:-.

.
Society of Petroleum Engineers

39927 “- ..

Practical Production Data Analysis

J. H. Frantz, Jr,, S. A. Holditch & Asswiat6s, Inc., J. P. Spivey, S. A. Holditch & Associates, Inc., and C. W. Hopkins,
S. A. Holditch & Associates, Inc.

~ lw( Sociaty
cfPetrch Eilg!s-raarl, k,

7his papar was ~ h ~amntatiin al W 1~ SPE Rx M~tain RagionaWow-

ParmeabiIity ReWira Sympm”m and ExWiin hld in Dam’ar, Colorado, 5-s April 1S98,
Introduction
This papar was Sa=ad ~r presentfiTyan WE Program COmmittaa folbwing reviaw of
information Wn- in an abstisrd aubrn~ by tha author(s). Contents of Iha papar, as
Many’ operators collect detailed information Ori their wells;’
Psmled,
~n
have mt baen raviawad by W S~ty
by the -s)
OFPetro!aum Enginaars and are s@act to
The material, as preserrtsd, doss not namsmrii raw any
production histories. This information can be analyzed to
~On M fhe SOCW M Pe~lem Enginears, its offi-s, or mernbars. Papars presentad at better understand the reservoir and we p~dent operating.
SPE maatings era * 10 pubr~ti rev-w by Edlwi81 Cwrmittaas of lha s~ty of
Petrolam Ersg”mm. ~ti-i “~ IM, &Win, or slwaga d my part of this paper decisions. The well data collected typically include daily or
fcf cmnmar~ ~ses ti fie written mnsant d tha Smiaty & Petmiaum Eng~ k
pmhibitad Parmiss”M to rap- in print k mstrictad to m abatrad of not more tharr 300
weekly gas, water, and condensate volumes, tubing, casing,
tiq tifratiom may @ bs mpisd, W Obstract md mntain mnspkuws and line pressures, days on line, choke settings, pl~ger-lif’(
~gmant & ~ ti by ~ h - was praaarrtad We LWars, SPE, PO,
SW S32S2S, Wdmrr, TX 7=-38X, U.S.A. fax 01-972-952-942S, cycle information, fluid levels, and occasional . .shut-in
pressures.
Some operators also perform pressure transient testing
Abstract occasionally to evaluate reservoir quality, wellbore damage or
This paper presents severaI production data analysis case stimulation, and reservoir pressure. In most cases, production
historia. The primary purpose of this work is to illustrate the data can be analyzed to provide similar information and
practical application of advanced production data analysis investigates more reservoir than short-term pressure transient
methods in low perm~bility gas reservoirs. Production data tests. Production data often show reservoir chtiacteristics
provide important information about reservoir quality and such as layered- or naturalIy-fractured behavior that are not
volume and the .s~on dectiveness. The analysis evident in pressure transient testing. A combination of
rdts can provide estimates for permeability-thickness production data and pressure transient analysis can be
product, skin factor or fracture half-Iength, gas in place, and effective tools to evaluate reservoir behavior. In this paper,
drainage area. Pressure trafisient information can also be we discuss several advanced production data anaIysis
integrated with production data analysis to further techniques and their application.
characterize the reservoh.
PrxH~pplications of production data analysis results
include ~ ~ort- and long-tern production rates, Analysis Methods
estimating reserves, designing/evaluating stimulation (and There are at least five objectives which may be achievd by
restimdation) treatments, predicting and evaluating production data analysis, pressure transient analysis, and
production increases from changes in flowing well pressure other means of formation evaluation. These are degree of
(pIunger-lifi and ,cornpressor installations), determining infiII stimulation, formation permeability, current average drainage
well potentiaI and optimaI ,weII spat.~ng, and identi&ing area pressure, original gas in place (OGIP) or drd-nage area,
natural &ture and/or Iayaed behavior, We will show and reserves. Since each analysis method ~ically
examples of the above applications using actual field data. investigates the reservoir at a different length scale,=some
New @ curves were also used in this work to analyze analysis methods are more useful for achieving a ‘given
difficult xts and to provide a starting point for more objective than are other methods.
sophisticated history-matching methods. The final portion of
this paper will include recommendations for field data Production Data Analysis. There are a number of
collection and database utilization. commonIy used methods for analyzing production @t.a,
including conventional decline curve analysis,~ adv&ced
dmline. curve analysis? automatic history matching, and
numticaI reservoir simulation.

