SPE 125031 Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

D. Ilk, Texas A&M University, J.A. Rushing, Anadarko Petroleum Corp., and T.A. Blasingame, Texas A&M University
Copyright 2009, Society of Petroleum Engineers This paper was prepared for presentation at the 2009 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, 47 October 2009. 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 This work formalizes the Arps' D and b-parameters using a semi-analytical flow relation for a well produced at a constant bottomhole pressure in an HP/HT reservoir. The semi-analytical formulation is given in rate-time and rate-cumulative forms, and the definitions of the D and b-parameters are derived directly from the semi-analytical form. The rationale for this work is the over-dependence that many analysts place on the Arps' D and b-formulations (i.e., the original hyperbolic rate-time model), tuned to specific (i.e., constant) values of the D and b-parameters. The purpose of this paper is to establish guidelines for reserve estimations and for production extrapolations in general. The relations developed in this work can be applied to a number of different HP/HT tight gas and shale gas field cases in this work we validate our proposed procedure with a simulated example and we apply the methodology to a fractured gas well in an HP/HT reservoir. A comprehensive workflow is provided, as well as the relevant details/formulations of the analytical flow solution for HP/HT gas cases. We also provide decline curve plots/plotting functions to yield reserve estimates and production extrapolations. Introduction Unconventional gas resources comprise a very important share of the domestic natural gas resource base and offer tremendous potential for future reserve and production growth. Unconventional gas resources in particular, tight gas sands are characterized by very-low/low permeability and low porosity reservoir properties. And as the natural gas industry continues to explore for more resources, exploration and development activities extend to much greater depths. These depths exceed 15,000 ft and are approaching to 25,000 ft. The reservoirs in these depths exhibit abnormally high initial pore pressure and temperature gradients resulting in high-pressure/high-temperature (HP/HT) reservoir conditions. The common practice in the industry in reserves assessment is to use the Arps' [1945] hyperbolic rate decline relations. Similar to conventional gas resources, tight gas sand reserves are frequently (often solely) assessed with Arps' hyperbolic rate decline relations. The Arps rate decline relations are only specifically valid for boundary-dominated flow conditions, and it is often noted that the decline curve parameter, b, should lie between 0 and 1. However, we frequently observe b-parameter values greater than 1 particularly in tight gas sands at HP/HT reservoir conditions, and it is well established that for cases of b>1, the reservoir is in transient (or transition) flow. It is inappropriate (in a strict reservoir engineering sense) to use the Arps models to estimate reserves for cases where b>1 as this will yield significant errors in reserve estimates particularly in unconventional resources (e.g., tight gas, shale gas, HP/HT gas reservoirs, etc). This is not to say that the hyperbolic relation cannot be used as an "interpolation" or "fitting equation" for transient flow data in fact, this appears to be a common practice in the work of economic reserves practitioners. While empirical fitting of transient flow data using the hyperbolic decline relation is "empirical" at best (i.e., an appropriate reservoir model should be used), our concern lies not in the use of the hyperbolic relation for "fitting" the transient flow data, but rather, in the extrapolation of the transient flow (hyperbolic) trend to estimate reserves. Extrapolation of transient flow data using any empirical model to estimate reserves is an inappropriate practice, and should never be conducted in any capacity other than building possible scenarios not for estimating reserves! Rushing et al [2007] present a study which was designed to evaluate the applicability of Arps' decline curve methodology for assessing reserves in tight gas sands at HP/HT reservoir conditions. Their approach showed that the errors in reserve estimates based on the hyperbolic rate decline relation were substantially higher (well over 100 percent error) when the hyperbolic rate decline relation is applied during transient/transition flow regimes. However, they indicated that Arps'

