Practical Decline Curve Analysis

Greg Reep

Chevron Corporation
Corporate Reserves
San Ramon, California
January 2008

Version 2008.01
 Version 2008.01

PREFACE

Decline Curve Analysis (DCA) is the most common technique used throughout Chevron
for preparing economic forecasts and reserve estimates. Practical Decline Curve
Analysis was written to improve world-wide standards, efficiency and technical
understanding into DCA. Insights provided will prove useful to both the experienced
engineer who has spent a career using DCA and the new hire who needs a technical
reference.

As with any guide book in our complex and diverse technical industry, no document can
cover every technique or technical situation. The ultimate responsibility rests with the
engineer or the Qualified Reserves Estimator (QRE) to ensure that any work is
technically sound and correct.

This document should be viewed as a work-in-progress that will be updated periodically.

Future versions will be expanded to cover missing sections such as thermal fields. Minor
updates to correct grammatical errors or mistakes will also occur. Please email any
mistakes found, improvement suggestions or changes to the author.

Finally, the author would like to thank the many technical staff, reserves coordinators and
Reserves Advisory Committee (RAC) members throughout Chevron for their many
suggestions, insights and comments which have made this guide possible. Special thanks
are offered to Ed Hrkel with the RAC and RAM team for his help in peer reviewing and
proof reading.

Greg Reep
Reserves Consultant
Global Exploration and Corporate Reserves

Chevron Confidential Use Only Page 2 of 244

 Version 2008.01

Table of Contents

1 Introduction to Decline Curve Analysis ..................................................................... 5

2 History of Decline Curve Analysis ............................................................................. 7
3 First Principle Equations for Petroleum Engineering ............................................... 13
3.1 Darcy’s Law...................................................................................................... 13
3.2 Radial Diffusivity Equation .............................................................................. 20
3.3 Material Balance ............................................................................................... 25
4 Production Declines - Why They Occur................................................................... 30
4.1 Declines Caused By Pressure Depletion........................................................... 30
4.2 Declines Caused By Flow Restrictions............................................................. 31
4.3 Declines Caused By Increasing Producing Back Pressure ............................... 32
4.4 Declines Caused By Transient Flow................................................................. 33
4.5 Declines Caused By Water Displacement ........................................................ 34
5 Arps Decline Curve Analysis.................................................................................... 35
5.1 Exponential Declines ........................................................................................ 35
5.2 Hyperbolic Declines.......................................................................................... 54
5.3 Harmonic Declines............................................................................................ 78
5.4 Comparison of Exponential, Hyperbolic and Harmonic Declines.................... 81
5.5 Hyperbolic Declines for Gas Wells .................................................................. 94
5.6 Secondary Product Stream Reserves – Solution Gas or Condensate................ 99
6 Decline Curve Analysis in Water Drive and Waterflood Reservoirs ..................... 107
6.1 Water Ratio Methods for Waterfloods and Water Drive Reservoirs.............. 108
6.2 Water Oil Ratio Curves................................................................................... 116
6.3 Rate Time Forecasts from Water Oil Ratio Curves ........................................ 121
6.4 RAC Guidance for DCA in Waterfloods and Water Drive Reservoirs .......... 132
7 Decline Curve Data and Assumptions .................................................................... 135
7.1 Data Regression - Least Squares Best Fit ....................................................... 135
7.2 Boundary Conditions ...................................................................................... 140
7.3 Data Accuracy and Uncertainty ...................................................................... 143
7.4 Well Groupings versus Individual Well DCA ................................................ 145
7.5 Equipment Operating Conditions ................................................................... 146
7.6 DCA Mapping................................................................................................. 150
8 Infill and Extension Well Drilling Effects .............................................................. 153
8.1 Drilling Wedge Method for Rate Time or Rate Cumulative .......................... 153
8.2 Drilling Wedge Method for Waterflood Fields .............................................. 159
8.3 Type Well Method for Drilling or Workover Programs................................. 163
9 Field versus Individual Well Decline Curve Analysis............................................ 169
9.1 DCA by Field Groupings ................................................................................ 171
9.2 Decline and Rate Statistics for a Field............................................................ 172
9.3 Advantages of Calculating Reserves by Well................................................. 175
9.4 Aggregating Wells or Reservoirs to a Hyperbolic Decline ............................ 179
10 Advanced Decline Curve Analysis – Rate Transient Analysis........................... 183
10.1 Blasingame Plot Type Curve .......................................................................... 184

Chevron Confidential Use Only Page 3 of 244

 Version 2008.01

10.2 Log-log Plot Type Curve ................................................................................ 186

10.3 Advanced Decline Curve Analysis Software.................................................. 188
11 Probabilistic Decline Curve Analysis ................................................................. 198
11.1 Reasons for Poor Success of Probabilistic DCA ............................................ 198
11.2 Improvements to Probabilistic DCA............................................................... 201
12 Material Balance in Gas Reservoirs.................................................................... 206
12.1 Ideal Gas Law and Gas Equations .................................................................. 206
12.2 Gas Reservoir Data Acquisition...................................................................... 211
12.3 Determining Average Reservoir Pressure....................................................... 218
12.4 Water Influx in Gas Reservoirs....................................................................... 224
13 Technical References .......................................................................................... 241
14 Index ................................................................................................................... 243

Chevron Confidential Use Only Page 4 of 244

 Version 2008.01

1 Introduction to Decline Curve Analysis

Decline Curve Analysis (DCA) is the prediction of future hydrocarbon production by

analyzing past production. Early petroleum engineers realized that by studying the trends

of past production from wells and fields, one could forecast future production. DCA is

the most widely used and many times the most misused petroleum engineering activity.

Over 80 percent of Chevron’s fields have their reserves calculated or checked using

DCA. Recognizing this importance, this guide is being written to help improve the DCA

methods, techniques and standards within the Chevron Corporation.

Much of the drudgery of DCA has been eliminated by modern digital scientific

computers. Gone are the days where the petroleum engineer or engineering assistant

spent many hours updating decline curve books by hand. Unfortunately, along with these

computers came a trend in which some petroleum engineers no longer recognize the

inherent inaccuracies that come from the DCA methods. The only thing that is certain

about a DCA production forecast is that it is most certainly wrong. Arguing passionately

that the correct decline from a field can be calculated to three or more decimal point

accuracy is fruitless and naive. History has repeatedly shown that the final ultimate

recovery of a field or well will only really be known when the property is plugged and

abandoned.

Warning – There are many forms and variations of DCA. Many of the methods,

standards, “rules of thumb” and suggestions in this manual are generalized. There will

Chevron Confidential Use Only Page 5 of 244

 Version 2008.01

always be situations requiring exceptions to the rules. It is the responsibility of the

petroleum engineer or qualified reserves estimator (QRE) to recognize these exceptions

and use the appropriate engineering judgment for the situation.

Chevron Confidential Use Only Page 6 of 244

 Version 2008.01

2 History of Decline Curve Analysis

It was recognized early in the history of petroleum engineering that calculating reserves

and production forecasts were possible by studying past production trends. The early

work was empirical and used mainly for proration and taxation.

In the Manual for the Oil and Gas Industry – Arnold (1919) presented a method to

calculate estimated ultimate recovery (EUR) from the first year oil production and the

decline observed during the first year production. Figure 2-1 shows Arnold’s method34 in

graphical form.

Chevron Confidential Use Only Page 7 of 244

 Version 2008.01

Figure 2-1 - Arnold’s Plot of Reserves versus First Year Production

In 1924, W. W. Cutler published Estimation of Underground Oil Reserves by Oil-

Well Production Curves. Cutler described the methods in use for calculating reserves at

that time. These included the barrels per acre method, saturation method, constant-

percentage decline method and percentage-decline curve method. Cutler discussed the

fundamentals of production decline curves on both a well and a tract basis and the effects

of increasing or decreasing well count and well spacing. Cutler also proposed the use of

semi-log and log-log curves in addition to the commonly used Cartesian plots. Figure

2-2 shows several of Cutler’s plots34.

Chevron Confidential Use Only Page 8 of 244

 Version 2008.01

Figure 2-2 – Cutler’s Plots

In 1944 J. J. Arps published Analysis of Decline Curves19 at the Houston AIME

meeting. This paper presented the exponential, hyperbolic and harmonic DCA equations

which are still used today for most DCA reserves estimates and production forecasts. It

is historically interesting that Arps pointed out that DCA was just being revived as a

petroleum engineering tool. This occurred because World War II eliminated many of the

state set production allowables that had been in place for decades. These allowables had

prevented the effective use of DCA. The paper is also interesting historical reading as it

details the laborious nature of DCA before digital computers. Several shortcuts and

graphical techniques are presented to solve for the DCA constants.

Arp’s use of the exponential, hyperbolic and harmonic equations was generally empirical.

He made little attempt to link DCA curve types to reservoir and well conditions. He did

recognize that reservoirs, where cumulative production has an inverse linear relation with

reservoir production, will produce with an exponential decline assuming productivity

indexes remain constant. It was later recognized there are some empirical reasons why

Chevron Confidential Use Only Page 9 of 244

 Version 2008.01

these declines work. For example, a constant boundary drainage liquid reservoir with a

constant fluid compressibility will produce with an exponential decline. The proof of this

concept is shown later in this manual. Similarly, gas wells, because of their changing gas

compressibility (1/P), their Pe2-Pw2 form of Darcy’s law, changing Z-factor and changing

gas viscosity, will produce a production forecast that can be approximated by a

hyperbolic decline.

In the 1970’s and 1980’s, considerable progress in DCA was made through the tireless

and passionate work of M. J. Fetkovich. He presented an approach29 that plotted well

performance in dimensionless time and then determined reservoir and decline curve

parameters using a set of type curves. The type curves have two stems. The first

represents the period while the well is in transient flow. The second covers the time

period after the well has reached boundary conditions. Figure 2-3 shows the type

curves34 used by the Fetkovich method.

Chevron Confidential Use Only Page 10 of 244

 Version 2008.01

Figure 2-3 – Fetkovich Curve

The next large improvement in DCA came in the 1990’s through the insightful work of

Tom Blasingame, a petroleum engineering professor from Texas A&M University. He

introduced the use of material balance time for variable rates and pressures along with the

use of rate integral functions on a type curve basis. With this method, pressure depletion

wells can be evaluated more accurately and in many cases reservoir conditions such as kh

and skin factor determined. Figure 2-4 shows Blasingame’s method34.

Chevron Confidential Use Only Page 11 of 244

 Version 2008.01

Figure 2-4 - Blasingame Plot

By 2000, the use of all the methods have been incorporated into fully integrated

production analysis (PA) computer program packages. Two of the most popular in use at

Chevron is Kappa Engineering Topaze-EcrinTM and the Fekete [Link]. RTA. These

are powerful tools when used by competent and knowledgeable petroleum engineers,

especially in wells with changing boundary conditions.

Chevron Confidential Use Only Page 12 of 244

 Version 2008.01

3 First Principle Equations for Petroleum Engineering

The production rates from most wells and fields begin declining as soon as they are fully

developed and place on production. In developing an understanding for the causes and

analysis of production declines, one is required to go back to some of the First Principle

petroleum engineering equations – Darcy’s law, radial diffusivity and material balance.

3.1 Darcy’s Law

The most important equation governing producing wells is called Darcy’s law. Equation

3-1 shows Darcy’s law8 for oil wells in pseudosteady-state flow with a radial drainage

area.

7 . 08 × 10 − 3 kh (p e − p wf )
q0 =
⎛ r ⎞
μ o Β o ⎜⎜ ln e − 0 . 5 + s ⎟⎟
⎝ rw ⎠
Equation 3-1 – Darcy’s Law for Oil Wells

qo = oil production rate – STB/day

k = permeability of formation, md

h = thickness of formation, ft

pe = pressure at the effective drainage radius re normally

approximated by pR, psia

Chevron Confidential Use Only Page 13 of 244

 Version 2008.01

pwf = flowing BHP at the producing formation face, psia

μo = oil viscosity, cp

Βo = oil formation volume factor, RB/STB

re = effective drainage radius, ft

rw = wellbore radius, ft

s = skin effect, dimensionless

Equation 3-1 is the most commonly used form of Darcy’s Law in pseudosteady-state

flow. It does, however, require that that pe be at the drainage radius. A more accurate

version of this equation8 is shown by Equation 3-2.

7 . 08 × 10 − 3 kh (p r − p wf )
q0 =
⎛ r r 2
⎞
μ o Β o ⎜⎜ ln − 2
+ s ⎟
⎟
⎝ rw 2 re ⎠
Equation 3-2 - Darcy's Law for Oil Wells at Any Radius

The radius r and pressure pr is the radius from the well. Slider advises that care should be

taken that Equation 3-1 not be misapplied if the pressure being used is not at the

boundary.

Equation 3-1 can be adjusted for flow geometries different from radial. It also can be

converted into a commonly used form called the productivity index or PI as shown by

Equation 3-3. This assumes that all variables and constants, except for reservoir pressure

Chevron Confidential Use Only Page 14 of 244

 Version 2008.01

and working bottom-hole pressure, can be consolidated into a single constant called the

productivity index or PI. Equation 3-3 has the units of STBOPD per psia.

7 . 07 × 10 − 3 kh
PI =
⎛ re ⎞
μ o Β o ⎜⎜ ln − 0 . 5 + s ⎟⎟
⎝ r w ⎠
Equation 3-3 - Productivity Index in Darcy Form

This is a very useful equation for the petroleum engineer as one can usually easily obtain

reservoir pressure, working or flowing bottom-hole pressure and production rate. The PI

can then be calculated using Equation 3-4.

qo
PI =
( p e − p wf )
Equation 3-4 - Productivity Index

Caution should be used when applying the PI equation because the PI for some oil wells,

especially those in solution gas drive reservoirs, will decrease as the well producing

bottom-hole pressure is reduced. J. V. Vogel36 with Shell Oil presented a method for

handling this PI tail off.

