez peut-être aussi