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

relations might be appropriate for assessing reserves in tight gas sands at HP/HT reservoir conditions if applied under the correct conditions (i.e., boundary-dominated flow conditions). In this work our objective is to reduce the uncertainty in reserve estimates by using a combined methodology which not only includes a semi-analytical relation proposed in this work, but also includes other (rigorous) reserves assessment techniques. Our methodology is driven by the use of the semi-analytical rate-time/rate-cumulative production relation, which is valid for boundary-dominated flow conditions in a moderate to HP/HT gas reservoir. We also employ the empirical, power-law exponential rate-time relation (Ilk et al [2008]) as a standard since this model has been shown to be more robust (and more applicable) than the typical Arps' exponential and hyperbolic families of rate decline relations. Lastly, where appropriate, we use model-based analysis techniques (i.e., decline type curve analyses and other similar "model-based" analysis techniques). Development of the Methodology Knowles [1999] presented an approach for linearizing the gas flow equation the nonlinear term (gcg) is assumed to be linear with pressure, which leads to a direct formulation for gas flowrate. The approach yields a (p/z)-squared form of the stabilized gas flow equation and when this relation is coupled with the gas material balance equation, a semi-analytic gas flowrate equation is obtained. Ansah et al [2000] generalized the concept (linearization of the gas flow equation) proposed by Knowles and developed several semi-analytical (i.e., direct) solutions for determining average reservoir pressure, rate, and cumulative production for gas wells produced at a constant bottomhole pressure during reservoir depletion. Although this relation was originally obtained for reservoir pressures lower than 6000 psia, we have found this relation useful for practical applications (including reservoir pressures much higher than 9000 psia see Blasingame and Rushing [2005]). Using a straight-line linearization scheme given by Knowles [1999] for the nonlinear (gcg) term, the following equation (in dimensionless form) is obtained.
................................................................................................................. (1) ((1 + pwD ) (1 pwD ) exp[ pwD t Dd ]) 2 Eq. 1 is the main equation in our work, and it is noted that Eq. 1 is only valid for the boundary-dominated flow conditions. Further details can be found in the work by Ansah et al [2000]. The dimensionless variables are defined as: qg ............................................................................................................................................................................ (2) q Dd = q gi q Dd =
2 4 pwD exp[ pwD t Dd ]

pwD = And t Dd =

pwf /z wf pi /zi Jg

.................................................................................................................................................................. (3)

cti G

t ........................................................................................................................................................................ (4)

Where Jg is the productivity index and defined as: 2kh ............................................................................................................................................ (5) Jg = 2.2458 A 141.2i Bgi ln 2 C A rwa Using the definitions of the "loss-ratio" and the "loss-ratio derivative" by Arps [1945] and Johnson and Bollens [1927], we define the dimensionless D-function (DD) and b-function (b) as: DD = b= 1 dq Dd ............................................................................................................................................................... (6) q Dd dt Dd

q Dd ................................................................................................................................................. (7) (dq Dd /dt Dd ) Using these definitions (i.e., Eqs. 6 and 7) and our proposed "HP/HT" rate model (i.e., Eq. 1), we obtain the following results for the DD and b-functions (after simplification): d dt Dd DD =
b=

pwD (1 pwD + (1 + pwD ) exp[ pwD t Dd ]) .................................................................................................................. (8) ( pwD 1 + (1 + pwD ) exp[ pwD t Dd ])
2 2 exp[ p wD t Dd ] (1 p wD )

(1 p wD + (1 + p wD ) exp[ p wD t Dd ]) 2

.......................................................................................................................... (9)

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

We present the behavior of Eqs. 1, 8 and 9 in Fig. 1, which is the so-called "q-DD-b" plot for various values of pwD. As noted earlier, these relations are only valid for boundary-dominated flow regime and this constraint is reflected in Fig. 1 by the near-constant values of q-DD-b at early times (typically of boundary-dominated flow models). We also note in Fig. 1 that each of the rate cases tend to 0 due to the quasi-exponential nature Eq. 1 for very late times. A significant conclusion drawn from Fig. 1 is that the b-parameter value is more or less constant and tends to converge to 0.5 for high drawdown cases. This plot could be considered as verification for the statement that for boundary-dominated flow conditions, b-parameter value is almost 0.5 for high drawdown cases in gas flow (this observation is in agreement with statements made by Fetkovich [1980] and Fetkovich et al [1987] that for gas flow cases, 0.4< b <0.6). Further, the propsoed gas flow relation (i.e., Eq. 1) yields identical results as the numerical solutions proposed by Carter [1985]. We believe that for HP/HT reservoir conditions, this relation should be useful in reserve estimates and production forecasts (assuming (obviously) that boundary-dominated flow conditions are established). Using Eq. 1, a "quadratic" rate-cumulative production relation can be obtained (see Blasingame and Rushing [2005] for details). This relation is given (in dimensionless form) as:
G 2 ................................................................................................................................................ (10) 2 pD Where the dimensionless variables are defined as: Gp G pD = ......................................................................................................................................................................... (11) G And G = .................................................................................................................................................................... (12) (q gi /Di ) q Dd = 1 G pD + where 2q gi