231
2 J. H. FRANTZ, JR., J. P. SPVEY, AND C, W. HOPKINS 39927

Convmtionai Decline Curve Ana@sis. Conventional Outer boundary geometries include infitite-acting
din--e analysis was developed during a period of our reservoirs, closed circular reservoirs, and closed rectangular
industry’s history whti only high-permeability reservoirs reservoirs of various aspect ratios. Fig. 2 compares the
were moiitic-lo deveIop. Although it was originally production from a closed bounded reservoir with that from an
developed on an empincd basis, it has subsequently been itilnite-acting reservoir. Note specifically how production
givm a more rigorous theoretical basis for boundary from the infinite-acting reservok is concave upwtid, while
dominated flow. Many of the unconventional gas resources that from the cIosed reservoir is concave downward, once
b@ exploited today do not exhibit boundarydominated boundarydominated flow is achieved.
flow for a substantial period of time, perhaps even years. me
primary objwtive of conventional decline curve analysis is
reserves estimation.
Advanced Decline Curve Ana&sis. Advanced decline
curve analysis was developed by Fetkovich, who combined
transient solutions to the flow equation with the conventional
Arp’s decline curve equationl on a single type curve? This
type curve has been one of the most useful tools for
production data analysis of both high- and low-permeability
reservoti. ‘In ~dition to providing reserves estimates as does
conv*onal--twIine curve anaIysis, advanced decline curve
l-m
analysis also may provide estimates of permeability, skin
Tfme, days
&tor or hydraulic fracture dimensions, and original gas in
:-. .
place and drainage area . .
Analytical Solutions. In contrast to conventional decline
curve analysis, advanced dmline curve analysis attempts to Fig. 2- Effect of outer boundary geometry on production
model the production res~nse for a specific reservoir shape rate for constant pressure, hydraulically fractured
or production ‘mechanism. To this end, analytical solutions case.
fm a vti~ -ti-r=aoir types and boundary conditions have
been dti’eloped. Reservoir ~ include single-porosity systems, fieudo-
Inner boundary gmmetries include vertical wellbores with steady-state dual porosity systems,s and trantient dual’
skin? hydraulically-fractured weIIs with either finite- or porosity systems.6’7 Fig. 3 compares the produc~ion from’ a
infinite-conductivi~ fractures? and horizontal weIlbores? homogeneous, sirtgle-Iayer system with that from a sin@e-
Fig. 1 shows a comparison of production at constant Iayer, pseudo-steady state dual porosity system. In this figure,
bottomhole pressure from a hydraulically fractured well ‘to there are two periods of boundary dominated flow fm the dual
that from a vertical well with no stimulation. The fractured porosity r&servoir. In the first of these two pai~s, -the
weII exhibits a much steeper dmline in production at early fracture system is providing practically all of ‘he fluid
times than does the unstimuIated well; at late times, the produced by the weII, and the Iow-ptieabili&” matrix has not
tipes of the production decline for the two completion yet begun to respond. In the second period, the total”’system
methods are practicably indistinguishable. has begun to respond. ~is type of reservoir r&pons’e ?s not
unique to naturally fractured reservoirs; it can also tic~ for
layered reservoirs where inter-Iayer crossflow is present:
For pressure transient analysis, analytical solutions are
usually developed in terms of wellbore pressure at constant
production rate. For production data analysis, solutions in
terms of production rate or cumulative production at constant
flowing bottomhole pressure are usually more convenient.
Most of the commonly available solutions may be easily
expressed in either form. Superposition may be usd to
obtain pressures or production rates for wells which produce
at a series of constant rates or a series of constant pressures.
. .
.- —- mq d~s
,.

Fig. 1- Effeet of inner boundary geometry on production

rate for constant pressure case.
-.

232
 .
39927 PRACTtCAL PRODU~ON DATA ANALYSIS 3

presented as log-log type curves, where the dimensionless

production rate is plotted as a finction of dimensionless time.
A manual type curve matching procedure, similar to that used
in pressure transient analysis, allows the appropriate
parameters of the particular reservoir model to be obtained
from the rate and time match points, and from the stem
matching parameters.z In addition to rate-time ~ curves,
type curves for cumulative production vs. time and rate vs.
cumulative production are often usefi.d.
me analytical solutions we have discussed were originally
developed under a number of rather restrictive assumptions.
Virtually all assume slightly compressible liquids, single layer
Thne, days reservoirs, produced at constant bottomhoIe pressure.
Fortunately, a variety of tools have been developed to dow
Fig. 3- Effect of @“@~roJi~ system on production rate these analytical solutions to be applied to a muck w-der
for consiant pressure case. variety of systems.
Appropriate definitions of adjusted pressure and adjusted
Auother form of solution is that of production rate for a time allow soIutions originally developed fm ‘slightly
linearly varying bottomhole pressure.8 ~is type of solution compressible liquids to be applied to gas reservoirs*” and even
may also be easily obtained for most reservoir geometries. to dewatered coal and naturall y-fractured shale reservoirs. 1*
Superposition may be appIied to obtain solutions for wells Other tools, including material balance deconvoIution,12
where the wellbore pressure changes continuously in a constant rate analog time,13+*4 and constant pressure analog
piecewise Iinmr fashion. time*5,allow production at variable rate and variabIe pressure
Fig. 4 compares the production for a welI produced at to be analyzed using techniques developed for analysis of
c~ pm to that where the welIbore pressure declined production at constant rate or constant pressure.
linearly fbr the first 30 days, then was held constant. Afier Automatic History Matching. Automatic history
three months or so, the difference in production between these matching analytical modeIs combine a forward method to
two cas= is rninimaI, Unfortunately, this poses a real predict reservoir pdormance for a given set of res~ir
problem in distinguishing the effwts of near-welIbore conditions, with a numerical procedure for changing the
conditions, i.e. stimulation, and reservoir effects such as parameters of the modeI to improve the agreement between
permeability. Most of the information distinguishing near- the performance predictd by the model and that obstied in
welIbore conditions and reservoir effects occurs early during the field.16 Automatic history matching can provide estimates
the Iife of a well. It is during this period that wellbore of formation properties with models which are too unti”eldy
Pressuramleast Iikely to obey the ideal constant pressure to handle with manuaI type-curve matching using paper type
assumed by the model, for a variety of reasons. curves, such as dual porosity and layered reservoirs.
Automatic history matching also aIIows the user to fix the
values of the parameters in which he has confidence, and
obtain a match by varying the unknown or uncertain
parameters.
Finite-Dlflerence Numerical Simulation. For even more
complex situations, finite-difference numerical simulation can
be used. Finitedifference simulation allows the engineer to
model reservoirs of arbitrary shape. It also allows simulation
of pressure drop in tubular goods and wellbore dynamics. It
can model reservoir effects such as non-Darcy flow or
pressure-dependent reservoir properties. Finally, it can model
1 1 r
multiphase or compositional fluid systems.
10 I J
-1 To Too low 1000O Di@cul~ of Separating Permeability and Near- WelBore
~ dqs Effects. me effects of permeability and skin ,factor on
production rate differ most at early times. At Iate times,’once
Fig. 4- Effect of bottomhole pressure history on production boundary-dominated flow has been established, the e~w-ts of
rate. permeability camot be distinguished from those of skin
&tOr. Thus, the earIy time data is extremeIy important for
~pe Curves. These analytical solutions are often obtaining unique estimates of permeability and skin factor
 ,.=-
4 J. H. FRANTZ, JR., J. P. SPVEY, AND C. W. HOPKINS ‘39927
...

