You are on page 1of 13

URTeC: 2888118

A Practical Way to Prepare Physical-Based Type Well Performance


Curves for Unconventional Reservoirs in the Permian Basin
Hongjie Xiong*, Tiejun Zhu, and James Forest, Texas Oil and Gas Institute, Cao Yang,
Jae Wook Lee, Harish Kumar, and Chen Li, Texas A&M University
Copyright 2018, Unconventional Resources Technology Conference (URTeC) DOI 10.15530/urtec-2018-2888118

This paper was prepared for presentation at the Unconventional Resources Technology Conference held in Houston, Texas, USA, 23-25 July 2018.

The URTeC Technical Program Committee accepted this presentation on the basis of information contained in an abstract submitted by the author(s). The contents of this paper
have not been reviewed by URTeC and URTeC does not warrant the accuracy, reliability, or timeliness of any information he rein. All information is the responsibility of, and, is
subject to corrections by the author(s). Any person or entity that relies on any information obtained from this paper does so at their own risk. The information herein does not
necessarily reflect any position of URTeC. Any reproduction, distribution, or storage of any part of this paper by anyone other than the author without the written consent of URTeC
is prohibited.

Abstract

It has been a huge challenge to generate type well curves for UR plays due to the reservoir heterogeneity,
complicated fracturing mechanisms, different completion designs, and production operations. The industry has been
trying to use analytical, statistic, and numerical modeling methods. Complexity of fracture networks limits the usage
of analytical method, and lack of physics mechanisms and long-term production data constrains the application of
statistic method. Numerical method often requires significant resources and expertise.

This paper will illustrate a practical approach to generate robust type well curves.
1) For a given geologically similar area (GSA), we build a sector numerical model, and then calibrate the
model with available production history with assistance of automatically-history matching method.
2) Identify the key uncertainties and investigate their possible ranges.
3) Study the effectiveness of past completion designs and investigate the possible completion designs.
4) Perform multiple runs with Monte-Carlo simulation method based upon the ranges and the distribution
types of those uncertainties and completion design parameters.
5) Compare the modeling results with the real production data from the same GSA, and perform multiple
runs again by adjusting the ranges and/or distribution shapes of the those uncertainties if needed.
6) Build type well curves based upon the calibrated multiple runs.

We have applied the workflow and built multiple type well curves for several GSAs of those URs in the Permian
Basin. The automatic process of history matching and forecasting significantly reduces the manpower requirement.
Since the approach enables us to incorporate the different flow mechanisms, formation heterogeneity, and
completion designs, we observed that the outcomes of multiple runs from those calibrated models are very
consistent with well production historical behaviors. We witnessed that the workflow was very easy to follow, and
less experience engineers were able to generate those type well curves very efficiently.

Our practical workflow is very easy to follow and can be applied into various UR plays in different basins. The
effective approach enables us to easily integrate various physics flow mechanisms in the unconventional reservoirs
with various completion designs. Therefore, the resulting type well curves should be more reliable to forecast the
overall performance of multiple wells in a given geologically similar area. It is very convenient to develop type well
curves for a subset of the GSA or completion designs, which enable us to get into the critical and insightful
information about the subsurface and/or completion effectiveness.

*Now with University Lands


URTeC 2888118 2

Introduction

The current technology to develop unconventional resource plays is horizontal wells with multistage hydraulic
fracturing treatments. Since permeability is extremely low (<0.001 md) in unconventional reservoirs (UR), multiple
fractures are needed to have economic well rates, as shown in the following image (Figure 1) (Xiong, 2017)

Figure 1 - Horizontal Well with Multi-stage Fracturing Completion

Type Well Performance Curves (Type Curves) are often used in unconventional reservoir exploration and
development (Rastogi et.al, Haskett). Each Type Curve ought to represent the well performance characters of a
group of wells for a given geological similar area (GSA), including initial rate, decline rate, and estimated ultimate
recovery (EUR). In addition to the geology, reservoir properties, and fluid PVT, a Type Curve (TC) also depends on
well lateral spacing and completion design. At the same time, operating conditions also impact well performance,
such as drawdown pressure management scheme and minimum flowing bottom-hole pressure (FBHP).

Xiong et.al (2017) has compared multiple methods to generate TCs, including

(1) Statistic method based upon production data only

This approach is basically to average well performance data of many wells in a GSA with very long history such as
10-30 years, which requires not only a statistically significant number of producers, but also extensive QA/QC on
production data (Freeborn et.al). The biggest drawback for the method is that the resulting profile is not a real
physical case, meaning we would lose the opportunities to exam and interrogate the “representative” case. Since the
completion design plays a very significant role on well performance, the method could mislead the estimation of
future wells that may have different completion design.