Let’s demonstrate the Darcy’s Law and the PI equation with an example.

Chevron Confidential Use Only Page 15 of 244

 Version 2008.01

Example - Calculate the oil producing rate and the productivity index for a newly

completed pseudosteady-state oil well which has the following conditions:

k = 15 md

h = 20 ft

pe = 3000 psia

pwf = 250 psia

μo = 0.5 cp

Βo = 1.035 RB/STB

re = 1500 ft

rw = 0.5 ft

s = 0

ce = 5 x 10-6 psi-1

Ф = 0.2

7 . 07 × 10 −3
× 15 × 20 × (3000 − 250 ) = 1 , 502
q0 = BOPD
⎛ 1500 ⎞
0 . 5 × 1 . 035 × ⎜ ln − 0 .5 + 0 ⎟
⎝ 0 .5 ⎠

7.07 × 10 −3 × 15 × 20
PI = = 0.55 BOPD / PSIA
⎛ 1500 ⎞
0.5 × 1.035 × ⎜ ln − 0.5 ⎟
⎝ 0.5 ⎠

Chevron Confidential Use Only Page 16 of 244

 Version 2008.01

1,502
PI = = 0.55 BOPD / PSIA
3000 − 250

Darcy’s Law for fluid producing situations assumes laminar flow. In gas wells with

radial drainage where the gas is highly compressible and the wellbore flow is generally

turbulent, a constant n is added. This constant is usually determined by open flow testing

and will range between 0.5 for very turbulent flow to 1.0 for laminar flow. For Darcy’s

Law in pseudosteady-state flow, Slider8 presented Equation 3-5 for gas wells which takes

a slightly different form.

qg =
0 . 703 × 10 −3
( 2
kh p e − p wf
2
) n

⎛ re ⎞
μ g T f z avg ⎜⎜ ln − 0 . 5 + s ⎟⎟
⎝ rw ⎠

Equation 3-5 - Darcy's Law for Gas Wells

qg = gas production rate, MCFD

pe = pressure at the effective drainage radius re normally approximated by pR, psia

pwf = flowing BHP at the producing formation face, psia

μg = gas viscosity, cp

k = permeability of formation, md

h = thickness of formation, ft

Tf = formation temperature, degrees Rankin

Chevron Confidential Use Only Page 17 of 244

 Version 2008.01

zavg = gas compressibility factor at the average pressure, dimensionless

re = effective drainage radius, ft

rw = wellbore radius, ft

s = skin effect, dimensionless

n = gas well backpressure test slope, dimensionless

Slider8 further points out that the -0.5 constant in the denominator can be combined with

the drainage distances to give Equation 3-6. This is the gas deliverability equation that is

commonly used in conventional deliverability tests.

qg =
0 . 703 × 10 − 3 kh p e − p wf ( 2 2
) n

⎛ ⎛ re ⎞ ⎞
⎜ ⎜
μ g T f z avg ⎜ ln ⎜ 0 . 606 ⎟ + s ⎟⎟
⎟
rw ⎠
⎝ ⎝ ⎠

Equation 3-6 - Gas Deliverability Equation

Some purists will point out that incorporating the exponent n into Darcy’s Law

invalidates the units of psia and MCFPD. A more complete form of Darcy’s Law for gas

wells is presented in the SPE Petroleum Engineering Handbook11 which incorporates the

non-Darcy flow factor into the denominator as shown by Equation 3-7. Here pR is the

average pressure in the drainage area and FnD is the non-Darcy flow factor.

Chevron Confidential Use Only Page 18 of 244

 Version 2008.01

qg =
( 2
0.703 × 10 −3 kh pR − p wf
2
)
⎛ r ⎞
μ g Tf z avg ⎜⎜ ln e − 0.75 + s + FnD q g ⎟⎟
⎝ rw ⎠

Equation 3-7 - Darcy's Law for Gas Wells using non-Darcy Flow Factor

To solve this for of the equation, we set the constants C1 and C2 as follows:

C1 =
(
0.703 × 10 −3 kh pR − p wf
2 2
)
μ g Tf z avg

re
C 2 = ln − 0.75 + s
rw

By substitution Equation 3-7 then becomes Equation 3-8 which can then be solved by the

quadratic equation as shown by Equation 3-9.

C1
qg =
C 2 + FnD q g
Equation 3-8 – Darcy’s Law for Gas Flow Using Constants

2
− C 2 + C 2 + 4 FnD C1
qg =
2 FnD
Equation 3-9 - Quadratic Solution to Gas Flow

For a known flow rate and set of reservoir parameters and flowing conditions, the non-

Darcy flow factor can be solved using Equation 3-8. Multiple flow rates can then be

solved for other conditions by calculating revised constants C1 and C2 and Equation 3-9.

Flow rates could also be solved using Equation 3-7 and the Excel solver functionality.

Chevron Confidential Use Only Page 19 of 244

 Version 2008.01

3.2 Radial Diffusivity Equation

Newly completed wells are in transient flow until the radius of investigation of the

pressure drop in the formation caused by the start of well production reaches the drainage

radius boundary. This process acts much like a ripple wave from a pebble thrown into a

pond. The initial pressure wave expands radially until the reservoir boundary is reached.

There the well drainage area stabilizes and the well begins pseudosteady-state flow. The

elapsed time for this stabilization to occur is calculated using the radial diffusivity

equation. For radial flow it takes the form of Equation 3-10.

0.04 × 103φμe ce re 2
ts =
k
Equation 3-10 - Radial Diffusivity Equation in Time in Days

ts = stabilization time, days

ce = effective compressibility, (1/psi)

k = permeability, md

μe = effective viscosity, cp

Ф = porosity, decimal

re = drainage radius, ft

Chevron Confidential Use Only Page 20 of 244

 Version 2008.01

This equation can also be rearranged into Equation 3-11 to calculate the radius of

investigation for a given time period.

t×k
r=
0.04 × 103φμe ce
Equation 3-11 - Radial Diffusivity Equation in Radius

All wells produce in transient flow until the stabilization time calculated using Equation

3-10 is reached. In high rate oil wells, the transient flow period can be quite short, many

times only a few hours. Gas wells can take much longer to stabilize because gas has a

much higher effective compressibility. Tight gas wells with permeabilities less than 0.5

md and large well spacings may be in transient flow for several years. During the

transient period, the production will drop significantly because the drainage radius is

expanding and the gas has further to flow for the given pressure drop.

Note that Equation 3-10 is independent of rate and wellbore flowing pressure. The

stabilization time is only influenced by the permeability, porosity, viscosity and effective

compressibility (reciprocal of reservoir pressure). This is similar to the speed of sound

which is a constant for a given set of conditions. For the speed of sound, a louder sound

does not travel faster than a softer sound. Similarly, a wellbore with a larger drawdown

does not spend less time in transient flow.

Chevron Confidential Use Only Page 21 of 244

 Version 2008.01

The accepted technique for predicting the drop in wellbore pressure during transient flow

is called the constant rate solution to the radial diffusivity equation. For this solution the

rate is held constant and wellbore pressure is solved. The technique for predicting rate is

called the constant pressure solution. For this solution, the wellbore pressure is held

constant and the rate is solved. Where rates or pressures are changing, superposition is

used to make the predictions.

The constant rate solution and constant pressure solution can be tedious. If only

engineering accuracy is needed, a very close prediction of rates during transient flow can

be made by calculating the effective drainage radiuses for a series of times. These

radiuses are then entered into Darcy’s law for pseudosteady-state flow to calculate the

flow rates.

Example - Calculate the gas production rate and the elapsed time in days when the

drainage radius equals 100, 200, 300, 500, 1000, 1500, 2000 and 2500 feet for the tight

gas well with the following conditions. Also generate a semi-log gas rate versus time

plot for the transient time period.

k = 0.5 md

h = 70 ft

pe = 5000 psia

pwf = 1500 psia

φ = 0.10

Chevron Confidential Use Only Page 22 of 244

 Version 2008.01

μg = 0.027 cp,

Tf = 710, degrees Rankin,

zavg = 1.058

re = 2500 ft

rw = 0.5

s = 0

n = 0.99

ue = 0.03 cp

ce = 0.000163

The calculation of rate and time for a drainage radius of 100 feet is made using Equation

3-5 and Equation 3-11. These equations are then used to populate the table.

qg =
0 . 703 × 10 −3
(
× 0 . 5 × 70 × 5000 2 − 1500 2 ) 0 . 99

= 5 , 655 MCFPD
⎛ 100 1 ⎞
0 . 027 × 710 × 1 . 058 × ⎜ ln − + 0⎟
⎝ 0 .5 2 ⎠

0.04 × 103 × 0.10 × 0.03 × 0.000163 × 100 2

ts = = 0.4 days
0.5

Drainage Radius - feet Elapsed Time – days Gas Rate - MCFPD

100 0.4 5,655

200 1.6 4,941

Chevron Confidential Use Only Page 23 of 244

 Version 2008.01

300 3.5 4,602

400 6.3 4,388

500 10 4,235

1000 39 3,821

1500 88 3,615

2000 156 3,482

2500 – re reached 245 3,385

Tight Gas Well in Transient Flow

10,000
MCFPD

1,000
0 50 100 150 200 250 300
Days

Figure 3-1 - Tight Gas Well in Transient Flow

Chevron Confidential Use Only Page 24 of 244

 Version 2008.01

Note the hyperbolic shape of the curve. Many petroleum engineers mistakenly

performed DCA using Arp’s equations on the transit period of tight gas wells. In most

cases, the hyperbolic exponent factors calculate larger than 1.0. This will give

exaggerated future production volumes causing reserves to be overestimated. Hyperbolic

decline curve analysis should never be performed during the transient flow period and the

hyperbolic constants used for the remainder of the well’s life. Other engineers, especially

early in the well life, see the high initial decline and assume that it will continue. Thus

they seriously under estimate the reserves. A better method is to review other wells in

the area with more production history to determine at which point the well can be

approximated by a normal hyperbolic or exponential decline. If other wells are not

available, a transient decline curve analysis should be performed using a well simulator

or finite differences reservoir simulator.

3.3 Material Balance

Our equations so far have focused on the prediction of producing rate. For a complete

understanding of declines, a basic understanding of the material balance equations is

necessary. These equations predict reservoir size and how reservoir pressure and fluid

saturations will vary with production. We will only cover the basics but any good

reservoir engineering text will have a thorough chapter on material balance methods.

The first equations are for material balance for a single reservoir fluid. The basic

equation from which all of the equations are derived states that the initial reservoir fluid

Chevron Confidential Use Only Page 25 of 244

 Version 2008.01

in place times the initial formation volume factor of the fluid is equal to the initial fluid in

place minus the cumulative production multiplied times the current formation volume

factor. Formation volume factor is a ratio of the surface volume of the fluid to the

reservoir volume of the fluid. For under saturated oil reservoirs or those at pressures

above the bubble point, the equation is shown as Equation 3-12.

N × Boi = (N − N p ) × Bo

Equation 3-12 - Material Balance Oil

N = original oil-in-place, STBO,

Np = cumulative oil production, STBO

Boi = initial oil formation volume factor, RB/STB

Bo = current oil formation volume factor, RB/STB

Equation 3-12 can be rearranged to calculate the OOIP as shown in Equation 3-13.

Bo N p
N=
Bo − Boi

Equation 3-13 - Material Balance Oil for OOIP

It can also be changed to solve for the cumulative production in Equation 3-14.

Chevron Confidential Use Only Page 26 of 244

 Version 2008.01

N × (B o − B oi )
Np =
Bo

Equation 3-14 - Material Balance Oil for Cumulative Production

Since cumulative production divided by the OOIP calculates the recovery efficiency,

Equation 3-14 can solve for recovery efficiency.

E ro =
Np
=
(B o − B oi )
N Bo

Equation 3-15 - Material Balance Oil Recovery

These equations take the exact same form for gas and water.

OGIP × B gi = (OGIP − G p ) × B g

Equation 3-16 - Material Balance for Gas

OWIP × Bwi = (OWIP − W p ) × Bw

Equation 3-17 - Material Balance for Water

Once free gas begins to breakout in the reservoir, the equations become more complex as

they must address the free gas in the reservoir and the gas that has been produced.

Equation 3-18 shows the material balance calculations for free gas.

Chevron Confidential Use Only Page 27 of 244

 Version 2008.01

Np (Bo + B g (Rp − R s ) )
N=
Bo − Boi + B g (R si − R s )

Equation 3-18 - Material Balance with Free Gas

Rs = current saturation GOR of the oil in the reservoir, SCF/STBO,

Rp = cumulative GOR = Cum gas/Cum oil, SCF/STBO,

Rsi = GOR at initial reservoir conditions, SCF/STBO

Rearranging Equation 3-18 for recovery efficiency of oil will yield Equation 3-19.

Np Bo − Boi + Bg (R si − R s )
Ero = =
N Bo + Bg (Rp − R s )

Equation 3-19 - Material Balance Oil with Free Gas Recovery

All of the above equations require an understanding of fluid conditions. The petroleum

engineer can obtain these through actual well fluid sampling or by using one of the

standard correlations that are widely available. Here are some of the typical ranges in

which reservoir fluids will exist.