-taneously. In order to use production data anaIysis Data Collection and Database Utilization
,.
tmhniques to obtain these estimates, capturing earIy
production rate and welIbore pressure transients is essential. Data Collection. Because of the importance of early time
Deterinining C3riginal Gas In Piac6 Production data data in distinguishing permeability and skin factor from
analysis is particularly suited to estimation of original gas in production data, our recommendations for data collection
place, once boundarydominated flow has been achieved. focus on accurate collection of early transient data.*7
Since the shape of boundary dominated production is For all weIls following completion, recompletion, stirntilation,
,..,..—
independent of the shape of the reservoir, production data or a change in backpressure, we recommend as a ~nimum
analysis can often provide accurate estimates of OGIP, even collecting production volumes of all phases and weIIhead
when pcrm=bility and skin factor cannot be determined pressures on a daily basis for the first month and on a weekly
independently. If good estimates of net pay thickness, basis for the next five months.
~sity, and water saturation are available, accurate Later in the life of the well, for a weIl producing
estimatm of drainage area can then be obtained. continuously against a constant back pressure, production
The time required to achieve boundarydominated flow is volumes of all phases and wellhead pressures shouId h
strongly dependent on the reservoir size, the formation collected on a monthIy basis. If production is interrupted, the
pmeability, and the compressibility of the fluids in place. number of days on line should also be recorded. -
In a high-permeability, undersaturated oil reservoir of limited For wells which are periodically curtailed for any reason,
areal extent, boundary-dominated flow may be achieved in a in addition to monthly production volumes and days on line,
matter &“minutm or hours. In a low-permeability, low routine data collection would ideally include wgllhead
pressure gas rmbti of large areaI extent, it may take years pressure at me end of the flow period, before shutting the weII
to achieve boundary dominated flow. Table 1 shows the time in, and the wellhead pressure at the end of the shutin peri~,
required to reach boundarydominated flow (stabilization before putting the weII back on production.
time) for a w;ell compIeted in a gas reservoir containing 0.67 -..
graviw-@ar140 “F and 2,000 psi initial pressure.17 Database Utilization. As the number of wells in the analysis’
increases, it tiII be nwessary to put the data in a data~ase for
Pressure Transient Analysis. Pressure transient analysis proper data management, especially if a Moving ~ornain19
(PTA) is much more usefil than production data analysis for type of analysis is performed. What we find most
investigating the near-weIIbore region. In pre-fracture tests, advantageous is to put production data, completion
PTA can readily distinguish the effects of skin factor from i~ormtion, well location information, and any other.
f~ ‘- perrrfeability.’s In post-fracture tests in low- pertinent data into a relational database such as Mcrosofi
perrtieabili{y reservoirs, PTA can provide estimates of fracture ACCESSm. In this type of database environment, we have
Cetivity and fractike half-length, provided an separate tables for production data, completion data, weII
independent estimate ofptieabiIity is available.18 location data, and operational data such as flowing “tubing
! pressures. We link the data so that we are able to identifi the.
Moving Doma=”Analysis. Moving domain analysis*9 takes production from any individual completion and also locate
a much broader, more statistical view of production data. It that well on a map. Once this data is in a relational database,
US= pubIicly available pr~uction and completion data to it is quite easy to extract subsets of the data for individual
identifi areas of interference between existing wells and to well analysis or to perform bulk processing to ,-caIcuIate
quantifi the impact of completion and stimulation practices parameters such as production indicators, ctiulative
on ti~l perfortice and drainage area. It may be used to volumes, and average production profile:. Once the &ta is in
quan~ ‘rnfiI1 &illing opportunities over large areas, to a relational database, spreadsheets such as Microsofi HCEL
evaluate completion practices and stimulation treatment sizes, can be used to perform the more complicated calculations.
,,
ad to locate areas for more indepth, detailed engineering Simple calculations can be performed wholly within the
studies. relational database. Typically, if certain features for each
Although moving domain analysis may be performed with individual well are caIcuIated, such as production indicators,
pubIicIy available production data alone, a combination of we will store these resuIts in the relational database. It is also
public and operator-suppfi~ data may be required for an easy matter to export data from the relational” database as
c~ietion or dirnulation anaIysis. Moving domain analysis ACSi’I files so that tiey can be imported into other so~are
rdso prow-d& an unbias”~ and consistent analysis across a packages. We find this works welI for the GEOGRAPW
study arm-.--Fifially, moving domain analysis is cost effective, mapping package. Once the data is loaded into the relational
i.e., m““some cases, moving domain anaIysis has provided database, we apply aII of our Moving Domain tools t&quickly
es-y the same ‘ “resuIts as much more detailed evaluate wells and make numerous comparisons. This-m of
@*-g studies at a mll fraction of the cost in terms of evaluation would not be possible without the data “in some
both dol~~s and rnruipower. type of database form.

234
 ,.