Di =

G The definitions of the "loss-ratio" and the "loss-ratio derivative" in dimensionless rate-dimensionless cumulative form (originally given by Ilk et al [2008]), we define the dimensionless D-function (DD) and b-function (b) as:
DD = dq Dd ................................................................................................................................................................... (14) dG pD

p /z 1 wf wf pi / zi

............................................................................................................................................... (13)

b = q Dd

d dG pD

1 .................................................................................................................................... (15) (dq Dd /dG pD )

Substituting Eq. 10 into Eqs. 14 and 15, we obtain the following results for the DD and b-functions for the "quadratic ratecumulative production" relation: DD = (1 G pD ) .............................................................................................................................................................. (16)
b= 2 2 G pD + G 2 pD 2 (G pD 1) 2 .................................................................................................................................................. (17)

In this work we utilize the rate-time/rate-cumulative production data and the associated D and b-data functions as plotting functions. Specifically, we show that using the DD and b-models add more insight into the analysis. The model parameters (e.g., qgi, Di, and G) are obtained using a simultaneous matching process. Our process is illustrated in the next sections. We also use the "power-law exponential" rate decline relation which has recently been introduced by Ilk et al [2008] for verifying the gas-in-place estimate obtained by the semi-analytical relation from an empirical standpoint. Ilk et al apply the "power-law exponential" rate decline model to various field cases (Ilk et al [2008] and Mattar et al [2008]) as well as to simulated data (see Mattar et al [2008]). In those efforts, the power-law exponential rate decline model is verified to be both accurate and robust matching transient, transition, and boundary-dominated flow data and yielding very consistent (i.e., reliable) reserve estimates. For reference the power-law exponential rate decline relation is given as: t n ] .............................................................................................................................................. (18) gi exp[ D t D qg = q i

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

The definitions of the so called "loss-ratio" and the "loss-ratio derivative" as presented by Johnson and Bollens [1927] and Arps [1945] help to describe the basis of the "power-law exponential" rate equation. The original definitions of the "lossratio" and the "loss-ratio derivative" are: 1 q (Definition of the Loss-Ratio) ........................................................................................... (19) D dq/dt

q (Derivative of the Loss-Ratio)........................................................................................... (20) dq/dt The "power-law exponential" rate decline relation (Eq. 18) is derived based on the observation that the D-parameter (actually D(t)) exhibits power-law behavior as a function of time. Ilk et al applied this new model to various field cases (Ilk et al [2008b] and Mattar et al [2008]), as well as to simulated data (see Mattar et al [2008]) and verify that the power-law exponential rate decline model is robust enough to match transient, transition, and boundary-dominated flow data. Further, the "power-law exponential" rate decline has been shown to yield consistent reserve estimates for field cases. b d dt d 1 D dt
This last step in our workflow is to apply analytical and numerical "model-based" analysis of production rate and pressure data to estimate reservoir properties and reserves. We believe that the incorporation of these three techniques (the proposed HP/HT formulation, the "power-law exponential" rate decline relation, and model-base analysis) provides a workflow that will yield reliable estimates of reserves/gas-in-place for HP/HT reservoir conditions.