Chevron Confidential Use Only Page 28 of 244

 Version 2008.01

Fluid Property Variable Low High Units

Formation Volume Factor Oil Bo 1.02 2.4 rb/stb
Formation Volume Factor Gas Bggg 0.003 0.1 rcf/scf
Viscosity Oil μo 0.2 30 cp
Viscosity Gas μg 0.01 0.05 cp
Viscosity Water μw 0.4 1.05 cp
Compressibility Oil (P>Pb) Co 5 20 x 10-6 psi-1
Compressibility Oil (P<Pb) Co 30 200 x 10-6 psi-1
Compressibility Gas Cg 50 1000 x 10-6 psi-1
Compressibility Water Cw 3 5 x 10-6 psi-1
Compressibility Formation (Normal) Cf 2 10 x 10-6 psi-1
Compressibility Formation (Abnormal) Cf 10 100 x 10-6 psi-1

Table 3-1 – Typical Fluid Properties

Chevron Confidential Use Only Page 29 of 244

 Version 2008.01

4 Production Declines - Why They Occur

Production declines occur for a number of reasons. Some are related to reservoir

conditions. Some are volumetric conditions because of variables such as porosity, water

saturation, drainage radius, etc. Others are performance related such as permeability and

viscosity. Other conditions are mechanical such as well drawdown. The following

sections investigate the causes.

4.1 Declines Caused By Pressure Depletion

From the First Principles equations and examples listed in the previous chapter, one can

see several reasons why declines occur. From Darcy’s law equations, especially

Equation 3-4 - Productivity Index, it is shown that with all other variables including

WBHP being held constant, the producing rate will decline as production reduces the

pressure of the reservoir through depletion. This is the major cause of most production

declines.

Example - A newly completed oil well has been producing for a period of one year.

When initially completed, the well had an average reservoir pressure of 3000 psia and

was producing with a bottom-hole pressure of 250 psia. One year later the reservoir had

declined to 2250 psia. What was the effective decline in production during the first year

of production using the following reservoir and well conditions?

Chevron Confidential Use Only Page 30 of 244

 Version 2008.01

k = 15 md
h = 20 ft
μo = 0.5 cp
Βoi = 1.035 RB/STB
Βo = 1.02 RB/STB
re = 1500 ft
rw = 0.5 ft
s = 0

Using Equation 3-1 – Darcy’s Law for Oil Wells, one can calculate the initial rate and the

one year rate. From these two rates, the decline is calculated.

7 . 07 × 10 −3
× 15 × 20 × (3000 − 250 ) = 1 , 502
q oi = BOPD
⎛ 1500 ⎞
0 . 5 × 1 . 035 × ⎜ ln − 0 .5 + 0 ⎟
⎝ 0 .5 ⎠

7 . 07 × 10 −3
× 15 × 20 × (2250 − 250 ) = 1 ,108
q o _ 1 yr = BOPD
⎛ 1500 ⎞
0 . 5 × 1 . 02 × ⎜ ln − 0 .5 + 0 ⎟
⎝ 0 .5 ⎠

1502 − 1108
Decline = = 0.26
1502

4.2 Declines Caused By Flow Restrictions

Chevron Confidential Use Only Page 31 of 244

 Version 2008.01

Production declines can also occur where the flow capability of a well or field decreases

with time. Scaling is a common situation. In the previous example, further suppose that

during the first year of production the completion developed a scaling problem which

damaged the well causing a skin factor of 2.0. What would the decline have increased

to?

7 . 07 × 10 −3
× 15 × 20 × (2250 − 250 )=
q o _ 1 yr = 875 BOPD
⎛ 1500 ⎞
0 . 5 × 1 . 02 × ⎜ ln − 0 .5 + 2 ⎟
⎝ 0 .5 ⎠

1502 − 875
Decline = = 0.42
1502

4.3 Declines Caused By Increasing Producing Back Pressure

Now further suppose that the pumping unit that was installed on this well had developed

a leaking ball and seat in the pump. This raised the WBHP to 750 psia. What would the

decline now have increased to?

7 . 07 × 10 −3
× 15 × 20 × (2250 − 750 )=
q o _ 1 _ yr = 656 BOPD
⎛ 1500 ⎞
0 . 5 × 1 . 02 × ⎜ ln − 0 .5 + 2 ⎟
⎝ 0 .5 ⎠

1502 − 656
Decline = = 0.56
1502

Chevron Confidential Use Only Page 32 of 244

 Version 2008.01

As demonstrated, production declines can occur due to both reservoir depletion and

mechanical effects. It is important that the engineer have a good understanding of all

aspects of a producing well, both reservoir and mechanical.

In practice not all of the other Darcy Law variables remain constant. As depletion

occurs, some of the fluid properties will change slightly. This can cause a change in the

decline, especially in oil wells that reach bubble point. Declines can also occur when

well damage occurs due to scale or fines plugging causing skin factor increases. Many

declines can be changed by wellbore cleanouts, re-perforations, workovers to open

additional net pay, acid stimulations and fracture treatments. These clean outs will effect

many of the variables used to calculate the productivity index.

Declines can also occur if the producing bottom-hole pressures do not remain constant.

A rising WBHP can occur with wellbore loading, sales line pressure increases, choke

changes, rod pump leakage or tubing leaks. It is important for the QRE to investigate

the operating conditions for a well or reservoir to ensure that operating conditions

are stable before calculating a decline that is used in a reserves calculation.

4.4 Declines Caused By Transient Flow

In Section 3.2 - Radial Diffusivity Equation, it was shown that wells experience

decreased production as their transient flow period progresses. This drop in production

occurs because the effective drainage radius is increasing and while the pressure draw

Chevron Confidential Use Only Page 33 of 244

 Version 2008.01

down remains constant. Since the production has a longer distance to flow through the

formation, the rate decreases.

From Figure 3-1 - Tight Gas Well in Transient Flow, one can see that new wells can

experience a changing decline during their transient flow period. Once the QRE

recognizes that the transient phase of the well is occurring, there are several excellent

decline curve programs available which use type-curve techniques developed by

Fetkovich, Blasingame and others to extract reservoir parameters including permeability,

OOIP, OGIP and skin factors. These have been shown to be very useful in calculating

reserves. The only requirement is that the petroleum engineer has a rough idea of

reservoir conditions and daily rate and surface or bottom-hole pressure information for

the well.

4.5 Declines Caused By Water Displacement

Another type of decline occurs on reservoirs that are under waterflood, gas injection or

experiencing a strong water drive. If voidage is maintained so the production in reservoir

barrels is balanced by injection, the average pressure in the reservoir remains somewhat

constant. The decline occurs because the displacement of oil by water causes the water-

oil-ratio (WOR) to increase. The increase occurs because of fractional flow, sweep

efficiency and the displacement nature of these injection processes. Stated simply, the oil

rate is being replaced with water rate. Declines caused by oil displacement will be

discussed in detail later in this manual.

Chevron Confidential Use Only Page 34 of 244

 Version 2008.01

5 Arps Decline Curve Analysis

Several excellent papers19 were published in 1944 and 1945 by J. J. Arps that

standardized the equations the industry were using for DCA. Although it has been over

60 years since these papers were published, the equations have essentially remained

unchanged and are still referred to the “Arps” equations. These papers are recommended

reading for all petroleum engineers. They have definitely withstood the “test of time.”

Arps categorized decline curves into exponential, hyperbolic and harmonic declines. The

hyperbolic equation is the universal equation. Exponential and harmonic declines are

two special hyperbolic cases when the hyperbolic constant b equals 0 for exponential

declines and 1 for harmonic declines. We will discuss each type decline separately.

All decline curve types are derived from the differential equation for the normal decline

rate given by Equation 5-1.

dq dt
d =−
q

Equation 5-1 - Basic Decline Equation

5.1 Exponential Declines

The most common decline used by petroleum engineers is the exponential decline. The

exponential decline popularity is primarily due because they are mathematically easier to

Chevron Confidential Use Only Page 35 of 244

 Version 2008.01

use than hyperbolic declines. They also produce production forecasts that are more

conservative than hyperbolic declines or harmonic declines. In reality, hyperbolic

declines are much more common than exponential declines. Harmonic declines are quite

rare.

Exponential declines will plot a straight line when the rate versus time curve is plotted on

a semi-log scale. Figure 5-1 shows a 5 year production history for an oil well with a

starting rate of 100 BOPD and a 20% effective decline.

Exponential Decline
100 BOPD IP - 20% Decline

100
BOPD

10
0 1 2 3 4 5
Years

Figure 5-1 - Exponential Decline on Semi-log Scale

As shown, the 20% decline results in a straight line. Production at year 1 equals 80

BOPD, year 2 equals 64 BOPD and year 3 equals 51 BOPD. The mathematician will

immediately recognize a simple mathematical series represented by the Equation 5-2.

Chevron Confidential Use Only Page 36 of 244

 Version 2008.01

Rate t = qi × (1 − D) t = 100 × (1 − .2) t

Equation 5-2 - Exponential Rate Forecast

Early petroleum engineers recognized this equation and used a convenient “mathematics

free” technique to determine exponential decline values by using two triangles. One

triangle was used to “eyeball fit” the exponential decline while the second triangle was

placed against the first triangle and used as a guide to shift the first triangle so a parallel

decline passed through the beginning of a year. The decline was then simply determined

by reading the intersection value at the end of the year and counting the cycles. In Figure

5-1 the intersection occurs at the second 10% section or a 20% exponential decline.

Constant percentage exponential declines are also commonly called effective declines

and are referred to as the symbol D. Henceforth in this text constant percentage

declines will be referred as effective declines. Another decline form is called the

nominal decline which uses the symbol “a.” The relationship between the nominal and

effective forms is:

a = − ln(1 − D)

Equation 5-3 - Nominal Decline

Chevron Confidential Use Only Page 37 of 244

 Version 2008.01

D = 1 − e −a

Equation 5-4 - Effective Decline

Effective Versus Nominal Declines

0.7

0.6

0.5
Nominal Decline

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5
Effective Decline

Figure 5-2 - Effective versus Nominal Declines

Figure 5-2 shows the relationship between the effective decline and the nominal decline.

Most petroleum engineers use the exponential decline using its exponential form. The

exponential equation using the base of natural logarithms form (2.718) is

q = qi e − at

Equation 5-5 - Exponential Decline

Chevron Confidential Use Only Page 38 of 244

 Version 2008.01

In the previous graph example, the producing rate at 3.5 years would then be calculated

by:

a = − ln(1 − 0.2) = 0.2231

q = 100 e −0.2231× 3.5 = 45.8 BOPD

The nominal decline has one very useful advantage over the effective decline that has to

do with changing the time units on the decline rate. Many engineers prefer to work with

monthly declines in fields that have high production rates and short lives. To convert an

annual nominal decline to a monthly decline, one simply divides the annual nominal

decline rate by 12. To convert a yearly effective decline to a monthly decline, the twelve

root of 1-D must be calculated. Although this is not hard using a scientific calculator, it

is not very intuitive.

Engineers can also calculate an effective decline by reading two points off a semi-log rate

time decline and calculating the slope using Equation 5-6:

⎛ q2 ⎞
⎜ ln( ) ⎟
⎜ q1 ⎟
⎜ t −t ⎟
⎜ 2 1 ⎟
D =1− e ⎝ ⎠

Equation 5-6 - Decline From Two Points

The exponential decline D is for the time units used. If time is years, the decline is per

year in decimal form. If time is months, the decline is monthly.

Chevron Confidential Use Only Page 39 of 244

 Version 2008.01

The equation for cumulative production between two producing rates with an exponential

decline is show by Equation 5-7.

365 × (q1 − q 2 ) 365 × (q1 − q 2 )

Q1 2 = =
a − ln(1 − D)

Equation 5-7 - Reserves from Exponential Decline

The rate is units per day. If the rates are units per month, the constant used would be 12

instead of 365. If an annual rate is used, the constant would be 1.0.

In our example, what reserves would be recovered in the first 5 years of production?

q 2 = 100 × (1 − 0.2) 5 = 32.8 BOPD

365 × (100 − 32.8)

Q1 2 = = 109,920 STBO
− ln(1 − 0.2)

If the economic limit of our example is 10 BOPD, what would the ultimate recovery be?

365 × (100 − 10)

Q1 2 = =147,215 STBO
− ln(1 − 0.2)

The producing life required to reach the economic limit of 10 BOPD is calculated using

the following formula.

Chevron Confidential Use Only Page 40 of 244

 Version 2008.01

⎛q ⎞ ⎛q ⎞
ln⎜⎜ 1 ⎟⎟ ln⎜⎜ 1 ⎟⎟
q ⎝ q el ⎠
t el = ⎝ el ⎠=
a − ln(1 − D)

Equation 5-8 - Exponential Decline Time to Economic Limit

For our example, the time to reach the economic limit of 10 BOPD is calculated as

follows.

⎛ 100 ⎞
ln⎜ ⎟
⎝ 10 ⎠
t el = = 10.32 Years
− ln(1 − .2)

The following table shows the producing rates and cumulative production for the life of

our example well.

Years BOPD Np - STBO

0 100 -
1 80 32,714
2 64 58,886
3 51 79,823
4 41 96,573
5 33 109,973
6 26 120,692
7 21 129,268
8 17 136,129
9 13 141,618
10 11 146,008
10.32 10 147,215

Chevron Confidential Use Only Page 41 of 244

 Version 2008.01

The value for the nominal decline “a” can easily be calculated because it is simply the

slope of the log rate versus time curve. From the above table, the slope can be calculated

taking the rate at years 0 and 2 or any other two points from Equation 5-9.

⎛ ln(q1 ) − ln(q 2 ) ⎞ ⎛ ln(100) − ln(64) ⎞

slope = a = −⎜⎜ ⎟⎟ = −⎜ ⎟ = 0.22314
⎝ t1 − t 2 ⎠ ⎝ 0−2 ⎠

Equation 5-9 - Nominal Decline from 2 Rates

The effective decline can be calculated using Equation 5-4.