399D” PRAcTtCAL WODUCmON DATA ANALYSIS 5’

— . . .—.
..
As with &y engineering &ta collection, the cost of 80 days of production until the flowing pressure reached line
obtaining the data must h. weighed against the potential pressure. ~is caused a linearly decreasing flowing ~ressure
bmefit to be gained from its acquisition, history and the increasing rates up to 35 &ys. In the
analytical model, we used a linearIy varying flo~ng ~ressure
and varied kh and Lf to match the data. We es-d k of
Application of Production-Data Analysis ‘“ 0,045 md and an effective fracture half-length of 150 ft. The
h this ~wtion, we present numerous individual well histories estimated permeability is reasonable for these formations and
to illustrate the practical applications of production data the fracture half-length consistent with hydrauIic <fracture
analysis. We used an analytical mode120in this work that modeling results for this treatment, Due to the ~flo@g
assumes a tigle-Iayer, homogeneous, single-phase reservoir. practices, the well does not appear to b: si~-ficktly
The model includes automatic and manual history matching stimulated (lack of steep initial fallo~, but it is highly
..
capabilities. stimulated.
A. Estimating Future Prodution Volumes and Reserves Figs. 5 and 6 show three, five year forecasts for the well.
We assum~ 4Q and 80 acre drainage areas for.ti~ model
1. Infiite-Acting-Data forecasts and used an arbitrary exponential decIine throi’~
Production data can & analyzed to estimate fiture production the last four datapoints for the decline curve forecast. The
volumes and reserves with a minimal degree of effort. It is difference in the three forecasts illustrate the impofiance of
corrmt, however, to ——— extrapolatee only boundary-dominated using boundarydominated data to accurately predict. Iong-
production - b“~a~or to forecast fiture performance. term recovery. An exponential dmline should onIy be used
Depending on the data trend, forecasting from infinite-acting when boundarydominat ed behavior strongIy influence the
data can lead to optimistic or pessimistic future production production response.
estimates. A gas in place (drainage area) estimate should k
assumed if for~asting from i~lnite-acting data. 100i)]

Appalachian Basin gas reservoirs are typically “tight”,

Iow-ptieability formations that must produce for several
~med
years before boundary-effects are evident. In practice, a ~
reserve estimate may be assigned when new welIs are brought f- 4 14% Exwential ~dlne

on-line based on a combination of the first monthslyear ~. ~ 100

80 Acre Drainw Area
actual production and the AFE’d reserve estimate. me %
a -------
.......Y. .. . ---
AFE’d estimate is ofien obtained from offset well analogies. ..................
. ......... ...........
b addition, simple declin?. curve extrapolations or “fieId” 40 Awe D&inage Area
type curves are ofim applied for forecasting and reserves
estimates from short-term data in lieu of more detailed 10
02C0400SOOSO0 loco lm 14W lm II m
analysis using analytical or finitedifference reservoir models.
..—.. TIMS, days
The fiple in this section shows several forecasts Horn
infiite-acting data to illustrate the range of forecasts that Ffg. 5- Gas rate projections for Ostrander No. 2.
..
cotid be obtained from the historical d~a. We analyzed.100
days of production data from Beld~ & Blake’s Ostrander-’No. If the exponential decline were used pr-tureIy, before
2 well located in @utauqua Cotity, New York for the boundary effects became evident, an erroneous forecast
puwse of projecting fiturk volumes and evaluating reservoir results. In this example, if a steeper exponential decline had
quaIity and effm~ivk fracture haIf-length, This short-term, b- used, a much lower forecast results. me 20-year
infinite-acting data is usefbl to estimate reservoir quality projections for this weIl indicate estimated recoveries of’ 133
(pmeability-thickness product - kh) and stimulation MMscf, 214 MMscf, and 223 MMscf for the 40 acre, 80 acre,
effectiveness when sufficient flowing pressure data is and exponential decline cases. The exponential decline
collected. The Ostrander No. 2 is compIeted in the low- fmecast shown causes an increase in estimated production
permeability Med[na/WhirIpool formation at 3,200 ft and was rate, recmery, and economics for the well since the
fractured with 100,000 Ibs of proppant with a nitrogen-foam production forecast is based on only the itilnite-acting data.
treatment. Table 2 shows the reservoir properties used in the .,,
analysis.
.,: !., -. . .
Due. ~ ~ short production period, the data e@iited
infinite-acting behavior.r ~gs. 5 and 6 show our rate and ,..
cumulative matches of the data. The, choke settin~s were
gradually increased to tise the well into line during the first
.-.
.
235
6 J. H. FRANTZ, JR., J. P. SPVEY, AND C. W, HOPKINS . . . 39627
————
.=
flowing pressure stabilized at approximately 200 to 400. psi,
the gas rate declined steeply.
..

Iom

lom
! !
4# ,00
a
!
*

nmi?, day

Fig. 6- Cumulative gas projections for Ostrander No. 2.

Fig. 7- History match and forecast for Huff 2 well.
,, . ..
2. Boundary-fimiriated Data
As discussed above, fiture volumes can be forwasted
accurately when boundary fiats significantly influence
production data. IrI addition to forecasting fiture rates, a gas
in place and a drainage area estimate can dso lx obtained
when analyzing boundarydominated data.
b this exampIe, we anaIyzed 275 days of production from
the Beldcn & Blake Huff No. 2 well. This well is completed
at 6400 fi in the Rose Run reservoir in Stark County, Ohio.
The well was initially acidized and produced at gas rates 0.1
ranging from 600 to 2,000 Mscf~ using a linearly decreasing
i
bottomhole pressure history. ~g. 7 and 8 show our history L
‘Ofi,
~
match and forecast fm the data. Fig. 7 is a semi-log plot of O.0001 0.001 1
the gas rate and ~g. 8 is a ~ curve Iog-log match of Olmunionlsss M “ -

dimensiordesi rate versus dimensionless cumulative

production. Note the downward bend in the log-log plot Fig. 8- Type curve match of Huff 2 well.
..
indicating boundary effects @ve significantly influenced the ..—.
production data. ‘~e--~ curve accounts for changes in
producing pressure and material balance over time. An
accurate future fwecast can thus be made. In this case, the 1s00

ROSeW“is a high permea6iIity reservoir (k = 0.77 md) with 1400.

a ~ fw.tor of -3.9 and a dainage area of 43 acres. The 1200.