Validation of New Analysis Methodology

In this section we use a numerical simulation case to verify our proposed methodology. We acquired the simulated data from the work by Rushing et al [2007]. The reservoir consists of a producing hydraulically-fractured well completed at an average depth of 18,000 ft in a HP/HT, low permeability, low-porosity tight gas sand. The reservoir fluid was asssumed to be a dry gas with water saturations which are only slightly higher than the irreducible values which yields a very small mobile water component. The reservoir and fluid properties are given in Table 1 below. The detailed information on the fluid properties correlations can be found in the work by Rushing et al [2007].
Table 1 Reservoir and fluid properties for the numerical simulation case (Rushing et al [2007]) hydraulically fractured gas well in a HP/HT reservoir.

Reservoir Properties: Net pay thickness, h Effective permeability, k Wellbore radius, rw Formation compressibility, cf Porosity, Initial reservoir pressure, pi Gas saturation, Sg Water saturation, Sw Skin factor, s Reservoir temperature, Tr Drainage area, A Fluid Properties: Gas specific gravity, g

= = = = = = = = = = = =

200 ft 0.009 md 0.30 ft 110-9 1/psi 0.061 (fraction) 16200 psia 0.636 (fraction) 0.364 (fraction) 0 (dimensionless) 400 oF 80 acres 0.633 (air = 1)

Hydraulically Fracture Model Parameters: = 200 ft Fracture half-length, xf = 28 (dimensionless) Fracture conductivity, FcD Production Parameters: Producing time, t = 50 years

In Fig. 2 we present the flowrate data for this well which spans over 50 years (i.e., the entire simulation run). We plot the gas flowrate data and the auxiliary functions (i.e., the rate-integral and the rate-integral derivative) for diagnosis of the flow regimes in Fig. 3. The half-slope trend exhibited by the rate-integral derivative indicates the linear flow regime (i.e., the signature of an infinite conductivity vertical fracture) and we also observe that the boundary-dominated flow regime is established after almost 2500-3000 days. Our first task is to employ the semi-analytical relations in order to estimate the gas-in-place. In Figs 4a and 4b we present the flowrate versus cumulative production and the logarithm of flowrate versus time plots (respectively); where the data are matched using the proposed "Ansah" semi-analytical rate model. We observe outstanding matches after boundary-dominated flow is establish clearly, the proposed "Ansah" model is appropriate for this case. The gas-in-place estimated using the "Ansah" relation was found to be 8.1 BSCF.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

In Figs 5a, 5b, and 5c we present the behavior of the computed D-parameter trend on different scales (Cartesian, semi-log, and log-log, respectively). In Figs 5a, 5b, and 5c, the D-parameter trend is modeled using Eq 8 or Eq. 16 (as appropriate). Our rationale for different scales is to understand the characteristic behavior of the computed D-parameter. As can be seen in Figs 5a, 5b, and 5c very good matches are obtained for the boundary-dominated flow regime. Similarly, in Figs 6a, 6b, and 6c we plot the computed b-parameter trend on different scales. At this point we should mention that we utilize the "Bourdet" [1989] derivative algorithm for the calculation of D and b-parameters. Since we perform the differentiantion twice to obtain the b-parameter, the "end-point" effects caused by the derivative algorithm can be significant (or even severe). Quite interestingly Figs 6a, 6b, and 6c verify our conclusion derived by observing behavior on Fig. 1a and Fig. 1b that for the drawdown cases where pwD tends to 0.1-0.2, the b-function exhibits an exponential decline behavior at late times. In fact, for this case we can conclude that at late times, the reservoir is in complete boundarydominated flow regime this observation can be verified by inspecting Figs 6a, 6b, and 6c. We also associate the increase in the computed b-parameter trend after 12,000 days to end point effects (not reservoir effects). Our next step is to apply the "power-law exponential" rate relation (i.e., Eq. 18) to estimate the reserves (i.e., the maximum cumulative gas production as t). Fig. 7 presents the "q-D-b" plot for this simulation case. We obtain excellent matches for the computed D-parameter data and flowrate data across all flow regimes. To distinguish the effect of the complete boundary-dominated flow regime, we also include the results when the D parameter in Eq. 18 is set to zero. It is worth noting that we obtain the same value for the reserves estimate as we estimated for gas-in-place using the "Ansah" semianalytical relation (Gp,max=8.1 BSCF).