D = 1 − e −( a ) = 1 − e −( 0.22314 ) = 0.20

In situations where the initial daily rate and exponential decline are known and the

engineer needs to calculate the production that will be recovered over a given number of

years, Equation 5-2, Equation 5-7 and Equation 5-8 can be combined by integrating the

daily rate over time generating Equation 5-10.

Years
Reserves = ∫ 365 × qi × (1 − D) t dt
0

(1 − D) years − 1
Reserves years = 365 × qi ×
ln (1 − D)

Equation 5-10 - Reserves Over a Time Period

Chevron Confidential Use Only Page 42 of 244

 Version 2008.01

Example – A well is producing 200 BOPD with a 15% exponential decline. How much

oil will be recovered by this well between now and when the production concession ends

in 5.6 years?

(1 − 0.15) 5.6 − 1
Reserves 5.6 years = 365 × 200 × = 268,392 STBO
ln (1 − 0.15)

Another very useful and in this author’s opinion under-utilized feature of exponential

declines is that they will plot a straight line when the producing rate is plotted using

linear coordinates with cumulative production on the abscissa and rate on the ordinate.

That the graph is linear can be shown by Equation 5-7.

365 × (q1 − q 2 )
Q 1 2 = Np =
a

The equation can be converted into the linear y = mx + b form by rearranging the

variables to Equation 5-11.

⎛ a ⎞
qn = qi − ⎜ ⎟ × Np
⎝ 365 ⎠

Equation 5-11 - Rate Cumulative

A plot of the data in the table for our previous example is Figure 5-3.

Chevron Confidential Use Only Page 43 of 244

 Version 2008.01

Rate Versus Cumulative Oil Example

120
100
80

BOPD
60
40
20
0
0 50,000 100,000 150,000 200,000
Cumulative - STBO

Figure 5-3 - Linear Rate versus Cumulative

The rate versus cumulative plot has several very useful features. One is that by extending

the rate production to 0, the moveable hydrocarbon volume can easily be determined.

Another is that it very graphically shows the incremental recoveries realized by

development plan changes such as infill drilling or waterflooding. Figure 5-4 shows an

example of incremental recoveries in a field that has been infill drilled.

Rate Versus Np Infill Plot Example

20000

15000
BOPD
BOPD

10000 20 Acre Infill

10 Acre Infill
5000

0
40 60 80 100 120
Np - MMBO

Figure 5-4 - Rate versus Cumulative Infill Drilling Example

Chevron Confidential Use Only Page 44 of 244

 Version 2008.01

Another major advantage of using rate versus cumulative plot is that any periods of

significant downtime or shut-in are eliminated. Most fields will occasionally experience

downtime periods. Sometimes downtime is caused by external events such as labor

strikes, storms or major equipment failure. Rate versus time DCA requires a time shift to

eliminate the shut-in period. Warning – Ignoring downtime for major shut-in periods

can cause the calculations of declines which are artificially to low. Figure 5-5 shows

a rate time decline incurred on one of the wells producing into the Mars Platform

operated by Shell Oil when the analysis is not adjusted for Hurricane Katrina downtime.

The effective decline is 12%.

Figure 5-5 - Hurricane Downtime Included in Decline

Chevron Confidential Use Only Page 45 of 244

 Version 2008.01

Figure 5-6 show a rate versus cumulative analysis for the same Mars well which

effectively eliminates the shut in time from the best fit analysis. This shows an effective

decline of 23%. Figure 5-7 shows the 23% decline transposed onto the rate time plot for

the well. As shown, ignoring the hurricane shut in time period would have significantly

increased the remaining reserves.

Figure 5-6 - Hurricane Downtime Eliminated from Decline Using Rate Cumulative Plot

Chevron Confidential Use Only Page 46 of 244

 Version 2008.01

Figure 5-7 - Hurricane Downtime Eliminated Rate Time Plot

Sometimes the downtime is caused because of a poorly designed production system

which causes production to fluctuate considerably. This results in a producing history

that is not always at capacity. In these fields, a straight line can be plotted on the graph

that acts as an upper boundary or the capacity of the field. This upper boundary line

helps the engineer eliminate data points that should not be used in the DCA because the

field is not producing at capacity. Figure 5-8 is an example of what an upper boundary

analysis may look like.

Care should be used when making production forecasts in these types of fields because

the production down time needs to be incorporated into the decline used to make the rate

Chevron Confidential Use Only Page 47 of 244

 Version 2008.01

forecast. DCA including rate time points that are curtailed will give an unrealistically

low decline.

Downtime Example

100
80
BOPD

60
40
20
0
0 20,000 40,000 60,000 80,000
Cumulative - STBO

Figure 5-8 - Linear Rate versus Cumulative with Downtime

Once the graph with a linear line is produced, two points can be read off of the straight

line and the base exponential decline that the field would produce at without any

downtime can be calculated. In Figure 5-8, two points are (20,000, 80) and (60,000, 38).

The effective exponential decline would be calculated using Equation 5-9 and Equation

5-4.

365 × (80 − 38)

a= = 0.38325
(60,000 − 20,000)

D = 1 − e −0.38325 = 0.32 = 32%

Chevron Confidential Use Only Page 48 of 244

 Version 2008.01

Again, care should be used when making production forecasts in this field because the

production downtime needs to be incorporated into the decline forecast.

Two very useful tables can be generated for exponential decline equations that can give

the QRE a quick rule-of-thumb estimate for exponential declines. Table 5-1 shows the

exponential declines for various current rate life (R/P) values and ratios of qel/qcurrent.

With this table, once the current rate life is known, a rough understanding of the

exponential decline needed to achieve the reserves can quickly be estimated. The author

uses this table extensively in RAC meetings to spot check declines for reasonableness.

Chevron Confidential Use Only Page 49 of 244

 Version 2008.01

Rule-Of-Thumb Exponential Decline D (Percentage)

Ratio qel/qi

R/P Years 0.05 0.10 0.15 0.2

1.0 61.3% 59.3% 57.3% 55.1%
1.5 46.9% 45.1% 43.3% 41.3%
2.0 37.8% 36.2% 34.6% 33.0%
2.5 31.6% 30.2% 28.8% 27.4%
3.0 27.1% 25.9% 24.7% 23.4%
3.5 23.8% 22.7% 21.6% 20.4%
4.0 21.1% 20.1% 19.1% 18.1%
4.5 19.0% 18.1% 17.2% 16.3%
5.0 17.3% 16.5% 15.6% 14.8%
5.5 15.9% 15.1% 14.3% 13.5%
6.0 14.6% 13.9% 13.2% 12.5%
6.5 13.6% 12.9% 12.3% 11.6%
7.0 12.7% 12.1% 11.4% 10.8%
7.5 11.9% 11.3% 10.7% 10.1%
8.0 11.2% 10.6% 10.1% 9.5%
8.5 10.6% 10.0% 9.5% 9.0%
9.0 10.0% 9.5% 9.0% 8.5%
9.5 9.5% 9.0% 8.6% 8.1%
10.0 9.1% 8.6% 8.1% 7.7%
11.0 8.3% 7.9% 7.4% 7.0%
12.0 7.6% 7.2% 6.8% 6.4%
13.0 7.0% 6.7% 6.3% 6.0%
14.0 6.6% 6.2% 5.9% 5.6%
16.0 5.8% 5.5% 5.2% 4.9%
18.0 5.1% 4.9% 4.6% 4.3%
20.0 4.6% 4.4% 4.2% 3.9%

Table 5-1 – Current Rate Life - Exponential Declines

Table 5-2 shows what the remaining life in years is for various R/P values and ratios of

economic limit rate to initial rate.

Chevron Confidential Use Only Page 50 of 244

 Version 2008.01

Rule-Of-Thumb Remaining Life - Years

Ratio qel/qi

R/P Years 0.05 0.10 0.15 0.2

1.0 3.2 2.6 2.2 2.0
1.5 4.7 3.8 3.3 3.0
2.0 6.3 5.1 4.5 4.0
2.5 7.9 6.4 5.6 5.0
3.0 9.5 7.7 6.7 6.0
3.5 11.0 9.0 7.8 7.0
4.0 12.6 10.2 8.9 8.0
4.5 14.2 11.5 10.0 9.1
5.0 15.8 12.8 11.2 10.1
5.5 17.3 14.1 12.3 11.1
6.0 18.9 15.4 13.4 12.1
6.5 20.5 16.6 14.5 13.1
7.0 22.1 17.9 15.6 14.1
7.5 23.7 19.2 16.7 15.1
8.0 25.2 20.5 17.9 16.1
8.5 26.8 21.7 19.0 17.1
9.0 28.4 23.0 20.1 18.1
9.5 30.0 24.3 21.2 19.1
10.0 31.5 25.6 22.3 20.1
11.0 34.7 28.1 24.6 22.1
12.0 37.8 30.7 26.8 24.1
13.0 41.0 33.3 29.0 26.2
14.0 44.1 35.8 31.2 28.2
16.0 50.5 40.9 35.7 32.2
18.0 56.8 46.1 40.2 36.2
20.0 63.1 51.2 44.6 40.2
Table 5-2 – R/P Rule-Of-Thumb for Remaining Life

Example - A field is producing with an exponential decline, an initial rate of 1000 BOPD

and a calculated economic limit of 100 BOPD. The current booked reserves are 2.92

MMBO. The calculated R/P is 8.0. Estimate the exponential decline that will match the

booked reserves using Table 5-1.

Chevron Confidential Use Only Page 51 of 244

 Version 2008.01

EL 100
Ratio = = = 0.1
IP 1000

From Table 5-1, the exponential decline required to match the reserves is approximately

10.6%.

Table 5-3 is also very useful for determining rule-of-thumb exponential declines. It uses

a ratio that shows how cumulative production relates to a given drop in daily producing

rate. The first step is to calculate the slope of the ratio using the following equation:

Np 2 − Np1
Slope =
q 2 − q1

This slope can be easily calculated from two points on a decline curve and the

exponential decline estimated. Note that the same volume units are required for both the

cumulative and the daily rate.

Chevron Confidential Use Only Page 52 of 244

 Version 2008.01

Rule-Of-Thumb Exponential Declines

Slope = (Np2-Np1)/(q2-q1)
Decline Slope Decline Slope
1.0 -36317 14 -2420
1.5 -24150 15 -2246
2.0 -18067 16 -2093
2.5 -14417 17 -1959
3.0 -11983 18 -1839
3.5 -10245 19 -1732
4.0 -8941 20 -1636
4.5 -7927 21 -1548
5.0 -7116 22 -1469
5.5 -6452 23 -1397
6.0 -5899 24 -1330
6.5 -5431 25 -1269
7.0 -5030 26 -1212
7.5 -4682 27 -1160
8.0 -4377 28 -1111
8.5 -4109 29 -1066
9.0 -3870 30 -1023
9.5 -3657 35 -847
10.0 -3464 40 -715
10.5 -3290 45 -611
11.0 -3132 50 -527
11.5 -2988 55 -457
12.0 -2855 60 -398
12.5 -2733 65 -348
13.0 -2621 70 -303
13.5 -2517 75 -263

Table 5-3 - Rate versus Cumulative Slopes and Declines

Example - A field has produced a cumulative of 1,000,000 STBO while production

dropped from 1500 BOPD to 650 BOPD. Estimate the exponential decline.

1,000,000
Slope = = −1,176
650 − 1500

Chevron Confidential Use Only Page 53 of 244

 Version 2008.01

From Table 5-3, the exponential decline is approximately 27%.

5.2 Hyperbolic Declines

It was recognized very early that the hyperbolic decline is the most common decline. J. J.

Arps in his 1944 paper19 documents that W. W. Cutler Jr. by 1924 had recognized the

widespread existence of hyperbolic declines. Table 5-4 shows an early distribution

analysis of hyperbolic exponents generated by Cutler. As shown by Arps, approximately

2/3rds of the cases have a hyperbolic constant between 0.1 and 0.4 suggesting a log

normal distribution.

Exponent b Between Number of Cases

0.0 and 0.1 19
0.1 and 0.2 41
0.2 and 0.3 27
0.3 and 0.4 34
0.4 and 0.5 15
0.5 and 0.6 9
0. 6 and 0.7 4
Above 0.7. None

Table 5-4 - Hyperbolic Exponent Frequency – Arps (1944)

The hyperbolic decline is similar to the exponential decline except that the decline

lessens with time. The relationship of the decline to the producing rates in hyperbolic

Chevron Confidential Use Only Page 54 of 244

 Version 2008.01

declines is given by Equation 5-12 where ai is the nominal decline at the time of the

initial rate qi and a is the nominal decline that changes with the rate q.

a qb
=
ai qi b

Equation 5-12 - Hyperbolic Equation Form

The variable b is the hyperbolic constant which determines the degree of change in the

decline. The hyperbolic constant b varies between 0 and 1. When b is equal to 0, the

decline remains constant and equation reverts to the exponential form. When b is equal

to 1, the decline becomes harmonic. The harmonic decline will be discussed later in this

chapter.

The hyperbolic equation for how rate changes with time is shown by Equation 5-13

where t is the time since the initial rate qi.

qi
q= 1

(1 + b × ai × t ) b

Equation 5-13 - Hyperbolic Equation Rate

Chevron Confidential Use Only Page 55 of 244

 Version 2008.01

The cumulative production between two different producing rates q1 and q2 is given by

Equation 5-14.

b (1−b ) (1−b )
q × (q1 − q2 )
Q12 = i × 365
a i × (1 − b)

Equation 5-14 - Hyperbolic Equation Reserves

Note that the rates q1 and qi can be the same rate but they are not required to be. If the

rates are in daily units, an adjustment constant of 365 is used. If the rates are monthly, a

constant of 12 is used. If the rate is annual, a constant of 1.0 is used.