permeability and skin factor are non-unique but within Install Plunger
r~onabIe r~ges given ,. ou~’knowledge of the Rose Run and
the acid breakdown. Table, 3 shows the reservoir properties
used in our analysis.
We alw reviewed tbe production data from a well c1
o
complq@ in the Rose Run formation in Licking County, o so 100 lm 2 D

O~o. The well was initiaIIy acidized and balled out. The TIM, days
.-
p~se of our evaluation was to estimate the reason for the
sharp production dmline recently observed. It was believed Fig. 9- Flowing pressures from well.
the well may be damaged. Fig. 9 shows the actual flowing
tubing and casing pressure history. A plunger was used
periodically to unload the tubing and was permanently We matched the production data varying k, s, and Al me
installed at approximately 150 days. As noted in Fig. 9, the matches are shown as Figs. 10 and 11 (type c~e ~tch).
well was produced in a ‘manner to reduce the flowing Based on our analysis of these data, the steep ~roduction
pressures gradually (linearly varying). This resulted in near- decline occ~ed because the well is completed. in a high
constant production rates ranging from 300 to 500 MscfD as permeability (1O md) reservoir with a liniited reservoir
observed from the actual ~’orical data in Fig. 10, After the

236
 S9927 PRACTICAL PRODUCTION DATA ANALYSIS 7

volume (81 MMsc~. A presme buildup test was pdormed

and verified the average reservoir pressure estimated from 1000,
Skin Factor -.3.5
production data analysis. Table 4 shows the reservoir Hkto~
Match
propfies used in our analysis.
i
.

5 -wed Dara “
100
Skin Factor. +2.64
E ●

Hist~
I !

loo~
200 400 600 600 low 12m 14m 16m lnm z
Tim, days
Observed * ●

Fig. 12- History match and forecast for

Quigley 2B weli.
9
1001’’’’’’’’’’’’’’’’’’’’”
#20mw* 100
“’’’’’’’’’”[
120 140 160 160
Figs. 12 and 13 shows the gas rate and cumulative
‘T&D, d~ prduction history match and three fiture forec~t~.
Boundary effects were evident in the data so an accurate
Eg. 10- Eistory match of well. forecast could be made. The two higher fgepasts ,a~me pp
,..
acid treatment could be performed to remove the yin damage
. . .. and slightly stimulate the weIl (s = -3.5). The fortia,sip
indicate an acid treatment could improve, and accel~ate weil
productivity during the next fm ymrs. ~ese two post-acid
#.-
‘;::’...,
“....., forecasts also show the impact of timing. If the well is

1
10 .... ......
..... .““..,,....,,,;...,..
. . ,.,,
.........?....,. observed acidized early in its life at 425 days, the production is
:.-:fi::*.;~*%*:m.ti.,.,*,,,, ~+,.
1 ““““”,*.+
*.. accelerated substantially compared to waiting until day 700 to
0.1 Type CUWaS perform the treatment. This example also assumes the well
\ can be acidized without fracturing down into any water zones
below the Beekrnantown.
IE.00S O.ml 0.001 0.01 0.1 1 10
~lmmionIees Time
——

I
360 . Win Factor --3,5
Fig. 11- Type curve match of well,
200. .

B. Evaluating Stimulation
. Treatments 260. .
Skin Factor - +2.64
200 .
1. Acid Treament” .
l!m .
This examplei~ustrates the use of production data analysis to
lW .
forecast po~tial “wellimprovement from an acid treatment.
The A. W. Tipka Quigley No. 2B well is an openhole m.
Cbs8wed Dara
Beekmantown completion in Tuscarawas County, Ohio. The o
0 2m4m5006m looo120014m 16m1sco2 m
weIl was drilled in 1993 to a depth of 75008. The well was ,.
TIK,days
flowed using a linearly decreasing pressure which kept rates .
WIy constant during the first year as shown in Fig. 12. Fig. 13- History match and forecast for
A pressure buiIdup test was run’ and analyzed in earIy Quigley 2B well.
1995 to evalmte if welIbore damage existed. We
subsequently matched 700 days of production data and 2.” Optimal Fracture Half-Length
obtained consistmt results compared to the buildup for k Ashland Exploration, Inc. participated with the Gas Research
(0.54 md) and s (+2.64), show@g the well was sIightIy Mtitute (GM) in a study of the Berea formation @ Logan
damaged. Table 5 shows the reservoir properties used in our County, West Virginia, The purpose of the study was to
amdysis. -, implement GM’s Advanced Stimulation Technology. Several

237
8 J. H. FR~~, JR., J. P. SPVEY, AND C. W. HOPKINS
___+..Y
39927
.
. . . . . _..—..-..__.’
..-. -. .
test wells were seIeeted for pre- and post-fracture analysis.
The Pardee Land Co. No. $7 study well is the 3800 ft Berea 3500.

example illustrated here for designing an optimal size 3000 .

. .—
Stimtition treatment.
me Baea was brok~ “~ownand produced for s~ren days
befbre performing a pr~:sure buildup test to obtain a reservoir
description. The weII averaged approximately 73 Msc~
during the pre-~acture ffow test. A 75,000 lb nitrogen-foam
1000 / \
stirm-datioritreatment was pumped in the well and de$gned to
achieve ‘“a 150 R fracture half-Iength based on fracture 500
t
modeling. Post-fracture production data and pressure buildup 01
analysis indicated that the treatment created an effective 0 50 100 150 200 250 <)00
.$

ficture haIf-Iength of approximately 100 ft. Table 6 shows FRACTURE HALF-LENOTH, ft