Figs 8a and 8b simply present the flowrate and cumulative production forecast as specified by these two semi-analytical analysis methods. We note that the behavior of the models is different (specifically, the "Ansah" relation does NOT model behavior during transient flow), but the gas-in-place/reserves estimates are consistent. The analysis results for this case are summarized in Table 2 below.
Table 2 Analysis results for the numerical simulation case.

"Semi-Analytic Relation" Analysis: = Initial gas production rate, qgi = Decline constant, Di Gas-in-place, G =

1450 MSCF/D 3.0x10-4 1/D 8.1 BSCF

"Power-Law Exponential Rate Decline Model": i = 4x105 MSCF/D Initial gas production rate, q Decline constant, D = 3.74 1/D i = 7x10-5 1/D Decline constant, D Time exponent, n = 0.065 Maximum gas production, Gp,max = 8.1 BSCF (D0)

Application to Field Data

We apply our proposed analysis methodology to field data acquired from a hydraulically fractured gas completed in a HP/HT tight gas reservoir. For reference, the estimated initial pressure for this case is approximately 14,000 psia and the formation temperature is estimated to be 260 Deg F. Approximately 3.5 years of daily wellhead and gas flowrate measurements are available for this well. Fig. 9 presents the base gas flowrate and the calculated bottomhole pressures for this well. Inspecting Fig. 9, we immediately observe the early well clean-up effects as well as liquid-loading and operational changes. In Fig. 10 we plot the flowrate and its auxiliary functions for diagnosis of the flow regimes. It is very important to note that we scrutinize the flowrate data and remove the erroneous/redundant data points. Because of the quality of these data at early times, it is difficult to clearly distinguish the individual flow regimes. We do observe the familiar half-slope trend in the rateintegral derivative function and we can also conclude boundary-dominated flow effects are established at late times which will enable us to deploy our semi-analytical flowrate models to estimate reserves and forecast production. In Figs. 11a and 11b we present the results for the "Ansah" semi-analytical rate relation. We obtain fairly good matches of the data with the "Ansah" model during boundary-dominated flow, and our analyses using this relation provide an estimate of gas-in-place of approximately 8.0 BSCF. For the computation of D and b-functions we remove the outlier data points and then we perform the numerical differentiation required for the D and b-functions. In Figs. 12a, 12b, and 12c we achieve quite reasonable matches of the D-function data using the "Ansah" model. However; for the computation of the b-function, the quality of the data significantly affects this computation as seen in Figs. 13a, 13b, and 13c. For this case, the matches of the b-function data with the "Ansah" model are "problematic" the data indicate no unique characteristic behavior. Next, we employ the "power-law exponential" rate decline relation. Fig. 14 presents the q-D-b plot for this case, and we observe a reasonable match of the D-parameter data trend with the power-law model. The match of the flowrate data is good (with the exception of the early time data affected by "cleanup") and the "power-law exponential" rate model yields a

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

maximum gas production value (Gp,max) of about 8.0 BSCF which is consistent with the value from semi-analytical relation. Our results for this case are summarized in Table 3.
Table 3 Analysis results for the field example (tight gas well).

"Semi-Analytic Relation" Analysis: = Initial gas production rate, qgi = Decline constant, Di Gas-in-place, G =

4700 MSCF/D 1.12x10-3 1/D 8.0 BSCF

"Power-Law Exponential Rate Decline Model": i = 3.2x104 MSCF/D Initial gas production rate, q Decline constant, D = 1.33 1/D i = 8.0x10-5 1/D Decline constant, D Time exponent, n = 0.15 Maximum gas production, Gp,max = 8.0 BSCF (D0) In Figs 15a and 15b we present production forecasts for this well generated using the "Ansah" semi-analytical and the powerlaw exponential rate decline models. In addition to these two models, we also use an analytical reservoir model for a vertical well with a finite conductivity vertical fracture (the "analytical" model posted on these plots). As can be observed from Figs 15a and 15b, all of the models approximate the data quite well and yield consistent reserve/gas-in-place estimates. Finally the history match of the flowrate and calculated bottomhole pressures generated using the analytical reservoir model is presented in Fig. 16. We observe an excellent match of the flowrate data and a reasonable match of the pressure history. For reference, the parameters estiamted using the analytical reservoir model are summarized in Table 4.
Table 4 Model parameters for the field example case analytical reservoir model.