It would also be hard to tie-in the type curve with production constraints, for example, different production pressure
management and/or artificial lifting methods. For UR plays, since we do not often have long-life production data,
we have to be very cautious to use the approach though it may be the simplest way.

(2) Empirical Decline Curve Analysis (DCA), Analytical or Semi-Analytical Method

We often use classic DCA methods (Ilk et.al) to analyze the limited production data and project well performance
for the rest of the well life, which requires long enough production history. The keyword here is “Long Enough” so
that the data set can demonstrate the decline characters. The biggest challenge overall for “DCA base approach” and
analytical-based approach (Fulford et.al) is that the data are never long enough to demonstrate the true characters for
the full life of the well.

Miller et.al proposed a very through workflow to build type wells, but it faces the very same challenges by requiring
large well data set and inability to incorporate multiple physical flow mechanisms (Lacayo et.al). Though those
methods are very simple to use, all of them suffer with the multiple flow regimes occurring in a full well life. Those
methods require long-term production history and are applicable for single phase only.
URTeC 2888118 3

(3) Numerical simulation models

Though it may require extra time and expertise, numerical modeling can rigorously simulate the well performance
behavior over the well life. Multiple uncertainties can be taken into consideration with those numerical models and
with multiple cases. The most notable advantage of numerical simulation is that it can fit in different life cycles of
UR plays, and can easily tie-in with different completion designs and operation constraints. With the right physical
mechanisms considered and simulated, the numerical model is supposed to be the best way to forecast the
production performance for the whole well life.

Complex Flow Mechanisms

Unconventional well performance depends not only on reservoir quality but also on well completion effectiveness.
The well flow mechanisms are quite complicated. As shown in the following chart (Figure 2), a well can experience
multiple flow regimes: Wellbore storage/fracture flow, bi-linear flow, earlier linear flow, boundary-dominate flow
(BDF) within stimulated rock volume (SRV), compound linear flow between wells, and final BDF within drainage
volume. Of those flowing regimes, the major flow regimes are linear flow, final BDF and regimes between those
two regimes. The duration of each flow regime varies and depends on reservoir quality and completion
effectiveness. The reservoir heterogeneity makes completion effectiveness vary even with the same completion
design, which compounds the complexity of flow regimes and their durations.

Figure 2 - Flow Regimes of a Horizontal Well with Multi-Stage Fracturing Completion


URTeC 2888118 4

Under the assumption of linear flow with infinite fracture conductivity, Wattenbarger’s Solutions—for “short-term”
approximations (Wattenbarger et al. 1996) can be written as the following equation (Eq. 1):

∅𝐶𝑡 𝑘𝑚
𝑞 = 𝑓(√ ∗ 𝐴√𝑘, 𝑡)∆𝑝 …… (1)
𝜇𝑜 𝐵𝑜

∅𝐶𝑡 𝑘𝑚
In the above equation (Eq.1), √ represents the reservoir quality, A is the fracture surface area, k is the
𝜇𝑜 𝐵𝑜

enhanced permeability, and ∆𝑝 is production draw down pressure. We may name 𝐴√𝑘 as the completion
effectiveness index, which depends on the completion design and execution effectiveness in addition to rock
mechanical properties. In another word, each well may have different completion effectiveness 𝐴√𝑘 even in the
same GSA because of the completion execution effectiveness.

It would be much easier to forecast the well performance if every parameter stays constant with time. The reality is
that every parameter in Eq.1 is a function of time.

∅𝐶𝑡 𝑘𝑚
𝑞 = 𝑓 (√ (𝑡) ∗ 𝐴(𝑡)√𝑘(𝑡), 𝑡) ∆𝑝…… (2)
𝜇𝑜 𝐵𝑜

The completion effectiveness index 𝐴(𝑡)√𝑘(𝑡) shrinks with production depletion, and the shrinking rate depends on
formation stress on proppant, rock mechanical properties, proppant strength, size, and concentration, etc. Similarly,
the reservoir quality will slightly reduce with depletion too.

Thus, complicated flow regimes and dynamic changes of completion effectiveness and reservoir quality
significantly challenge forecasting long-term well production performance. Incorporating multiple physics flow
mechanisms into the forecast modeling is absolutely necessary and definitely enhances the confidence and
reliabilities of the well performance forecast.