l=--
the reservoir properties used in our analysis. g ___
Fig. 14 sliows pre- and post-fracture production data and Fig. 15- Example fracture half-!en~th
a history match of the poii-fracture ‘data using a 100 fi half- optimization plot.
Iength. We,aIso show the predicted performance if the 150 ft .. .
half-length had be~ achieved. Fig. 15 shows an example Meridian Exploration Corp. operates Queen;ton for~tion
fracti haIf-l;ritih optimization plot. The NPV is calculated wells in Senec~ and- Cayug~ Counties, ‘New York. ~ese
for each predict~ flowstream using the pre-fracture restioir wells are completd at 2,000 R and produce associa~~ brine
description,
. . W“OUS‘fi~e haIf-Iengths, and accounting for water. The example shown in Fig. 16 is for Metidian’s
the diffe;ent stimulation costs. This plot is not specific to this Delaney 1087 well that had produced for 1,400 dafi” before
treati~t,-but illustrates that fracture haIf-length can ‘b installing a casing plunger. The bottomhole flowing pr~sure
optimi~~using this meth~ology. was estimated at 350 psi before adding the plunger and’ 150
psi after. We history matched the data to ~orecast ~fiture
production and to estimate the impact of addin~-cornprkssion
..
lao to rtiuce the pressure to 75 psi at 2,700 days: The pfimary
utility fm Meridian of course is to evaluate gas rate

i
improvements on other plunger lift candidates in the field.
Table 7 shows the reservoir properties used in our analysis.

‘“t=‘“--
# Icm
“’’’
..... Pmdiid

: ------------
~
Li-lsoft

. . . . . ... ... . . .. .. .. .

! &wed ~wed
$ “ ~Btoy[a,chL,.lm,
Pm-Frac Post-Fmc

Compression
~w
a

F Obsefved ‘.-
Fig. 14- Pre- and post-fracture Berea production 4
10
data horn the ~ Pardee Land 87 weI1. 0 w“ 1000 1s00 2000 qo 3W0
..
. ..+
Ttme, day ‘=
C. Plunger-Lift and Compressor Installation . ,.
.—
Fig. 16- History match and forecast for
Many Appalachian Basin weIIs produce fluid (water or Delaney 1087 well.
condensate) along with the gas. Tubing or casing plungers ..-
are ofien M to IiR” these fluids and maintain a low D. In@l l~ell Potential
bottomhole pressure. Production data can be analyzed to
predict the increased gas volumes associated with adding Noduction data can be analyzed to estimate drainage area
plunger-lifi if the flowing pressure history is available. and infill well potential using some assumptions for
Compressor insfalIation- can aIso be evaItited to estimate the volumetric, i.e., net pay thickness, @rosity, etc~ We
gas ____ ent ~ Iowering the bottomhole producing evaluat+8 Z years of production data from the OHO L&M
pressure. fiuck No.’3 well located in ~ashington Codty, ‘O~o. “tie
well is completed in the Clintotiedina formations at 5,900

238
 -—,-
39927 PRACTICAL PRODUCTION DATA ANALYSIS ._ 9

----
R. Reanalyzed theproduction data toestimate k,s, and A. produced for 16 days measuring daily volumes and
Fig. 17 shows t~e data is significantly affected by boundary tubing/casing pressures. The well prtiuced at a folly
effws. -ant rate of 197 Msc~. A five-day pressure buiIdup test
was also performed at the end of the flow period.
F%. 19 shows our production data analysis resul~~. We
matched the production data aIong with the pressure %uildup
data to estimate kh, s, and initial reservoir pressure. From
i
these data, we forecasted future performance using similar
well properties and well spacings of 40 to 320 acres. We
Obsetved “.
performed a simplified before federal income tax (BFIT)
.
economic analysis to calculate a net present value for each of
F the performance predictions. We then calculated the net
G ,~
1 10 100 Im lomo
present value per acre for each case and plotted it versus well
llme. dav
spacing as shown in Fig. 20. Using net present value per acre
Fig. 17- ~story malch of Huck 3 well. as an optimization criteria, our results indicafe”that a well
spacingof approximately 160 acres is optimal for this cake.’
From our production data analysis in Figs. 17 and 18, we
estimated k (0.08 red),s (4), Md an original gas in place of u:
230 MMscf. Based on the volumetric parameters assumed in

‘m
gj :
our analysis, this equates to ti effective drainage area of 26
acres. Since this’ area is driIIed on 100 acre spacing, it may 8
be possible to drill idllI weIIs. Further anaIysis is History
Obsewed
recommended before drilling a test well including a geologic. Match
8$
assessment tmd log anrdysis to vaIidate the volumetric
assu~”~; - Additional CIinton/Medina wells in this field k“
*
-“”--+-””----i-

could be anaIyzed to tither estimate drainage areas. In a 2

I I I I 1 1 1
100
large fieId with many producers, a moving domain studylg 02 4 6 6 10 12 14 16

could be pdormd in addition to individual well production Time, day “- -

data analysis to further quanfifi infill potential. Table 8

shows the reservoir properties used in our analysis. Fig. 19- History match of G. W. Kazee weII.
t.
. ..-.
.
m
Iao
160 t
140 History
120 Wtch
100 Observed
60
2
so
40
~ 1
Qtbnal Future Well ~acing
n
2
:~
Omolooo lm20w25imsooom
nm, dw

Fig. 18- &story match of Huck 3 well.

se lti - 2W “m
,,. .- ,~.+—,
E. Optimal \Vell Spacing WELL S.PACING, ACRES
:,, .-~,

Production data can k analyzed when new wells are brought ‘Fig. 20- WeII spacing optimization piot. “ ‘
... ..
on line to perform a well spacing optimization study. In the
Appalachian Basin, most weIIs are developed on 160 acres or F. Identi&ing Layered and Dual-Porosity EeKa-tier
-,
smaller well spacing. For any given reservoir, the true
optimal weIl spacing depends on storage capacity, flow This example illustrates the production response ‘fiorn=a weII
capacity, and economic considerations. producing in an acidized, layered, Rose Run reservoir. Fig.
~s example is from the G. W. Kazee #l well completed 21 shows the production data, calculated bottomhole flowing
in the Rome Sqd @ Elliott County, KY. me well was pressure, and surface choke settings for 370 days of