The time to the economic limit is calculated by Equation 5-15.

b
⎛ q1 ⎞
⎜⎜ ⎟⎟ − 1
q
t el = ⎝ el ⎠
ai × b

Equation 5-15 - Hyperbolic Remaining Life

Example – An oil well is producing 100 BOPD with an initial nominal decline of 0.2231

and a hyperbolic constant of 0.5. What will the producing rate be at 5 years?

Chevron Confidential Use Only Page 56 of 244

 Version 2008.01

100
q= 1
= 41.2 BOPD
(1 + (0.5 × 0.2231 × 5)) 0.5

How much production will be produced in the first 5 years?

100 0.5 × (100 (1−0.5 ) − 41.2(1−0.5 ) )

Q12 = × 365 = 117,148 STBO
0.2231 × (1 − 0.5)

How many years of production must occur before the producing rate reaches the

economic limits of 10 BOPD?

0.5
⎛ 100 ⎞
⎜ ⎟ −1
10
t el = ⎝ ⎠ = 19.4 years
0.2231 × 0.5

Construct a table showing the production forecast and graph the semi-log rate versus time

and linear rate versus cumulative plots.

Chevron Confidential Use Only Page 57 of 244

 Version 2008.01

Year Rate q BOPD Np - STBO

0 100.0 -
1 80.9 32,836
2 66.8 59,682
3 56.1 82,040
4 47.8 100,948
5 41.2 117,148
6 35.9 131,182
7 31.5 143,459
8 27.9 154,287
9 24.9 163,910
10 22.3 172,518
11 20.2 180,264
12 18.3 187,271
13 16.7 193,639
14 15.2 199,453
15 14.0 204,782
16 12.9 209,684
17 11.9 214,208
18 11.0 218,396
19 10.3 222,285
19.4 10.0 223,691

Hyperbolic Example

100
BOPD

10
0 5 10 15 20
Years

Figure 5-9 - Hyperbolic Decline on Semi-log Coordinates

Chevron Confidential Use Only Page 58 of 244

 Version 2008.01

Hyperbolic Example

120
100
80

BOPD
60
40
20
0
0 50 100 150 200 250
Cumulative Oil - MBO

Figure 5-10 - Hyperbolic Linear Rate versus Cumulative

Neither graph is linear. The slopes of both lines become less negative with time and

cumulative production. This is the nature of hyperbolic declines.

The shape of the hyperbolic curve is controlled by both the initial decline and the

hyperbolic constant b. Figure 5-11 shows the shape of the production forecast assuming

the 0.2231 initial nominal decline but varying the hyperbolic constant from 0.5 to 0.9.

Chevron Confidential Use Only Page 59 of 244

 Version 2008.01

Example Varying Hyperbolic Constant

100

b=0.5

BOPD
b=0.7
b=0.9

10
0 5 10 15 20
Years

Figure 5-11 - Effects of Hyperbolic Exponent

Note that the higher the b factor, the lower the decline, the longer the life to the economic

limit and the higher the ultimate recovery. In some extreme cases with hyperbolic

constants greater than 0.5, the hyperbolic equations will produce declines that are a very

low decline late in the producing life.

It has been the author’s experience that considerable confusion exists when using

hyperbolic declines over the relationship between initial effective and initial nominal

declines. If the initial effective decline is calculated using a method which determines

the slope of the decline of a line that is tangential to the semi-log plot at the instantaneous

point of the initial decline, then the equation that calculates the nominal initial decline is

a i = − ln(1 − D i ) . This is the same equation that was presented for exponential declines

and is called the instantaneous effective decline by most software programs.

Chevron Confidential Use Only Page 60 of 244

 Version 2008.01

Some engineers will determine the effective initial decline by reading the initial and one

year producing rates from the decline curve and calculate the effective decline rate using

the equation D i = (q 0 − q1 ) q 0 . When this method is used, the nominal decline is

calculated using Equation 5-16.

ai =
(1 − Di ) − 1
−b

Equation 5-16 - Initial Nominal Decline from Two Rates

The relationship between the tangential method and the two rate method is easier to

understand by observing Figure 5-12. The two rate method actually will calculate rates

above the hyperbolic rate for the first year of production and will intersect the hyperbolic

rate curve exactly at time equals year 1. The exponential tangent decline will always be

below hyperbolic rate except where they are equal at year 0. NOTE: Throughout the

rest of this manual, the effective decline used in any equation is the exponential

decline that is tangential to the hyperbolic rate curve at time zero unless specifically

declared.

Chevron Confidential Use Only Page 61 of 244

 Version 2008.01

Hyperbolic Initial Decline Methods

1000
Daily Rate

100
0 1 2
Year

Daily Rate Tangent Method Two Point Method

Figure 5-12 - Initial Hyperbolic Decline Methods

DCA software packages are not always clear on which decline needs to be entered for the

initial decline rate. A good way to determine the expected decline type is with a

hyperbolic example with an initial daily rate of 1000 BOPD, a year 1 rate of 373 BOPD,

a hyperbolic exponent of 0.4 and a one year life. The cumulative production for the first

year of production equals 225,045 STBO. To achieve these conditions, the nominal

decline would need to be 1.2040; the two rate effective decline would need to be 0.6257;

and the effective instantaneous tangential decline would need to be 0.7000.

One of the characteristics of hyperbolic declines is that on long life wells they can

generate declines that are relatively low. Low value declines cause very long producing

Chevron Confidential Use Only Page 62 of 244

 Version 2008.01

lives and can be unreasonable for some fluid systems and producing reservoirs. This

effect was listed as a common major reserves error in the 2004 SPE paper “Oil and Gas

Reserves Estimates: Recurring Mistakes and Errors” by Ron Harrell, Ryder Scott.32 To

adjust for this situation, engineers should set a minimum decline after which the

production forecast will default to an exponential decline.

The producing rate at which a given decline is reached is calculated using Equation 5-17.

One can quickly see that this equation is taken directly from Equation 5-12.

1/ b
⎡ a × qi b ⎤
qmin = ⎢ min ⎥
⎢⎣ ai ⎥⎦

Equation 5-17 - Minimum Nominal Decline

In its effective form where the decline is treated as an instantaneous decline that is

tangential to the rate time curve, the producing rate at which this decline is reached can

be calculated by Equation 5-18.

1/ b
⎡ (− ln(1 − Dmin ) × qi b ⎤
q min =⎢ ⎥
⎣ − ln(1 − Di ) ⎦

Equation 5-18 - Rate at Minimum Decline

For the previous example, calculate the producing rate at which a 10% effective decline

is reached.

Chevron Confidential Use Only Page 63 of 244

 Version 2008.01

1
⎡ ( − ln(1 − 0.10) × 100 0.5
⎤ 0.5
qmin = ⎢ ⎥ = 22 BOPD
⎣ − ln(1 − 0.20) ⎦

The reserves for the previous example would then be calculated using Equation 5-7 -

Reserves from Exponential Decline and Equation 5-14 - Hyperbolic Equation Reserves.

365 × 100 0.5 × (100 (1−0.5 ) − 22(1−0.5 ) ) 365 × (22 − 10 )

Q12 = + = 215,271 STBO
− ln(1 − 0.2) × (1 − 0.5) − ln(1 − 0.1)

This compares to a reserves estimate of 223,692 STBO for the full hyperbolic decline to

an economic limit of 10 BOPD.

Similar to Equation 5-10 for exponential declines, the reserves for a hyperbolic

production stream producing for a set numbers years can be calculated by integrating rate

over time. These equations assume that the initial daily rate qi, the initial nominal decline

ai and the hyperbolic constant b all occur at time 0. If the production between two time

points is required, the reserves from time zero to each time can be calculated and the

reserves determined by subtraction.

Chevron Confidential Use Only Page 64 of 244

 Version 2008.01

years qi
Reserves = ∫ 365 × 1
dt
(1 + b × a i × t )
0
b

1 + b × a i × t − (1 + b × a i × t ) b
Reserves years = 365 × qi × 1
(1 + b × a i × t )b × a i * (b − 1)

Equation 5-19 - Hyperbolic Reserves for a Time Period

As demonstrated, the calculation of the hyperbolic constant b is important. The usual

acceptable range is between 0 and 1. However, as discussed earlier, many times

engineers will find tight gas wells that exhibit hyperbolic declines with hyperbolic

constants greater than 1.0. The high hyperbolic constant is caused by the transient flow

nature of the well in its early production history. One should use good judgment when

calculating reserves with b factors greater than 0.5. The reason for the high exponent

needs to be well understood. One should also consider using one of the evaluation tools

that analyze transient flow behavior.

Hyperbolic equations should generally not be used to analyze wells producing in the

transient flow period. The following example generated with a transient flow well

simulator shows the error that such an analysis can cause. The well is a gas well with a

depth of 10,000 foot, a permeability of 0.025 md, an initial pressure of 6,500 psia and a

1,200 psia WBHP. Figure 5-13 shows the production which exhibits hyperbolic traits

for the first several years and then began producing with approximately a 6% exponential

decline. Producing the well to 2020, the EUR is 4.9 BCF with a final producing rate of

120 MCFPD.

Chevron Confidential Use Only Page 65 of 244

 Version 2008.01

Figure 5-14 shows the same well data but with a hyperbolic analysis performed after two

years of production becomes available. The transient flow period exhibits a hyperbolic

constant of 3.05 and the hyperbolic forecast generates an EUR of 14.1 BCF with a final

producing rate of 300 MCFPD. The EUR exceeds the OGIP and the estimation

technique significantly overestimated the initial reserves primarily because the hyperbolic

equations will eventually produce an exponential decline that is unreasonably low.

Figure 5-13 - Tight Gas Well Example

Chevron Confidential Use Only Page 66 of 244

 Version 2008.01

Figure 5-14 - Tight Gas Well Evaluated Using Hyperbolic Analysis

The example in Figure 5-14 shows where late time hyperbolic analysis greatly over

estimates the actual reserves. The reverse case can also occur where the analysis can

underestimate the reserves. This is especially the case where the engineer is using an

exponential analysis during the transient flow period where the early declines are large.

Figure 5-15 shows another example of an actual tight gas well exhibiting a hyperbolic b

greater than 1.0. One reserve estimation technique for gas wells such as this is to use the

hyperbolic formulas until a decline is reached that can be supported by other wells in the

area and then use that exponential decline for the remainder of the producing life.

Chevron Confidential Use Only Page 67 of 244

 Version 2008.01

Figure 5-15 - Hyperbolic Exponent Greater Than 1.0

Much has been written in the petroleum engineering literature and text books on how to

solve for the hyperbolic constants b, ai and Di. This was due to the very laborious nature

of the hyperbolic solution in the period before digital computers became widely

available. Today most DCA software packages can easily solve for the initial rate, initial

decline and hyperbolic exponent necessary to define a hyperbolic decline forecast. Most

of the solution techniques use iteration to minimize the variance between the production

data and the hyperbolic line equation.

Sometimes engineers need to calculate the hyperbolic constants using simple and

approximate methods. One method is called the “Three Point Method.”14 Using this

method the engineer picks three points on the production curve that provide an “eyeball”

fit of the hyperbolic curve. At each point the production rate and cumulative production

Chevron Confidential Use Only Page 68 of 244

 Version 2008.01

is recorded. The rates and cumulative production are then inserted into a rearranged

Equation 5-14 for each of the points assuming point 1 is also the initial point.

(1−b ) (1−b )
( q1 − q2 )
(a i × (1 − b)) / q1 b
=
Q12

(1−b ) (1−b )
(q 2 − q3 )
(a i × (1 − b)) / q1 b
=
Q 23

Treating the variables left of the equality sign as a constant; each equation can be set

equal to each other as per Equation 5-20.

(1−b ) (1−b ) (1−b ) (1−b )

(q1 − q2 ) (q2 − q3 )
=
Q12 Q 23

Equation 5-20 - 3 PT Method

The value of b can then be solved iteratively. The solver feature in Microsoft Excel will

quickly and easily solve for b. After b is known, the initial nominal decline ai can be

calculated.

Another simple practical method to solve for the hyperbolic constants is called the

“Initial Rate and Decline Method.” The method uses the exponential decline equations

and the early data to establish the initial nominal decline ai and initial rate qi. The method

is graphically shown by Figure 5-16 where the red line shows the initial annual effective

exponential decline of 35% and an initial rate of 1000 BOPD. These constants are then

used in Equation 5-14 to match the cumulative production between the initial rate and a

Chevron Confidential Use Only Page 69 of 244

 Version 2008.01

selected date by iteratively solving for the hyperbolic constant b. The cumulative

production is used because it ensures a good fit that matches the overall history. A

comparison check can also be made between the hyperbolic forecasted rate and the

historical rate. One could also fit the hyperbolic constant b to minimize the variance

between the hyperbolic fit and the historical production.

Hyperbolic Example

1,000
BOPD

100
1990 1991 1992 1992 1993
Year

Figure 5-16 - Hyperbolic Constants Solution Example

Another very innovative and insightful solution is shown in Determination of Oil and Gas

Reserves – Petroleum Society Monograph Number 1 by Canadian Institute of Mining,

Metallurgy and Petroleum. It was based on a work by Agbi and Ng (1987) who showed

that the hyperbolic decline can be expressed as a nonlinear equation with the hyperbolic

exponent b as the only unknown. The equation acts as a cubic equation with three real

Chevron Confidential Use Only Page 70 of 244

 Version 2008.01

roots two of which are 0 and 1. To solve the equation one needs to know two rates (qi &

q) and the time in years and cumulative production between the two rates.