Of all of those three methods mentioned above, the numerical modeling approach is the only method to take
multiple flow regimes into the consideration.

Methodology/Workflow

We have been applying numerical simulation method to generate type curves for different GSAs for the University
Lands Acreage in the Permian Basin. The following chart (Figure 3) illustrates our methodology – how to prepare
Type Well Performance Curves with physics-based numerical modeling method.
URTeC 2888118 5

Figure 3 - Workflow to Prepare Type Curves with Physics-Based Modeling Approach

Step 1 By examining the Eq.1, there are many factors affecting well performance, including reservoir quality and
completion effectiveness. The completion effectiveness depends on not only completion designs but also reservoir
rock/geomechanical properties. By performing necessary geological and reservoir engineering studies, one can
determine different geologically similar areas (GSAs) with the consideration of those critical parameters in
reservoir, geomechanical, and completion. For example, we could define GSAs with the information, such as
reservoir quality index k/ϕ, reservoir fluid PVT, pore pressure, and in-situ stress gradients, etc.

In the appraisal and early piloting development phase, we may define a GSA for a wide area. With more data, we
can then divide a big GSA into several smaller GSAs.

Step 2 For a given GSA, we identify an existing well with a good data set and build a sector numerical model, and
then calibrate the model with available production history with automatically-history matching method. Based upon
the history matching results and the understandings of the reservoir, and completion practices and technology trends,
we identify the key parameters/uncertainties and investigate their possible ranges. At the same time, we study the
effectiveness of past completion designs in the GSA and investigate the possible future completion designs. To
incorporate the fracture completion effectiveness shrinkage with time, one of techniques is to incorporate the
fracture conductivity reduction with time into the model.

Considering the modeling time to perform the multiple run and flexibility to vary different completion effectiveness
parameters, we usually build a half-cluster and half-fracture sector model as shown in the following chart (Figure 4),
including one (half) fracture, half fracture spacing (cluster spacing), and an enhanced zone around the fracture. So
that, we can specify the properties of those 3 different zones: fracture, enhanced zone, and matrix, including
permeability, porosity, and dimensions.
URTeC 2888118 6

Figure 4 - One-fourth sector model setup

Step 3 Perform multiple simulation runs with Monte-Carlo simulation method based upon the ranges and the
distribution types of those uncertainties and completion design parameters. Compare the simulation results with the
production history of multiple wells existing in the same GSA. If needed, we may need to go back to Step 2 to
modify the model and perform multiple runs again by adjusting the ranges and/or distribution shapes of the
uncertainties, and/or even add more critical parameters, if needed.

Step 4 Build type well curves based upon the calibrated multiple runs (Xiong et.al 2017). We can build the
different subset type curves if the studying GSA is quite large or with long history, such as by completion intensity
since completion intensity changes with time, etc…

Application Example

We have applied the methodology on multiple unconventional reservoirs on the University Lands Acreage in the
Permian Basin. In the section, we illustrate how to apply our methodology with a case history in the Southern
Midland Basin.

Step 1 – Define Geologically Similar Area (GSAs)

University Lands acreage in the Southern Midland Basin covers a very big area with multiple stacking
unconventional reservoirs. Through a series of geological and reservoir engineering studies, we chose to use
reservoir fluid PVT as the main factor to define the GSAs for the example at the time to prepare the type curves. We
had also considered the pore pressure, and other geological factors. The chosen GSAs are illustrated in the following
chart (Figure 5).
URTeC 2888118 7

Figure 5 - Defining Geologically Similar Areas (GSAs)

Step 2 – Build and Calibrate the Numerical Simulation Model for Each GSA

With the basic information from each GSA, including geology, reservoir fluid PVTs, and basic completion data, we
built a numerical simulation model and then calibrated the model with a representative well data for each GSA. The
following chart show one of the models with history matching results.

Figure 6 – Modeling Calibration Example

During the history matching process, we found out the following parameters very sensitive for the given GSA,
including
1. Matrix permeability;
2. Fracture length and conductivity;
3. Enhanced zone (EZH) width and permeability;
4. Propped fracture height; and
5. Fracture/cluster spacing.

We also found out that the formation porosity, water saturation, rock compressibility, and pay thickness caused less
variance. In those ultra-low permeability reservoirs, wells can’t deplete much beyond fracture network (Xiong, et.al
2018). Thus, lateral well spacing is a business decision based upon the corresponding well completion design. In the
model setup, we chose to fix the lateral well spacing in the whole process.