239
 ,;—.
.,
10 J. H. FRANTZ, JR., J. P. SPVEY, AND C. W. HOPKINS 3*27

productim” data from the A. W. Tipka Oil & Gas

Swartzentruber No. 2-A well, located in Wayne County, -----
Ohio. me log through the Rose Run at 6400 R appears to E Obseirved

indicate two md layers s@arated by a shale break. The Iog

shows a high and low porosity layer. The well produced at
rates _%orrt 500 to 1,200 Mscf/D for approximately
170 days and then exhibited a steep decline as shown in Fig.
21. The production rate stabilized at a near constant 120
~~ until” the present time. The well was produced by
‘“0.1 1 10 100 1000
gradually increasing the choke settings resulting in a IinearIy
Tim, day
decreasing pres=e- history for the first 220 days. At
appro-tiRIy 220 days, constant pressure production began. Fig. 22- History match of Swartzentrubr 2-A we~.

.-

- 1.4 CONCLUSIONS
- 1.3
1. Production data, flowing pressures, and all well
information can be collected and stored in a database to
- 1.2 ~
provide easy fiture access.
- 1.1
i 2. Numerous tools are available to analyze production data
- 1.0 ~ and estimate formation permeability, degree o
o
- 0.9
stimulation, current average reservoir pressure, original
gas in place, drainage area, and reserves.
, 0.8
3. Production data analysis can be used routinely as a cos
‘ -- io~d, Iw 200
effective method to predict short- and long-term rates,
.-=ti ——
M days estimate reserves, design stimulation treatments, predict
production increases from changes in from-ng weI
Fig. 21- Production and flowing pressure pressures @lunger-lift and compressor installations),
data for-%artzentruher 2-A well. evaluate optimal well spacing, and identi$ layered
and/or natural fractured behavior.
We modeled the first 230 days of production to obtain an
~ivaImt singIe-Iayer reservoir description for the completed
intervaI. This match yielded a permeability estimate of 4 md NOMENCLATURE
and a small drainage area of 17 acres. Boundary effects were k = Permeability, md
evident based on the log-log plot of rate versus time shown in kh= Permeability-Thickness Product, md-fi
Fig. 22. Note in the match the forecasted production data = Skin Factor, dimensionless .
MIs well below the actual stab~ized gas flow rates from&@ i= Drainage Area, ac
230 to 370 ~Ig. ~1). The 120 Mscf~ can & maintained if a Average Reservoir Pressure, psia
F, =
secotid, lower permeability layer is contributing to -,. .
production. We investiga~ed this scenario using a 2-layer,
finite-difference, reservoir simulator capable of modeling
multiple layers with different drainage areas. We successfully ACKNOWLEDGEMENTS
We would like to thank the numerous companies who allowed
history matched all the flowing pressure and production data
using a 2-layer model incorporating a high permeability and a us to use the examples shown in this paper including Ashland
Exploration, Inc., A. W. Tipka Oil & Gas, Belden & Blake
low permeability layer. The Iower permeability Iayer appears
to be itiite acting, Our analysis results are non-unique, i.e., Corporation, Carson and Associates, Eastern States
Exploration, Meridian Exploration Corp, and Ohio L&M.
the production data and pressure history could also be
explained by other combinations of high and low permeability
layers at different drainage areas. A naturally-fractured
REFERENCES
reservoir could also exhibit the same production profile as the
1. Arps, J. J.: “Analysis of Decline Curves,” fians. AIME
layered example shown here, however, the stabilization point
is ~icaIly much earlier than 220 days at these high rates. (1945) 160,228.
2. Fetkovich, M. J.: “Decline Curve Analysis Using Type
CSuves,” &T(June. 1980) 1065-1077.
3. BIasingame, T. A., and Lee, W. J.: “Prop&Tes of