Effective permeability, k Fracture half-length, xf Fracture conductivity, FcD Gas-in-place, G

= = = =

0.012 md 80 ft 3.26 (dimensionless) 8.0 BSCF

Summary and Conclusions

Summary: In this work we propose a new approach for the direct estimation of gas-in-place and reserves using only rate-time and rate-cumulative production data. We utilize a semi-analytical rate relation given by Knowles [1999] and generalized by Ansah [2000] for the direct estimation of gas-in-place. We have formulated the "Ansah" semi-analytical rate relation to in terms of the Arps' D- and b-functions for diagnosis and analysis. Using the definitions for the q-D-b functions we obtain the optimum values for the model parameters and as such we can estimate the gas-in-place. It is important to note that the "Ansah" semi-analytical gas flow relation is only valid for boundary-dominated flow. We also use the "power-law exponential" rate relation as proposed by Ilk et al [2008] to augment and valid our other analyses. The utility of the power-law exponential rate relation is that this generalized model can represent essentially any flow regimes (including transient and transition flow regimes). However; one should note that this relation is empirical and uncertainty in its application and results can increase for poor data quality. When applicable (i.e., when we have reservoir data and a competent pressure history), model-based analysis is applied to data to further validate the results obtained from the two semi-analytical/empirical methods. In this work we validate our methodology using a numerical simulation case and we estimate the reserves/ gas-in-place for a tight gas well completed in an HP/HT reservoir. Conclusions: We state the following conclusions based on this work: 1. The most important conclusion from this work is that for high drawdown cases, the value for the Arps b-parameter should converge to approximately 0.5 during boundary-dominated flow regime. 2. The straight-line linearization scheme proposed by Knowles [1999] and Ansah [2000] is validated for practical purposes including application to cases of HP/HT gas reservoirs. 3. The use of the D and b-data functions provides a unique insight to flow regime identification and the simultaneous matching of the data with the q-D-b models (both the "Ansah" model and the "power law exponential" model is a robust mechanism for estimating gas-in-place/gas reserves. However, this process (i.e., the computation of the Dand b-functions) is severely affected by the data quality therefore, vigilance is required when applying these methods to avoid the analysis of data artifacts.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

4. As the "Ansah" semi-analytical rate relation is only applicable for the boundary-dominated flow regime, in some cases it can be difficult to identify the onset of boundary-dominated flow. Therefore, we propose the use of the empirical power-law exponential rate decline relation in conjunction with the semi-analytical relation in order to achieve more consistent estimates of gas-in-place/gas reserves. In addition, if wellbore pressure data are available and an appropriate well/reservoir model can be identified, then model-based analysis should be performed to validate the results of the semi-analytical/empirical methods addressed in this work. We strongly suggest using all three techniques jointly to decrease the uncertainty/non-uniqueness associated with the assessment of gas reserves for wells in HP/HT gas reservoirs.