The equation was originally presented in two parts. The first solved for a variable x

which is shown by Equation 5-21. The second solved for is variable y as shown by

Equation 5-23.

qi
x=
q

Equation 5-21 - Hyperbolic Cubic Solution X

Np
y=
qi × 365 × Δt years
Equation 5-22 - Hyperbolic Cubic Solution Y

( ) (
f (b) = y × x b − 1 × (1 − b ) − b × 1 − (x )
b −1
)
Equation 5-23 - Hyperbolic Cubic Solution Function

If the equations for x and y are substituted into Equation 5-23, the resulting Equation

5-24 shows the function for daily rate. If monthly rates are used, the 365 constant would

be replaced with 12.

Np ⎛ ⎛ q ⎞b ⎞ ⎛ ⎛ q ⎞ b −1 ⎞
f (b) = × ⎜ ⎟ − 1 × (1 − b ) − b × ⎜1 − ⎜⎜ i ⎟⎟ ⎟
⎜ i ⎟
qi × 365 × Δt ⎜ ⎜⎝ q ⎟⎠ ⎟ ⎜ ⎝q⎠ ⎟
⎝ ⎠ ⎝ ⎠

Equation 5-24 - Hyperbolic Two Rate b Solution

Chevron Confidential Use Only Page 71 of 244

 Version 2008.01

The plot of f(b) versus a range of b values is then constructed and the solution root other

than 0 or 1 is read from the plot. Once b is known, the nominal initial decline is

calculated using Equation 5-25.

b
⎛ qi ⎞
⎜⎜ ⎟⎟ − 1
q
ai = ⎝ ⎠
b×t

Equation 5-25 - Solving for Initial Nominal Decline

Once ai is known, Di can be calculated using Equation 5-4. A word of caution, in some

cases mathematical solution to this technique can be a negative exponent which makes

the hyperbolic solution invalid probably because of bad input data.

Example – Let’s use this technique on an example from Slider’s8 Reservoir Engineering

text. A well has the production history shown in Table 5-5. Solve for the hyperbolic

exponent and initial decline using the two rate solution graphing the f(b) function

between 0 and 1 and plotting the raw data against the hyperbolic forecast.

Month STBO Year BOPD Np

1 2,580 0.083 84.9 2,580
2 2,100 0.167 69.1 4,680
3 2,090 0.250 68.8 6,770
4 1,780 0.333 58.6 8,550
5 1,860 0.417 61.2 10,410
6 1,470 0.500 48.4 11,880
7 1,510 0.583 49.7 13,390
8 1,250 0.667 41.1 14,640
9 1,330 0.750 43.8 15,970
10 1,220 0.833 40.1 17,190
11 1,090 0.917 35.9 18,280
12 1,150 1.000 37.8 19,430
13 982 1.083 32.3 20,412

Chevron Confidential Use Only Page 72 of 244

 Version 2008.01

14 940 1.167 30.9 21,352

15 883 1.250 29.0 22,235
16 850 1.333 28.0 23,085
17 713 1.417 23.5 23,798
18 700 1.500 23.0 24,498
19 743 1.583 24.4 25,241
Table 5-5 - Slider Example Production History

25241 ⎛ ⎛ 84.9 ⎞ b ⎞ ⎛ ⎛ 84.4 ⎞ b −1 ⎞

f (b) = × ⎜⎜ ⎟ − 1⎟ × (1 − b ) − b × ⎜1 − ⎜ ⎟
19 ⎜ ⎝ 24.4 ⎠ ⎟ ⎜ ⎝ 24.4 ⎟⎠ ⎟
84.9 × 365 × ⎝ ⎠ ⎝ ⎠
12

Solving Hyperbolic Exponent

0.002

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

-0.002

-0.004
f(b)

-0.006

-0.008

-0.01

-0.012

-0.014
b

Figure 5-17 - f(b) Function for Slider Example

As shown on Figure 5-17, the third root value is 0.821. We then solve for ai using

Equation 5-25.

Chevron Confidential Use Only Page 73 of 244

 Version 2008.01

0.821
⎛ 84.9 ⎞
⎜ ⎟ −1
⎝ 24.4 ⎠
ai = = 1.3720
0.821 × 1.583

D i = 1 − e −1.372 = 0.7464

Figure 5-18 shows the best fit example for both rate and Np.

Example - 2 Rate Method Best Fit

100 30000

25000

20000

Np - STBO
BOPD

15000

10000

q bestfit
q Data
Np bestfit 5000
Np data

10 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Years

Figure 5-18 - 2 Rate Example Best Fit

What does the f(b) plot look like assuming that the hyperbolic exponent is greater than

1.0? Figure 5-19 shows the f(b) plot for the tight gas well analysis for 2 years gas

production on the tight gas well example show in Figure 5-14. Note that the three roots

are 0, 1 and 3.05.

Chevron Confidential Use Only Page 74 of 244

 Version 2008.01

Solving Hyperbolic Exponent

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5 4

-0.1
f(b)

-0.2

-0.3

-0.4

-0.5

-0.6

-0.7
b

Figure 5-19 - f(b) Function for Tight Gas Well Example

Another powerful tool that many engineers do not make maximum use of is the Solver

function in Microsoft Excel spreadsheets. With the Solver one can easily find a good fit

of the ai and b hyperbolic constants by calculating the variance between each data point

and the best fit value, summing the variances and then using the Solver to find the

hyperbolic best fit that minimizes the total variance. Variance is the difference between

the best fit and the data squared.

Let’s use our previous Slider example to solve using the solver. Figure 5-20 shows the

settings used on the Solver. The cell for b is C1 and ai C2. Thus the Solver “By

Chevron Confidential Use Only Page 75 of 244

 Version 2008.01

Changing Cells” setting is $C$1:$C$2. Note that the Min is checked and that the

hyperbolic exponent b cell $C$1 is constrained between 0 and 1.

Figure 5-20 - Solver Hyperbolic Settings

The Solver results are listed as follows.

Solver Results
b 0.8367
ai 1.4931
di 0.7753

Table 5-6 shows the results for the Best Fit rates and the variance using the Solver

results. Note that there are three data points that have some significant variance.

Month Years Data BOPD Best Fit BOPD Variance

1 0 84.9 84.9 0.000
2 0.0830 69.1 75.5 40.407
3 0.1660 68.8 67.8 1.044
4 0.2490 58.6 61.4 7.963
5 0.3320 61.2 56.1 26.206

Chevron Confidential Use Only Page 76 of 244

 Version 2008.01

6 0.4150 48.4 51.5 9.827

7 0.4980 49.7 47.6 4.316
8 0.5810 41.1 44.2 9.755
9 0.6640 43.8 41.2 6.531
10 0.7470 40.1 38.6 2.207
11 0.8300 35.9 36.3 0.142
12 0.9130 37.8 34.2 13.056
13 0.9960 32.3 32.3 0.000
14 1.0790 30.9 30.6 0.084
15 1.1620 29 29.1 0.005
16 1.2450 28 27.7 0.112
17 1.3280 23.5 26.4 8.296
18 1.4110 23 25.2 4.846
19 1.4940 24.4 24.1 0.081
Total Variance 134.878
Table 5-6 - Best Fit Table for Solver Example

Figure 5-21 shows the best fit rates and the data points.

Hyperbolic Best Fit Using Excel Solver

100
Data BOPD
Best Fit BOPD

Solver Results
b = 0.8367
ai = 1.4931
di = 0.7753

Daily Rate

10
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
Years

Figure 5-21 - Best Fit Graph for Solver Example

Chevron Confidential Use Only Page 77 of 244

 Version 2008.01

5.3 Harmonic Declines

The harmonic decline is a special case of hyperbolic decline, where the decline rate is

proportional to the production rate. This occurs when the hyperbolic constant b is equal

to 1. The production at any point, the production between two points, and the time to the

economic limit can be calculated using Equation 5-26, Equation 5-27 and Equation 5-28.

The 365 constant in Equation 5-27 is used if the rate is in daily units. If monthly units are

used, the constant 12 is used. If annual units are used, the constant 1.0 is used.

qi
q=
1 + ait
Equation 5-26 - Harmonic Decline Rate

⎛q ⎞
qi × ln⎜⎜ i ⎟⎟
Np = ⎝ q ⎠ × 365
ai

Equation 5-27 - Harmonic Decline Reserves

Chevron Confidential Use Only Page 78 of 244

 Version 2008.01

⎛ qi ⎞
⎜⎜ ⎟⎟ − 1
q
t el = ⎝ el ⎠
ai

Equation 5-28 - Harmonic Decline Remaining Life

With the exception of strong water drive (or waterflood) reservoirs, harmonic declines

are rarely observed.1 (SPE Reserves Manual). Brons25 also recognized that harmonic

declines will mostly occur in water drive reservoirs with heavy oil and a poor mobility

ratio.

Fekete in their RTA documentation state “Harmonic decline is a special case of

hyperbolic decline, with b = 1, i.e., the decline rate, D, is proportional to q. This means

that the decline rate, D, goes to zero when q approaches zero. This type of performance

is expected when very effective recovery mechanisms such as gravity drainage are

active.”

According to the SPE PE Handbook11, harmonic declines do not occur very often and

extrapolation of them usually provides a projection that is too optimistic even for reserves

classified as possible unproved reserves. Harmonic declines should only be used where it

has been firmly demonstrated that the best fit of the production history results in a

hyperbolic constant of b equal to 1.0 and the best fit is adequately justified.

Chevron Confidential Use Only Page 79 of 244

 Version 2008.01

Example – For our previous example, calculate the ultimate recovery at the economic

limit of 10 BOPD. Also generate the production forecast and plot the rate versus

cumulative recovery using a semi-log graph scale.

⎛ 100 ⎞
100 × ln⎜ ⎟
⎝ 10 ⎠ × 365 = 376,638 STBO
Q1 2 =
− ln(1 − 0.2)

Harmonic Example

100
BOPD

10
0 100 200 300 400
Cumulative Production - MBO

Figure 5-22 - Harmonic Decline on Semi-log Coordinates

As shown by Figure 5-22, a harmonic decline will plot a straight line on a semi-log rate

versus cumulative production curve. An easy check to determine if a field is exhibiting a

harmonic decline is to construct a semi-log rate versus cumulative plot and visually look

for straight lines.

Chevron Confidential Use Only Page 80 of 244

 Version 2008.01

5.4 Comparison of Exponential, Hyperbolic and Harmonic

Declines

As shown there is considerable difference in reserves volumes between the three types of

declines. The exponential decline example had 147 MBO of ultimate recovery. The

hyperbolic example generated 224 MBO, 1.5 times the exponential. The harmonic

example increased the recovery to 377 MBO, 2.6 times the exponential. Given the wide

range, one can easily see that it is important that the correct decline type is selected.

Why do reservoirs exhibit certain types of decline? Viewed in the macro, there are many

factors that control how a reservoir will deplete. These factors categorize easily into

reservoir characteristics, reservoir fluids, drive mechanisms and operational activities.

Some are provided by nature and are largely uncontrollable. Reservoir characteristics

include depth, drainage area, thickness, formation dip, lithology, permeability,

wettability, relative permeability, layering and compartmentalization. Reservoir fluids

include fluid type, phase behavior, fluid compressibility and viscosity. Primary drive

mechanisms are volumetric, water drive or encroachment, gravity drainage and rock

compaction. Others factors are more controllable. They include well drawdown, surface

pressures, prorated production, workovers, drilling schedules, well spacing and

configuration, fluid or gas injection, downtime, scale, paraffin, completion practices,

equipment sizing and limitations, operating practices and personnel changes.

Chevron Confidential Use Only Page 81 of 244

 Version 2008.01

Although the factors that affect depletion are complex and difficult to represent

mathematically, we can gain some insight from the simple use of Darcy’s Law for oil

wells. To determine this insight let’s ask the question what must the reservoir pressure

depletion history be to generate an exponential or a hyperbolic decline?

Let’s investigate with an example. An oil well with a 4000 psia initial reservoir pressure

was placed on production with a 250 psia working bottom-hole pressure at a rate of 400

STBOPD. Assuming fluid, reservoir and completion variables remain constant, calculate

what reservoir pressure depletion forecast must be if the well produces with a 20%

exponential decline until the reservoir pressure reaches the bubble point at 1500 psia.

Compare that depletion forecast to a hyperbolic decline with an initial decline of 20% and

a hyperbolic constant 0.5 using Equation 5-13.

First calculate the productivity index using Equation 3-3.

400
Productivity Index (PI) = = 0.107 BOPD/PSI
(4000 − 250)

Using the 0.107 BOPD/PSI productivity index, the producing rate can be calculated for

various reservoir pressures. These rates can then be used to calculate the cumulative

production for both the exponential and hyperbolic cases using Equation 5-7 and

Equation 5-14. The following table and graph shows the results.

Chevron Confidential Use Only Page 82 of 244

 Version 2008.01

MSTBO
Pressure BOPD Exponential Hyperbolic
Cumlative Cumulative
4,000 400 0 0
3,750 373 44 44
3,500 347 87 90
3,250 320 131 138
3,000 293 174 188
2,750 267 218 240
2,500 240 262 295
2,250 213 305 353
2,000 187 349 415
1,750 160 393 481
1,500 133 436 553

Table 5-7 - Exponential versus Hyperbolic Decline

Exponential Versus Hyperbolic Example

4,500
4,000
Reservoir Pressure - PSI

3,500
3,000 Exponential
2,500 Cumlative
2,000 Hyperbolic
1,500 Cumulative
1,000
500
-
0 200 400 600
Np - MSTBO

Figure 5-23 - Exponential versus Hyperbolic

Chevron Confidential Use Only Page 83 of 244

 Version 2008.01

As shown, the exponential case requires a linear relationship between reservoir pressure

and cumulative production. The hyperbolic case requires a flattening and extension of

the curve. Experienced reservoir engineers will recognize that the exponential curve

mimics oil reservoirs producing above the bubble point and the hyperbolic curve mimics

oil reservoirs below the bubble point or those producing with some water encroachment.

