Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

8/30/2013

Production Forecasting for


Unconventional Resources
Lecture 7
Continuous EUR Estimation
Summary of Currie, Ilk,
Blasingame Method
SPE 132352

Learning Objectives
You will be able to
 Explain the Currie et al. “Continuous Estimation of Ultimate
Recovery” methodology
 State the reasons why the Arps hyperbolic model is likely to
overstate EUR and reserves for low-permeability gas wells
 Explain the similarities between the Ilk et al. “Power Law” model
and the Stretched Exponential Decline Model (and the difference
in the Extended Power-Law model)
 Explain why the Power-Law model is more likely than the Arps
hyperbolic model to provide an upper estimate of EUR
 Explain why the Arps exponential decline model is likely to
provide a lower limit estimate of EUR
 Critique the Continuous Estimation of Ultimate Recovery
methodology

1
8/30/2013

Continuous Estimation of UR Methodology


 Method actually applied at discrete time intervals as
more production data become available (so not truly
‘continuous’)
 Data analyzed with
 Arps’ hyperbolic model

 Power-law model (SEDM) and expanded power-


law model
 Arps’ exponential model (using rate-cumulative
production relations)
 Results plotted vs. time, allowing examination for
approach to convergence to constant EUR estimate,
stabilization

Review of Arps Model Development


 Instantaneous decline rate, D, related to “loss ratio”
1 𝑞
≡- 𝑑𝑞
𝐷
𝑑𝑡
 Decline parameter, b, related to derivative of loss
ratio
𝑑 1
𝑏≡
𝑑𝑡 𝐷
 Integration for constant b leads to Arps’ hyperbolic
decline model
𝑞𝑖
q(t) = 1
1+𝑏𝐷𝑖 𝑡 𝑏

2
8/30/2013

Observations: Arps’ Hyperbolic Decline Model


 Most-used relationship in industry for reserves
estimation (often inappropriately)
 Requires constant b and requires b ≤ 1
 Constant b required in derivation of model

 Finite reserves as 𝑡 → ∞ requires b < 1

 Extrapolation using Arps’ model prior to onset


of BDF will cause overestimation of reserves
 b continuously decreases with time during
transient flow

Power-Law Decline Model


 Basic power-law model
𝑞 = 𝑞𝑖 exp −𝐷𝑖 𝑡 𝑛
 Actually identical to stretched-exponential decline
model
𝑡 𝑛
𝑞 = 𝑞𝑖 exp − 𝜏
 Extended power-law model
𝑞 = 𝑞𝑖 exp −𝐷∞ 𝑡 − 𝐷𝑖 𝑡 𝑛
 𝐷∞ term improves match of some simulated data
during BDF
 No longer identical to SEDM
𝑛
 Approaches exponential decline when 𝐷∞ 𝑡 ≫ 𝐷𝑖 𝑡

3
8/30/2013

Arps’ Exponential Decline Model


 Rate-time equation
𝑞 = 𝑞𝑖 𝑒𝑥𝑝 −𝐷𝑖 𝑡 , 𝐷 = 𝐷𝑖 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
 Rate-cumulative production equation
𝑞 𝑡 = 𝑞𝑖 -𝐷𝑖 Gp(t)
 Indicates plot of q(t) vs. Gp(t) will be
straight line, with EUR found by
extrapolation to q(t) = 0 (or, alternatively,
to economic limit rate or rate at specified
time, such as 30-year well life)

Application of Continuous EUR Method (SPE 132352)

4
8/30/2013

Application of Continuous EUR Method (SPE 132352)

Detail for Arps Hyperbolic Model at 2,000


Days: EUR = 3.10 Bcf

5
8/30/2013

Detail for Power-Law Model Analysis at


2,000 Days: EUR = 3.05 Bcf

Detail for Arps’ Exponential Model Analysis


at 2,000 Days: Gp,max = 2.58 Bcf

6
8/30/2013

Application of Continuous EUR Method (SPE 132352)


 Identify upper limit for EUR from power-law model
 Identify lower limit for EUR from Arps’ exponential
model

Validation with Synthetic Data


 Data from simulated east Texas gas well
(SPE 84287)
 Tight gas well
 Vertical well
 k = 0.005 md
 FCD = 10
 Gp (30 years) = 2.42 Bscf
 Gp (q = 0) = 2.65 Bscf