Nomenclature
Field Variables A b Bgi CA cf cti D Di D D i G Gp Gp,max h Jg k n p pi pwf rw rwa qg qgi
gi q

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Drainage area, ft2 Arps' decline exponent, dimensionless Gas formation volume factor (initial), RB/MSCF Shape factor, dimensionless Formation compressibility, psi-1 Total (initial) compressibility, psi-1 Arps' "loss ratio," D-1 Initial decline constant, D-1 Decline constant at "infinite time" [i.e., D(t=)], D-1 Decline constant, D-1 Original (contacted) gas-in-place, MSCF Cumulative gas production, MSCF Maximum gas production (t), MSCF Formation thickness, ft Productivity index, MSCF/D/psi Formation permeability, md Time exponent Pressure, psia Initial reservoir pressure, psia Flowing bottomhole pressure, psia Wellbore radius, ft Apparent wellbore radius (rwa=rw e-s), ft Gas production rate, MSCF/D Gas initial production rate, MSCF/D or STB/D Rate "intercept" [i.e., qg(t=0)], MSCF/D Skin factor, dimensionless Time, days Reservoir temperature, oF Fracture half length, ft Gas compressibility factor, dimensionless Dimensionless D-parameter Dimensionless fracture conductivity Dimensionless cumulative production, (GpD=Gp/G) Dimensionless pressure Dimensionless decline rate Dimensionless decline time

s t Tr xf z DD FcD GpD pwD qDd tDd

g g

Dimensionless Variables

Greek Variables = Rate-cumulative production relation characteristic parameter, =G/(qgi/Di) by definition = Gas specific gravity, dimensionless = Gas viscosity, cp

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

References
Ansah, J., Knowles, R.S., and Blasingame, T.A. 2000. A Semi-Analytic (p/z) Rate-Time Relation for the Analysis andPrediction of Gas Well Performance. SPEREE. 3 (6): 525-533. Arps J.J. 1945. Analysis of Decline Curves. Trans. AIME: 160, 228-247. Blasingame, T.A. and Rushing, J.A. 2005. A Production-Based Method for Direct Estimation of Gas-in-place and Reserves. SPE paper 98042 presented at the SPE Eastern Regional Meeting, Morgantown, West Virginia. 14-16 September. Bourdet, D., Ayoub, J.A., and Pirard, Y.M. 1989. Use of Pressure Derivative in Well-Test Interpretation. SPEFE 4 (2): 228-293-302. Carter, R.D.1985. Type Curves for Finite Radial and Linear Gas Flow Systems: Constant-Terminal Pressure Case. SPEJ. 25 (5): 719-728. Fetkovich, M.J. 1980. Decline Curve Analysis Using Type Curves. JPT. 32 (6): 1065-1077. Fetkovich, M.J., Vienot, M.E., Bradley, M.D. and Kiesow, U.G. 1987. Decline Curve Analysis Using Type Curves: Case Histories. SPEFE. 2 (4): 637-656. Johnson, R.H. and Bollens, A.L. 1927. The Loss Ratio Method of Extrapolating Oil Well Decline Curves. Trans. AIME 77: 771. Ilk, D., Perego, A.D., Rushing, J.A., and Blasingame, T.A. 2008. Exponential vs. Hyperbolic Decline in Tight Gas Sands Understanding the Origin and Implications for Reserve Estimates Using Arps' Decline Curves. Paper SPE 116731 presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, 21-24 September. Knowles R.S. 1999. Development and Verification of New Semi-Analytical Methods for the Analysis and Prediction of Gas Well Performance. M.S Thesis, Texas A&M University, College Station, Texas. Mattar, L., Gault, B.W., Morad, K., Clarkson, C.R., Freeman, C.M., Ilk, D., and Blasingame, T.A. 2008. Production Analysis and Forecasting of Shale Gas Reservoirs: Case History-Based Approach. Paper SPE 119897 presented at the SPE Shale Gas Production Conference, Fort Worth, TX, 16-18 November. Rushing, J.A., Perego, A.D., Sullivan, R.B., and Blasingame, T.A. 2007. Estimating Reserves in Tight Gas Sands at HP/HT Reservoir Conditions: Use and Misuse of an Arps Decline Curve Methodology. Paper SPE 109625 presented at the 2007 Annual SPE Technical Conference and Exhibition, Anaheim, CA., 11-14 November.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

Fig. 01a (Log-log Plot): "qDd-DD-b" plot. Definition of the DD and b-functions governed by the semi-analytical relation.

Fig. 01b (Semi-log Plot): "qDd-DD-b" plot. Definition of the DD and b-functions governed by the semi-analytical relation.