Many solution gas drive oil reservoirs will produce with an exponential decline above the

bubble point where the oil compressibility is relatively constant and with a hyperbolic

decline below the bubble point when the reservoir effective compressibility begins

increasing when free gas forms in the reservoir. Figure 5-24 shows a solution gas drive

field that exhibits these characteristics.

Chevron Confidential Use Only Page 84 of 244

 Version 2008.01

Figure 5-24 - Hyperbolic Decline due to Solution Gas Breakout - Williston Basin Example

As shown by Figure 5-23, the relationship between cumulative production and reservoir

pressure is linear for the exponential decline case. Figure 5-25 shows examples from a

well simulator of how exponential declines change with varying fluid compressibility.

This particular case is for a well with the following conditions:

k = 20 md

h = 10 ft

μo = 0.5 cp

Chevron Confidential Use Only Page 85 of 244

 Version 2008.01

Βo = 1.1 RB/STB

re = 3000 ft

rw = 0.375 ft

s = 0

Note that each example gives an exponential decline once the transient effects have

dissipated and the well is in pseudo-steady-state. Also notice that as compressibility

increases, the decline becomes less.

Exponential Declines from Varying Fluid Compressibilities

1000

D = 3.77% per Month

D = 6.78% per Month

BOPD

D = 32.9% per Month

100
0 100 200 300 400 500 600 700 800
DAYS

3.00E-05 1.65E-05 3.00E-06

Figure 5-25 - Declines Varying with Fluid Compressibility

Chevron Confidential Use Only Page 86 of 244

 Version 2008.01

We have shown that an exponential decline exists in a volumetric depletion with a

constant fluid compressibility. Now let’s investigate how oil compressibility would have

to change for various hyperbolic b factors.

Example – A well is producing in pseudosteady-state with the conditions shown below.

Calculate the exponential decline and develop a rate time and rate cumulative forecast

assuming the oil compressibility remains constant and the pressure is depleted down to

1400 psia. Assuming that the hyperbolic initial decline is equal to the exponential

decline, calculate the required fluid compressibility increase necessary for the well to

produce with a hyperbolic factor of 0.001 (exponential), 0.5 and 0.999 (harmonic).

Bo = 1.15, RB/STB

μo = 2, CP

h = 25, feet
k = 50, md
re = 3000, feet

rw = 0.4, feet

s = 0

sw = 0.25

Φ = 0.22

pi = 4000, psia

pwf = 400, psia

acres = 650, acres

OOIP = 18,063, MSTBO

co = 3.00E-05, psi-1

7.07 × 10 −3 × 50 × 25 × ( 4000 − 400)

qi = = 1,642 BOPD
⎛ 3000 ⎞
1.15 × 2 × (ln⎜ ⎟ − 0.5 + 0)
⎝ .4 ⎠

Chevron Confidential Use Only Page 87 of 244

 Version 2008.01

7.07 × 10 −3 × 50 × 25 × (1400 − 400)

q= = 456 BOPD
⎛ 3000 ⎞
1.15 × 2 × (ln⎜ ⎟ − 0.5 + 0)
⎝ .4 ⎠

Np = c o × OOIP × (p i − p a ) = 0.00003 × 18063 × ( 4000 − 1400) = 1,409 MSTBO

365 × (1642 − 456)

ai = = 0.3072
1409000

D i = 1 − e −0.3072 = 0.265

Table 5-8 shows the oil compressibilities necessary for hyperbolic constants of

exponential (0.001), hyperbolic (0.5) and harmonic (1.0). Column 1 contains the

reservoir pressure. Column 2 calculates the Darcy Law oil rate at that reservoir pressure.

The Np column calculates the cumulative production necessary to match the b factor, the

initial decline and the Darcy oil rate for each reservoir pressure. The Co column is the oil

compressibility necessary to produce the cumulative volume for that b factor using

material balance.

Darcy b=0.001 b=0.5 b=0.999

PSIA BOPD Np - MBO Co Np - MBO Co Np - MBO Co

4000 1,642 0 3.00E-05 0 3.00E-05 0 3.00E-05

3800 1,551 108 3.00E-05 110 3.04E-05 111 3.09E-05
3600 1,460 217 3.00E-05 223 3.09E-05 230 3.18E-05
3400 1,369 325 3.00E-05 340 3.14E-05 356 3.28E-05
3200 1,277 434 3.00E-05 461 3.19E-05 490 3.39E-05
3000 1,186 542 3.00E-05 586 3.24E-05 635 3.51E-05
2800 1,095 650 3.00E-05 716 3.30E-05 791 3.65E-05
2600 1,004 759 3.00E-05 852 3.37E-05 960 3.80E-05
2400 912 867 3.00E-05 993 3.44E-05 1146 3.97E-05

Chevron Confidential Use Only Page 88 of 244

 Version 2008.01

2200 821 976 3.00E-05 1143 3.51E-05 1352 4.16E-05

2000 730 1084 3.00E-05 1301 3.60E-05 1581 4.38E-05
1800 639 1193 3.00E-05 1468 3.70E-05 1842 4.63E-05
1600 547 1301 3.00E-05 1649 3.80E-05 2142 4.94E-05
1400 456 1410 3.00E-05 1845 3.93E-05 2497 5.32E-05
Table 5-8 - Oil Compressibilities for Various Hyperbolic Constants

Figure 5-26 graphs the oil compressibility versus reservoir pressure for the range of

hyperbolic b constants. As expected, the exponential case has a constant compressibility.

As the b factor increases thus raising the volume recovered for a given reservoir pressure,

the oil compressibility must increase. When the decline is harmonic, the oil

compressibility has to increase the initial compressibility by a factor of almost 1.8 when

the reservoir reaches 1400 psia. There are very few PVT systems that can achieve that

increase without adding energy from water injection or water influx. Thus one would

expect harmonic declines in volumetric depletion systems to be rare.

Chevron Confidential Use Only Page 89 of 244

 Version 2008.01

Oil Compressibility Required for Hyperbolic Constants

5.5E-05
b=0.001
b=0.1
b=0.3
5.0E-05 b=0.5
b=0.7
b=0.999
4.5E-05
Oil Compressibility - PSI-1 .

4.0E-05

3.5E-05

3.0E-05

2.5E-05

2.0E-05
1000 1500 2000 2500 3000 3500 4000
Reservoir Pressure

Figure 5-26 - Oil Compressibilities versus Hyperbolic Constants

Hyperbolic declines also occur because of reservoir layering. Many wells will

demonstrate a log normal distribution for permeability. In these wells the net pay with

low permeability exceeds the pay with high permeability. When these wells are placed

on production, the high permeability streaks produce at high rates but rapidly deplete.

The lower permeability sections produce at a lower rate but have a longer depletion time.

When all of the sections are summed, the effect will be a production trend which appears

hyperbolic.

Chevron Confidential Use Only Page 90 of 244

 Version 2008.01

To demonstrate this effect, a multi-zone well was analyzed using a well simulator. The

well net pay included 1 foot @ 10 md, 2 feet @ 5 md, 4 feet @ 2.5 md, 8 feet @ 1.25 md

and 16 feet at 0.625 md. All of the other parameters were left constant with the exception

of water saturation which was varied by permeability. Table 5-9 gives all of the

parameter for each zone. It also includes an arithmetic total or average.

Zone 1 2 3 4 5 Total or Average

Initial Pressure (psia) 4200 4200 4200 4200 4200 4200
Porosity 0.18 0.18 0.18 0.18 0.18 0.18
Sw 0.5 0.45 0.4 0.35 0.3 0.46
Permeability (md) 0.625 1.25 2.5 5 10 1.613
Net Pay (Feet) 16 8 4 2 1 31
Skin -1 -1 -1 -1 -1 -1
Oil Gravity 0.8 0.8 0.8 0.8 0.8 0.8
Reservoir Temperature (F) 195 195 195 195 195 195
Sgor (SCF/BO) 375 375 375 375 375 375
Gas Gravity 0.7 0.7 0.7 0.7 0.7 0.7
Drainage Radius (feet) 1500 1500 1500 1500 1500 1500
Drainage (Acres) 162 162 162 162 162 162
Bo 1.21 1.21 1.21 1.21 1.21 1.21
OOIP - MSTBO 1,498 824 449 243 131 3,135
Table 5-9 - Layered Well Example

Figure 5-27 show the production for each of the zones along with the total summation of

all the zones. Since the kh was kept constant for each zones, all zones have a similar

starting production rate. Note that the 10 md zone depletes in 1.5 years while the 0.625

md zone is still producing at the end of the five year period. The total summation for all

of the zones demonstrates a distinct hyperbolic nature with an initial decline of 53% and

a b factor of 0.7.

Chevron Confidential Use Only Page 91 of 244

 Version 2008.01

Layered Well Example

1000
0.625 md 1.25 md 2.5 md 5 md 10 md Total

Hyperbolic Constants
Di = 53%
b Factor = 0.70

100
BOPD .

10

1
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Years

Figure 5-27 - Layered Well Example - Rate by Zone

Figure 5-28 compares the production for the layered solution with a single zone solution

using the same well simulator. Note that the single zone produces with an exponential

decline once the transient flow period is over. The rates are comparable to start with but

the single zone predicts a higher rate for the first 3.5 years.

Chevron Confidential Use Only Page 92 of 244

 Version 2008.01

Layered Versus Single Zone

1000

1 Zone Layered

BOPD .

100

10
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Years

Figure 5-28 - Layered Well versus Single Zone Well

Figure 5-29 summarizes the characteristics for each of the Arps decline curve

classifications, exponential, hyperbolic and harmonic. The example well has an initial

production rate of 1000 BOPD with a 30% initial effective decline rate. The hyperbolic

well was given a hyperbolic constant b of 0.5. The harmonic well by definition was

given a hyperbolic constant of 1.0. The four plots show rate versus time, semi-log rate

time, rate versus cumulative production and semi-log rate versus cumulative production.

Note that only three plots, exponential semi-log rate time, exponential rate cumulative

and harmonic semi-log rate cumulative generate straight lines.

Chevron Confidential Use Only Page 93 of 244

 Version 2008.01

Rate Time Rate Cumulative

1200 1200

1000 Exponential 1000 Exponential

Hyperbolic Hyperbolic
800 800
Harmonic Harmonic

BOPD
BOPD

600 600

400 400

200 200

0 0
0 2 4 6 8 10 0 500 1000 1500

Years Np - MSTBO

Semi-log Rate Time Semi-log Rate Cumulative

1000 1000

BOPD
BOPD

100 100
Exponential
Exponential
Hyperbolic
Hyperbolic
Harmonic
Harmonic

10 10
0 2 4 6 8 10 0 500 1000 1500

Years Np - MSTBO

Figure 5-29 - Comparison of Arps DCA Classifications

5.5 Hyperbolic Declines for Gas Wells

The decline for gas wells is slightly different because Darcy’s Law for gas wells has the

(Pe2-Pw2) component rather than the oil well (Pe-Pw) component. This change along with

the fact that gas compressibility, gas viscosity, z factor and formation volume factor

change with reservoir pressure makes most gas wells produce with a hyperbolic decline.

Chevron Confidential Use Only Page 94 of 244

 Version 2008.01

Because of the single phase nature of most gas reservoirs, the material balance equations

are much simpler provided that water influx is not occurring. A plot of the reservoir

pressure adjusted for gas compressibility versus cumulative gas production is linear.

Thus we can easily solve for the rate using Darcy’s Law.

Let’s investigate with an example. A 10,000 foot gas well contains 11.8 BCF of dry gas

with a specific gravity of 0.65 and an initial pressure of 5000 psia. The reservoir

temperature is 250 degrees Fahrenheit. The net pay is 19.6 feet and the permeability is 3

md. The drainage area is 640 acres and of the wellbore is 0.5 feet. Assume laminar

conditions (not turbulent) and pseudosteady-state flow. Make a production forecast plot

assuming a constant bottom-hole pressure of 1500 psia and a reservoir abandonment

pressure of 1750 psia.

The summary Table 5-10 - Hyperbolic Gas Well Example, shows the pressures, rates and

fluid conditions for 250 psia increments. The time in years was determined by

integrating with pressure increments.

Avg
Pe Pe Z Z ug MCFPD Bg GIP Gp Pe/Z RE Years D
5,000 1.025 0.976 0.0244 6,781 0.00412 11,815 0 4,880 0.0% 0.00
4,750 1.009 0.968 0.0238 6,260 0.00427 11,393 422 4,706 3.6% 0.18 0.363
4,500 0.995 0.961 0.0232 5,739 0.00445 10,951 864 4,523 7.3% 0.38 0.350
4,250 0.981 0.954 0.0225 5,218 0.00464 10,487 1,328 4,331 11.2% 0.61 0.336
4,000 0.969 0.948 0.0219 4,699 0.00487 10,000 1,815 4,130 15.4% 0.88 0.323
3,750 0.957 0.942 0.0213 4,181 0.00513 9,489 2,326 3,919 19.7% 1.19 0.310
3,500 0.946 0.937 0.0207 3,666 0.00544 8,954 2,861 3,698 24.2% 1.57 0.297
3,250 0.937 0.932 0.0200 3,157 0.00580 8,395 3,420 3,467 28.9% 2.02 0.284
3,000 0.930 0.928 0.0194 2,654 0.00623 7,812 4,003 3,226 33.9% 2.57 0.271
2,750 0.924 0.925 0.0188 2,162 0.00676 7,205 4,609 2,976 39.0% 3.26 0.257
2,500 0.920 0.924 0.0183 1,683 0.00740 6,577 5,237 2,717 44.3% 4.15 0.244
2,250 0.919 0.923 0.0177 1,221 0.00821 5,931 5,884 2,450 49.8% 5.37 0.231

Chevron Confidential Use Only Page 95 of 244

 Version 2008.01

2,000 0.919 0.923 0.0172 783 0.00924 5,269 6,545 2,176 55.4% 7.18 0.218
1,750 0.922 0.924 0.0167 374 0.01059 4,597 7,218 1,899 61.1% 10.36 0.207
Table 5-10 - Hyperbolic Gas Well Example

Hyperbolic Gas Well

10,000 50%

Rate
D
45%

40%

Decline - %
MCFPD .