240
39927 PRACTtCAL PRODUCTION DATA ANALYSIS 11
. . -+ .--—

Homogeneous Reservoirs, Naturally Fractured presented at the Gas Technology Symposium, Houston,
Reservoirs, and HydrauIicaI1y Fractured Reservoirs From Texas, January 23-24, 1991.
Decline Curve Analysis,” paper SPE 15018 presented at 16. Murtha, J. A., Gatens, J. M., Lancaster, D. E., Lane, H.
the Permian Basin Oil and Gas Recovery Conference, S., Lee, W. J., Olarewaju, J. S., and Watson, A. T.:
Midland, Tetis, March 13-14, 1986. CTractical Methods for Analyzing WeII Test and
4. MutaIik, P. N., and Joshi, S. D.: “Decline Curve Analysis Production Data from Devonian Shale Reservoirs,w S. A.
Predicts. Oil Recovery From Horizontal Wells,” OGJ Holditch & Associates, Inc. Topical Report to the Gas
(Sept. 7, 1992) 42-48. Research Institute, GRI Contract No. 5084 -213~0986,
5. Da Prat, G., Cinco-Ley, M, and Ramey, H. J. Jr.: Oct. 1987.
“D&line Curve hlysis Using Type Curves for Two- 17. Spivey, J. P.: “Production Data: Advanced Analysis
Porosi&Systems,” FEJ[J une ~ X41362. Methods,” Petro Systems World. May/June 1995,26-28.
6. Kazemi, H.: ‘Tressure Transient Analysis of Naturally 18. Lee, W. J., Rollins, J. B., and Spivey, J. P.: Presst4re
Fractured Reservoirs With Uniform Fracture Transient Testing, SPE Textbook Series, in preparation.
Distribution: SPEJ (Dee. 1969) 451-561; Trans. AlME, 19. Voneiff, G. W. and Cipolla, C.: “A New Approach to
2%. Large-Scale InfiIl Evacuations Applied to the Ozona
7. de Swaan, O. A.: “AnaIytic Solution for Determining (Canyon) Gas %nds~ SPE 35203, Permian Basin Oil &
Naturally Fractured R“&ervoir Properties by Well Gas Recovery Conference, Midland, TX (1996).
Testing: WEJ(June 1976) 117-122; Trans. AI’E, 261. 20. “PROMAT for Windows, A Production Data History
8. Spivey, J. P., and Frantz, J. H.: “History Matching Matching and Performance Forecasting Program for ~
Production Data Using Analytical solutions for Linearly and Gas Wells,” S. A. Holditch & Associates, Inc. User’s
Varying Bottomhole Pressure,” paper SPE 29167 Manual, Version 1.04, 1995.
presented at the SPE A“fiual Technical Conference and
Etibition, New Orleans, Louisiana, September 25-28,
1994. _._L__-—.—..——-—
9. Lewis, P. E...—and Hoff, J. G.: “Rate-Cumulative Type .
Cties for Forecasting ‘the Performance of Curtailed
WelI~w-paper SPE 1899U presenied at ‘tKe-To~W&ky
Mountain “‘-‘m-@ihw Permeability Reservoirs
Symposium ani Exhibition, Denver, Colorado, March 6-
8,1989.
10. Fraim, M. L., and Wattenbarger, R. A.: “Gas Reservoir
Decline-Curve Analysis Using Type Curves With Real
Gas Pseudopressure and Normalized Time: SPEFE
(Dec. 1987) 671-682.
11. Spivey, J. P., and Semmelbeck, M. E.: “Forecasting
Long-T- Gas Production of Dewat ered Coal Seams
and Fractured Gas Shales,” paper SPE 29580 presented
at the SPE Rocky Mountain Regional/Low-Permeability
Reservoirs Symposium, Denver, Colorado, March 20-22,
1995. ——...
.————
12. Johnston, J. L.: ‘variable Rate Analysis of Transient
Well Test D;ta Using Semi-Analytical Methods,” MS
~esis, Texas A&M University, August, 1992,
13. Blasingame, T. A., and Lee, W. J.: “Variable-Rate
Reservoir Limits Testing? paper SPE 15028 presented at
the Permian Basin Oil & Gas Recovery Conference,
Wdland, Texas, March 13-14, 1986.
14. Blasingame, T. A. and Lee, W. J.: “me Variable-Rate
Reservoir Limits Testing of Gas Wells,” paper SPE
17708 presented at the Gas Technology Symposium,
Dallas, Texas, June 13-15, 1988.
15. Blasingame, T. A., McCray, T. L., and Lee, W. J.:
“Decline Curve Wysis for Variable pressure
Drop~ariable Flowrate Systemsy paper SPE 21513

241
12 J. H. FRANTZ, JR,, J. P. SPVEY, AND C. W. HOPKINS 3W7

~.
,1--~, -.—”
“’ TABLE 1:
.,-. :.r: . .
EFFE~ OF PERMEABILITY AND DRAINAGE AREA ON TIME REQUIRED TO REACH
BOUND~Y DOMINATED FLOW

Permeabilityy Porosity Time to boundary dominated flow

for given drainage area
40 acres 160 acres 640 acres
1000 25.0 1.2 hours 4.7 hours 18.6 hours
.. 100 18.8 8.7 hours 1.5 days 5.8 days
10 14.1 2.7 days 10.9 days 1.4 months
1 10.5 20.5 days 2.7 months 10.8 months
0.1 7.9 5.0 months 1.7 years 6.7 years
0.01 5.9 3.2 years 12.6 years 50 years
~ 0.001 4.4 23.6 eMS 95 ears 380 YWS
.. .
TABLE 2

&=RVOHZ PROP&TIES USED IN OSTRANDER No. 2 EVALUATION

TABLE 3
.“
~SERVOIR PROPER= USED IN ~ NO. 2 EVALUATION

Initial Reservoir Pressure, psi = 2,790

Format ion Temperature, ‘F = 120

Porosity , 0/0 =
10
Water Saturation, y. = 0.35
Final Bottomhole Flowing Pressure, psi = 440
.
TABLE 4 ;,.

RESERVOIR PROPER~S USED IN ~ EVALUATION

Formation Temperature, “F . 110

Net Pay, fi = 26
Porosity , 0/0 = 16.8
Water Saturation, VO .
17.8
Final Bottomhole Flowing Pressure, psi = 325

242
 . .
3W7 PRACTICAL PRODUCTION DATA ANALYSIS 13

TABLE 5

~SERVOIR PROPERTIES USED IN QUIGLEY 2B EVALUATION

\ .—
Reservoir Pressure, psi
Initial = 3,400
Formation Temperature, “F = 130
Net Pay, R = 10
Porosity, Y. = 25
Water Saturation, VO = 20
Final Bottomhole Flowing Pressure, psi = 1,000

TABLE 6

RESERVOIR PROPERTIES USED IN PARDEE 87 EVALUATION

.. .
Initial Reservoir Pressure, psi = 780
Formation Temperature, “F = 95
Net Pay, fi = 15
Porosity, YO =
8.4
Water Saturation, Y. = 30
Final Bottomhole Flowing Pressure, psi = 155
-.

TAB& 7

~VOIR PROPERTIES USED IN DELANEY 1087 EVALUATION

Initial Reservoir Pressure, psi = 600

Formation Temperature, ‘F = 60
Net Pay, f? = 80
Porosity, VO = 7.7
Water Saturation, Y. . 40
~ = 150

TABLE 8

=SERVOIR PROPERTIES USED IN HUCK 3 EVALUATION

Initial Reservoir Pressure, psi = 2,025

Formation Temperature, ‘F = 126
Net Pay, R = 40
Porosity, VO . 5.2
Water Saturation, V. = 35
~ = 400

,.,

243