Step 3 – Perform Multiple Simulations with Monte-Carlo Simulation Approach

With the calibrated numerical model and the identified uncertainties and key parameters as we understand from last
couple steps, we built those uncertainty distributions listed in Table 1.
URTeC 2888118 8

Table 1- Uncertainty Distributions of Those Key Parameters

Figure 7 - Examples of Chosen Parameter Sampling Distributions

With those uncertainty distributions, we then performed multiple simulations. Some of the sampling distributions
are shown in Figure 7. The results of those multiple simulation runs are shown in the Figure 8. We tried to run as
many cases as possible so that there would be enough data sampling density to cover the existing wells and different
subsets possible for different completion design sensitivity or reservoir quality sensitivities.
URTeC 2888118 9

Figure 8 - Cumulative Oil Production of Multiple Simulation vs Existing Well Production

It is critical to compare the production rates of the simulation cases with those from the existing well production in
the same GSA. As shown in Figure 8, blue lines represent the cumulative production from those pseudo wells, and
the gray-color curves are the real production data from those wells in the GSA. Right-hand chart in Figure 8 zooms
into the time period when the existing wells produce so far. A visual inspection should be close enough to tell if the
calibrated model can predict the production of all of those wells in the GSA. Each individual well performs
differently possible due to completion and/or reservoir properties.

If there is a significant difference between the pseudo wells from multiple simulation cases and the existing wells,
one may need to expand the ranges of those key parameters listed in Table 1, and re-run the Monte-Carlo
simulations, and compare the results with the existing well performance. In some cases, for example, the existing
well performance is outside of the pseudo well performance cloud, we may have to expand our key parameters, such
as fluid PVT properties, which can happen in the early development phase or appraisal phase when the data is
limited.

One of the advantages from the calibrated model is to forecast those three phases in the well production stream,
including oil, gas, and water.

Step 4 Choose a Representative Case as the Type Curve for the GSA

We have detailed how to choose a representative case from those multiple cases for a GSA in the technical paper
URTeC 2668394 (Xiong et.al, 2017). For the example here, the chosen representative case is shown as the thick
black line in the following (Figure 9), in which, the left-hand chart illustrates the oil phase type curve and the right-
hand chart shows the gas phase type curve.

Figure 9 - Choose a Representative Case for Type Curves of Both Oil and Gas Phases for the GSA
URTeC 2888118 10

We recommend converting the numerical format of the representative case to Classic/Arp’s style format with multi-
segment DCA method (Xiong et.al, 2017) for the convenience of typical type curve applications. The segments can
be based upon flow regimes and timing determined by diagnostic plots, such as, log rate – log time plot, SQRT time
plot for telf , “b” & “D” plots vs time.

1. Determine flow regime timing by diagnostic plots, such as, log rate – log time plot, SQRT time plot for telf ,
“b” & “D” plots vs time,…, with normalized rates (Left-hand side chart in Figure 10);
2. Divide the production profile into two or three segments based upon the timing when flow regime changes
and decline characters (middle chart inFigure 10);
3. Determine the values of “b” and “Di” for those segments by curve-fitting; Suggesting
• “b”=2 for the first segment (until the end of the first linear flow)
• “b”<2 for the second segment (transitional flow period)
• “b”<1 for the last segment (depletion period)
4. Compare the rates and EURs between the numerical model and multi-segment DCA (right-hand side chart
in Figure 10);
5. Last step is to validate the new type curve by evaluating diagnostic plots.

Figure 10 - Convert the Representative Case into a Multiple-Segment Type Curve (URTeC 2668394)

Discussion

As illustrated in the workflow and the example, the advantages of using a calibrated numerical simulation model to
generate well performance type curves include
(1) capable of taking multiple flow physical mechanisms into consideration; and
(2) capable of forecasting multiple phase flow rates, which minimizes the possible uncertainties that could be
brought in by using different ratios, such as using cum GOR for gas production, and/or using cumulative
water cut for water production based upon oil phase production only.
URTeC 2888118 11

Another advantage of the method includes generating a subset type curve in the same GSA. Since the completion
design has very significant impact on UR well performance, an operator might have tried several different types of
completion designs in the same GSA. We can generate the subset type curves by narrowing down the completion
design parameters, such as by cluster spacing, etc. Figure 11 and Figure 12 show the examples of subset type curves
for two fracture cluster spacing sets.