10

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 02 (Semi-log Plot): qg versus Gp. Flowrate data plot simulation case.

Fig. 03 (Log-log Plot): qg and auxiliary functions versus t. Diagnostic plot flow regime identification.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

11

Fig. 04a (Cartesian Plot): qg versus Gp. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

Fig. 04b (Semi-log Plot): qg versus t. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

12

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 05a (Cartesian Plot): D versus t. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

Fig. 05b (Semi-log Plot): D versus t. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

13

Fig. 05c (Log-log Plot): D versus t. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

Fig. 06a (Cartesian Plot): b versus t. Model validation plot almost constant behavior (at early times) for the computed bparameter (b0.5 for early times). Note the end-point effects are caused by numerical differentiation.

14

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 06b (Semi-log Plot): b versus t. Model validation plot almost constant behavior (at early times) for the computed bparameter (b0.5 for early times). Note the end-point effects are caused by numerical differentiation.

Fig. 06c (Log-log Plot): b versus t. Model validation plot almost constant behavior (at early times) for the computed bparameter (b0.5 for early times). Note the end-point effects are caused by numerical differentiation.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

15

Fig. 07 (Log-log Plot): "q-D-b" plot. example.

Application of the "power-law exponential" rate decline relation for the simulated

Fig. 08a (Log-log Plot): qg and Gp versus t. Analysis/forecast plot results from the semi-analytical and the power-law exponential rate relations are shown.

16

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 08b (Semi-log Plot): qg and Gp versus t. Analysis/forecast plot results from the semi-analytical and the power-law exponential rate relations are shown.

Fig. 09 (Semi-log/Cartesian Plot): qg and pwf versus t. Production history plot field example case. Note the well clean-up and liquid loading effects.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

17

Fig. 10 (Log-log Plot): qg and auxiliary functions versus t. Diagnostic plot flow regime identification.

Fig. 11a (Semi-log Plot): qg versus Gp. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

18

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 11b (Semi-log Plot): qg versus t. Model validation plot the data are matched with the proposed model across the boundary-dominated flow regime.

Fig. 12a (Cartesian Plot): D versus t. Model validation plot computed D-parameter data are matched with the proposed model across the boundary-dominated flow regime.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

19

Fig. 12b (Semi-log Plot): D versus t. Model validation plot computed D-parameter data are matched with the proposed model across the boundary-dominated flow regime.

Fig. 12c (Log-log Plot): D versus t. Model validation plot computed D-parameter data are matched with the proposed model across the boundary-dominated flow regime.

20

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 13a (Cartesian Plot): b versus t. Model validation plot erratic b-parameter data behavior caused by the data quality and numerical differentiation. Average match of the data with the model (b=0.5).

Fig. 13b (Semi-log Plot): b versus t. Model validation plot erratic b-parameter data behavior caused by the data quality and numerical differentiation. Average match of the data with the model (b=0.5).

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

21

Fig. 13c (Log-log Plot): b versus t. Model validation plot erratic b-parameter data behavior caused by the data quality and numerical differentiation. Average match of the data with the model (b=0.5).

Fig. 14 (Log-log Plot): "q-D-b" plot. Application of the "power-law exponential" rate decline relation for the field example.

22

D. Ilk, J.A. Rushing, and T.A. Blasingame

SPE 125031

Fig. 15a (Log-log Plot): qg and Gp versus t. Analysis/forecast plot results from the semi-analytical and the power-law exponential rate relations are shown.

Fig. 15b (Semi-log Plot): qg and Gp versus t. Analysis/forecast plot results from the semi-analytical and the power-law exponential rate relations are shown.

SPE 125031

Decline Curve Analysis for HP/HT Gas Wells: Theory and Applications

23

Fig. 16 (Semi-log/Cartesian Plot): qg and pwf versus t. Analysis summary plot (history match) field example case. Flowrate and bottomhole pressure values are obtained using the analytical solution for a fractured well with finite conductivity. Very good flowrate match and reasonable bottomhole pressure matches are observed.