7
8/30/2013

Validation with Synthetic Data: q, Gp vs. t

EUR Estimates with Hyperbolic Decline Model

8
8/30/2013

EUR Estimates with Power-Law Model

EUR Estimates with Exponential Model

9
8/30/2013

Summary: EUR vs. Time

Summary: Parameters from Best Fits


All parameters from final fit at 2,000 days
Hyperbolic: qi = 4,400 Mscf/D,
Di = 0.00497 d-1, b = 0.75
Power-Law: 𝑞𝑖 = 812,662 Mscf/D, 𝐷𝑖 =
4.158 𝑑−1 , n = 0.07, 𝐷∞ = 0.000460 d-1
Exponential: Slope = 738*e-6 d-1,
Intercept = 1,601 Mscf/D, Gpult = 2.17 Bscf

10
8/30/2013

Field Example: East Texas Gas Well (SPE 84287)


 Hydraulically fractured well, tight gas
reservoir

Analysis of East Texas Gas Well (SPE 84287):


Hyperbolic Model
 Off-trend points removed prior to analysis
 No evidence of BDF
 b-values range between 2.2 (earliest data) and 1.57 (all data)

11
8/30/2013

Analysis of East Texas Gas Well (SPE 84287):


Power Law Model

Analysis of East Texas Gas Well (SPE 84387):


Exponential Decline

12
8/30/2013

Summary: EUR vs. Time

Summary: Parameters from Best Fits


All parameters from final fit at 2,000 days
Hyperbolic: qi = 4,100 Mscf/D,
Di = 0.0152 d-1, b = 1.57
Power-Law: 𝑞𝑖 = 66,500 Mscf/D, 𝐷𝑖 =
1.684 𝑑−1 , n = 0.15, 𝐷∞ = 0 d-1
Exponential: Slope = 443*e-6 d-1,
Intercept = 1,145 Mscf/D, Gpult = 2.58 Bscf

13
8/30/2013

Field Example 2: Fractured Gas Well

Field Example 2: Hyperbolic Decline Model

14
8/30/2013

Field Example 2: Power Law Model

Field Example 2: Exponential Decline

15
8/30/2013

Field Example 2: EUR vs. t


 Transition from transient to BDF likely – but transition not
complete (b>1)

Summary: Parameters from Best Fits


All parameters from final fit at 1,930 days
Hyperbolic: qi = 3,495 Mscf/D,
Di = 0.0121 d-1, b = 1.90
Power-Law: 𝑞𝑖 = 40,421 Mscf/D, 𝐷𝑖 =
1.93 𝑑−1 , n = 0.11, 𝐷∞ = 0 d-1
Exponential: Slope = 242*e-6 d-1,
Intercept = 863 Mscf/D, Gpult = 3.57 Bscf

16
8/30/2013

How Can We Obtain Software to Apply Method?


 Free software (latest version) available
to implement method at

http://www.pe.tamu.edu/blasingame
/data/z_Rate_Time_Spreadsheet/

Concluding Observations
 Uncertainty in reserves and EUR estimates for
unconventional reservoirs relatively large
 Systematic methodology needed to quantify
uncertainty and identify likely range
 Continuous EUR procedure provides required
methodology
 Procedure evaluates EUR as function of time

 EUR eventually stabilizes when complete BDF


achieved
 EUR from Power Law model usually stabilizes even
during transient flow

17
8/30/2013

Concluding Observations
 EUR from terminal exponential decline model
usually continues to increase with time until
BDF achieved
 EUR from terminal exponential decline model
provides lower limit estimate
 EUR from power-law model provides upper
limit estimate
 EUR from hyperbolic model usually larger than
from power law (especially if no BDM), and
can provide unrealistic upper limit estimate

Accomplishments
You are now able to
 Explain the Currie et al. “Continuous Estimation of Ultimate
Recovery” methodology
 State the reasons why the Arps hyperbolic model is likely to
overstate EUR and reserves for low-permeability gas wells
 Explain the similarities between the Ilk et al. “Power Law” model
and the Stretched Exponential Decline Method (and the
difference in the Extended Power-Law model)
 Explain why the Power Law model is more likely than the Arps
hyperbolic model to provide an upper estimate of EUR
 Explain why the Arps exponential decline model is likely to
provide a lower limit estimate of EUR
 Critique the Continuous Estimation of Ultimate Recovery
methodology

18
8/30/2013

Production Forecasting for


Unconventional Resources
Lecture 7
Continuous EUR Estimation
Summary of Currie, Ilk,
Blasingame Method
SPE 132352
End

19

You might also like