Figure 11 - Subset Type Curve for Cluster Spacing 20-30 ft

Figure 12 - Subset Type Curve for Cluster Spacing 10-15ft

Fracture conductivity usually diminishes with pressure depletion, which means that the fracture dimension can
reduce with depletion, which in turn will reduce well performance. Depending on the rock and geomechanical
properties and proppant pumped in the formation, we may need to incorporate the fracture conductivity reduction to
prepare the type curves. When we build and calibrate the model, we should evaluate if we need to consider that in
the modeling process.
URTeC 2888118 12

The challenge to apply the proposed method is the data, time, and experience to build and calibrate the model.
However, it is our experience that it may take some time for the first-time application, and our workflow really
speedups the process, and helps in-experienced individuals prepare type curves. Once the first case (model) is
complete for a GSA, and following cases should be very easy for the rest of GSAs.

Conclusions

Considering the pros and cons of different methods to generate type curves, our method may present the best way to
easily integrate various physics flow mechanisms in the unconventional reservoirs with various completion designs,
and forecast the production rates of three phases, which is very critical to the UR plays in the Permian Basin.
Therefore, the resulting type well curves should be more reliable to forecast the overall performance of multiple
wells in a given geologically similar area.

Our practical workflow is very easy to follow and can be applied into various UR plays in different basins. It is very
convenient to develop type well curves for a subset of the GSA or completion designs, which enable us to get into
the critical and insightful information about the subsurface and/or completion effectiveness.

References

Freeborn, F., Russell, B., and Keinick, W.: Creating Analogs, Fact and Fiction, paper SPE 162630 presented at the
SPE Hydrocarbon, Economics, and Evaluation Symposium held in Calgary, Alberta, Canada, 24-25 September
2012.

Fulford, D.S. and Blasingame, T.A.: Evaluation of Time-Rate Performance of Shale Wells using the Transient
Hyperbolic Relation, paper SPE 167242 presented the at the SPE Unconventional Resources Conference-Canada
held in Calgary, Alberta, Canada, 5-7 November 2013

Haskett, W.: Unconventional Type Curves: Useful, or Sirens of Destruction? paper SPE 147059 presented at the
SPE Annual Technical Conference and Exhibition held in Denver, CO, USA, 30 Oct – 2 Nov 2011.

Ilk, D., Rushing, J.A., Perego, A.D., and Blasingame, T.A.: Exponential vs Hyperbolic Decline in Tight Gas Sands –
Understanding the Origin and Implications for Reserve Estimate Using Arps’ Decline Curve, paper SPE 116731
presented SPE 2008 ATCE held in Denver, CO., USA, 21-24 September 2008
Lacayo, J. and Lee, J.: Pressure Normalization of Production Rates Improves Forecasting Results, paper SPE
168974 presented at the SPE Unconventional Resources Conference – USA held in the Woodlands, TX, USA, 1-3
April 2014

Miller, P., Frechette, N., Kellett, K.: Building Type Wells for Appraisal of Unconventional Resource Plays, paper
SPE 185053 presented at the SPE Unconventional Resources Conference held in Calgary, Alberta, Canada, 15-16
Feb., 2017.

Rastogi, A. and Lee, J.: Methodology for Construction of Type Wells for Production Forecasting in Unconventional
Reservoirs, paper URTeC 2152273 presented at the Unconventional Resources Technology Conference held in San
Antonio, TX, USA, 20-22 July 2015

Wattenbarger R., et. al 1996: Production Analysis of Linear Flow into Fractured Tight Gas Wells, paper SPE 39931
presented at the SPE Rock Mountain Regional/Low Permeability Reservoirs Symposium and Exhibition held in
Denver, CO, USA 5-8 April 1996

Xiong, H.: Optimizing Cluster or Fracture Spacing: An Overview, the article published at the SPE
THEWAYAHEAD, May 2017, https://www.spe.org/en/twa/twa-article-detail/?art=3007
URTeC 2888118 13

Xiong, H., Gao, S., and Li, H.: Generate Type Well Performance Curves by Combining Multi-Segment Decline
Models and Calibrated Numerical Simulation Models for UR Wells in Permian Basin, paper URTeC 2668394
presented at the Unconventional Resources Technology Conference held in Austin, Texas, USA, 24-26 July 2017

Xiong, H. Wu, W., and Gao, S.: Optimizing Well Completion Design and Well Spacing with Integration of
Advanced Multi-Stage Fracture Modeling & Reservoir Simulation - A Permian Basin Case Study, paper SPE
189855 presented at the SPE Hydraulic Fracturing Technology Conference held in The Woodlands, TX, USA, 23-
25 January 2018.

You might also like