1,000 35%

30%

25%

100 20%
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Years

Figure 5-30 - Hyperbolic Gas Well

The semi-log rate time plot Figure 5-30 shows that the gas well has a hyperbolic shape.

It also shows the instantaneous exponential decline and how it changes with time. The

initial effective decline was calculated from the curve to be 36%. The best fit hyperbolic

constant to minimize variance is 0.39. You will also note by observing the rate of change

in the decline that in later years the decline will stabilize around a 20% decline. In 1973

Fetkovich published a remarkable paper in which he showed that exponential decline is

the long time solution to the diffusivity equation with constant wellbore pressure.

Chevron Confidential Use Only Page 96 of 244

 Version 2008.01

The hyperbolic nature of gas wells is further shown by Figure 5-31. It shows a type

curve well for the average USA gas wells drilled in the lower 48 states in 1996 that was

compiled by the Energy Information Agency. As shown, this average well shows a

definite hyperbolic decline with an initial decline of 53% and a hyperbolic constant of

0.56.

Lower-48 1996 Vintage Average Gas Well

10,000

Best Fit
Hyperbolic Exponent = 0.56
1,000 Initial Decline = 53% 1996 Average Well
Hyperbolic Curve Fit
Mcf/day

100

10
0 50 100 150 200 250
Months

Figure 5-31 - USA Lower 48 States Average Gas Well Showing Hyperbolic Effects

As stated previously, the reason for the hyperbolic shape is that the gas compressibility

(cg=1/P), viscosity, z factor and Bg change with the depletion of the reservoir. Plots of

Chevron Confidential Use Only Page 97 of 244

 Version 2008.01

these variables versus depletion pressure for our gas well example are shown by Figure

5-32.

Formation Volume Factor versus Pressure Z Factor versus Pressure

6,000 6,000
5,000
Reservoir Pressure

5,000

Reservoir Pressure
4,000 4,000
3,000 3,000
2,000 2,000
1,000 1,000
0 0
0.004 0.005 0.006 0.007 0.008 0.009 0.01 0.011 0.92 0.93 0.94 0.95 0.96 0.97 0.98
Formation Volume Factor Bg Z Factor

Gas Viscosity versus Pressure Gas Compressibility versus Pressure

6,000 6,000
5,000 Reservoir Pressure 5,000
Reservoir Pressure

4,000 4,000
3,000 3,000

2,000 2,000

1,000 1,000

0 0
0.0150 0.0170 0.0190 0.0210 0.0230 0.0250 0.0002 0.0003 0.0004 0.0005 0.0006
Gas Viscosity - cp Gas Compressibility - 1/PSI

Figure 5-32 - Gas PVT Properties versus Depletion Pressure

As shown, the hyperbolic solution, while not exact for gas wells, can be used as a rough

approximation for forecasting reserves. In some cases, hyperbolic DCA will slightly over

predict the reserves but the difference is well within the quality of the data that one is

generally required to use. As shown by the instantaneous decline in Figure 5-30, the

exponential decline is continuously getting lower with time. For this reason, exponential

declines will under predict the reserves and the petroleum engineer will constantly find

oneself having to annually add reserves. A good pragmatic compromise between the two

methods is to use hyperbolic declines until a reasonable decline is reached and then

default to a continuous exponential decline.

Chevron Confidential Use Only Page 98 of 244

 Version 2008.01

5.6 Secondary Product Stream Reserves – Solution Gas or

Condensate

The secondary product streams that are generated during DCA include solution gas from

oil reservoirs and condensate from wet gas reservoirs. The projection of secondary

product streams is more difficult than the projection of primary product.5 There is no

totally satisfactory single method of making the projection but several useful

approximations are available.

Under-saturated oil reservoirs that are pressure depleted pass through four phases. Phase

I is when the reservoir is above the bubble point. During this period the reservoir drive is

provided by liquid expansion and the reservoir will generally produce with an

exponential decline. Phase II begins when the bubble point is reached but the gas

saturation is below critical and the gas is not mobile. The free gas supplements the fluid

expansion reservoir drive lessening the decline. Phase III occurs once the critical gas

saturation is reached and gas become mobile. It continues while the gas evolving

exceeds the gas production and the GOR increases. Phase IV begins when the gas

evolving is less than the gas being produced causing the GOR to decrease. Figure 5-33

shows the four phases for a typical solution gas oil reservoir.

Chevron Confidential Use Only Page 99 of 244

 Version 2008.01

Solution G as Reservoir Phases .

Bubble
Point Phase
III
Phase Phase
Phase IV
I II Critical Gas
Saturation

Cumulative Oil

Oil Rate G OR

Figure 5-33 - Solution GOR Oil Reservoir Phases

It shows the difficulty of predicting the solution gas production from an oil reservoir

because of the tremendous variability in PVT properties on different fields. On large

reservoirs, material balance or simulation should probably be used. Material balance

however may not be time or cost efficient on smaller reservoirs requiring alternate

methods.

The most common method used to predict the remaining gas reserves is to multiply the

oil reserves by the latest GOR. As one can see from Figure 5-33, this will usually give a

conservative gas estimate except in Phase IV when the GOR begins decreasing and the

method produces an overly optimistic estimate. At this point, however, the oil production

is generally close to depletion making the gas reserves quite low.

Chevron Confidential Use Only Page 100 of 244

 Version 2008.01

The constant GOR method also works very well in waterfloods once fill-up has been

reached and no free gas exists in the reservoir causing a stable GOR.

Another method for predicting solution gas was originally proposed by J. J. Arps in 1956.

He plotted logarithmic cumulative oil production versus logarithmic cumulative gas

production and extrapolated the gas ultimate recovery to the latest ultimate oil recovery.

Thompson and Wright5 point out that this method is represent by Equation 5-29 where m

is the slope and K is the y value as x equals 1. The slope m can be calculated using two

points from the best fit line. K can be solved as the unknown variable using one point

and the calculated slope. They also noted that when the GOR is constant, the slope of the

line will be 1.0. Figure 5-34 shows an example plot for the method proposed by Arps.

G p = K × (Np )
m

Equation 5-29 - Secondary Product Equation

Chevron Confidential Use Only Page 101 of 244

 Version 2008.01

Table Rock Almond Lewis

1,000
Cummulative Gas - BCF .

100

10
0.1 1.0 10.0
Cummulative Oil - MMBO

Figure 5-34 - Arp's Secondary Product Method

Cronquist in the SPE Reserves Manual1 proposed using this method but suggests that the

cumulative oil scale be linear instead of logarithmic. Care must be taken when using

either of these methods to ensure that the falling GOR in late life of a field does not cause

an overly optimistic reserves estimate. Figure 5-35 shows what the plot using Croquist’s

suggested semi-log plot.

Chevron Confidential Use Only Page 102 of 244

 Version 2008.01

Table Rock Almond Lewis

1,000
Cummulative Gas - BCF .

100

10
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Cummulative Oil - MMBO

Figure 5-35 - Cronquist Secondary Product Stream Method

In many cases, a simple linear plot provides a very clear trend which can easily be

extrapolated to get a secondary product. The author prefers this plot where it is useable.

Figure 5-36 shows the linear plot for the example field.

Chevron Confidential Use Only Page 103 of 244

 Version 2008.01

Table Rock Almond Lewis

160

140

120
Cummulative Gas - BCF .

100

80

60

40

20

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Cummulative Oil - MMBO

Figure 5-36 - Linear Plot for Secondary Products

In many gas fields liquid yield remains constant or increases slightly with time. In these

fields calculating the condensate reserves using a constant liquid yield is reasonable.

Some fields have GORs and liquid yields that change and can be represented by

exponential declines on semi-log versus time plots or linear plots. These plots can be

combined with a primary product exponential rate versus time plots to calculate the

remaining gas or condensate reserves. Where both the primary product rate and the ratio

produce with exponential declines, the secondary product reserves can be generated by

integrating the product of rate and ratio for the remaining life in years as shown in

Equation 5-30.

Chevron Confidential Use Only Page 104 of 244

 Version 2008.01

∫ ((365 × q × e )× (ratio × e ))× dt

life
−arate × t −aratio ×t
Secondary Reserves = i i
0

Equation 5-30- Secondary Reserves Integration

After performing the integration between the current time 0 and the remaining life in

years, the secondary reserves can be calculated using Equation 5-31 where qi is the

current rate, ratioi is the current GOR or liquid yield, arate is the nominal exponential

decline for the primary product and aratio is the nominal exponential decline of the GOR

or liquid yield.

e −life × (arate +aratio ) − 1

Secondary Re serves = −365 × qi × ratio i ×
a rate + a ratio

Equation 5-31 - Secondary Reserves Exponential Decline

Example – An oil reservoir is producing 900 BOPD with an effective annual decline of

20% and 9 years of remaining life. The GOR is 450 SCF/STBO and is decreasing with

an effective annual decline of 10%. Calculate the remaining oil reserves, gas reserves

and average GOR.

a rate = − ln(1 − 0.20) = 0.2231

a ratio = − ln(1 − 0.10) = 0.1054

q el = 900 × e −9×0.2231 = 121 BOPD

Chevron Confidential Use Only Page 105 of 244

 Version 2008.01

0.365 * (900 − 121)

Qo = = 1274 MSTBO
0.2231

e −9 × (0.2231 + 0.1054 ) − 1
Qg = −365 × 900 × 0.450 × = 426600 MCF
0.2231 + 0.1054

426600
Average GOR = = 335 SCF / STBO
1274

Equation 5-31 can also be used for a series of GOR declines. For example, a field may

have solution gas projection that includes a 2% effective decline for 3 years through

Phases I and II, an incline for 7 years at a -10% effective decline through Phase III and a

50% decline for 2 years during Phase IV. This forecast could be combined with an

exponential decline for the oil forecast where the oil rates can be calculated at the start of

each decline and gas reserves under each section determined and summed.

A similar equation can be developed where the GOR forecast is linear instead of

exponential. This equation is longer and more complicated to use. Since the GOR is

linear and thus increases or decreases a fixed volume per year, this calculation can easily

be performed using a spreadsheet.

Chevron Confidential Use Only Page 106 of 244

 Version 2008.01

6 Decline Curve Analysis in Water Drive and Waterflood

Reservoirs

As introduced briefly in Section 4.5, another type of decline occurs on reservoirs that are

under waterflood or experiencing a strong water drive. If voidage is maintained so the

production in reservoir barrels is balanced by injection, the average pressure in the

reservoir and the withdrawal rate remains somewhat constant. The decline occurs

because of the displacement of oil by water and thus the water-oil-ratio or water cut is

increasing. This occurs because of fractional flow, sweep efficiency and the

displacement nature of these injection processes. Simply put, the oil rate is being

replaced with water rate.

The concept of declines caused by water displacement is quite different from declines

that occur because of pressure depletion. Once a waterflood is underway and voidage is

being maintained, any primary recovery in the reservoir ceases and all recovery is due to

displacement. In waterfloods, production can be increased or decreased and the

production decline changed simply by increasing or decreasing injection. The engineer

can also make the injection more efficient through good reservoir management practices.

The amount of effective injection in a waterflood is commonly called the reservoir

processing rate.

Chevron Confidential Use Only Page 107 of 244

 Version 2008.01

Effective use of injection is an important concept. Waterflood need to be managed from

the injection wells while depletion reservoirs need to be managed from the producing

wells.

6.1 Water Ratio Methods for Waterfloods and Water Drive

Reservoirs

Although the conventional Arps equations can be used for waterflood recovery

predictions, for the reasons discussed earlier in this section, most evaluations also

includes a forecast of one of the water versus oil ratio methods. The most common

plotting methods used are:

ƒ Semi-log Water-Oil-Ratio (WOR) versus linear cumulative oil (Np)

ƒ Semi-log oil cut (fo) versus linear Np

ƒ Semi-log water cut (fw) versus linear Np

ƒ Linear oil cut (fo) versus linear Np

ƒ Linear water cut (fw) versus linear Np

ƒ Ershaghi X plot versus linear Np

There is much confusion in the engineering community as to which method to utilize and

with good reason. Although intuitively one would think that the methods would be close,

because logarithms are used in the equations, the reserves estimates can be very different.

Chevron Confidential Use Only Page 108 of 244

 Version 2008.01

To test the relationship between the various methods, the following exercise was

conducted using a randomly generated 20,000 point synthetic dataset. The test method is

described as follows:

Test method - An oil field has been under water injection for a decade.

After a cumulative oil production represented by variable A has been

produced, the water cut is variable B. Several years later when the

cumulative production had increased to variable C, the water cut had

increased to variable D. Assuming that the points (A,B) and (C,D)

represent a straight line for each of the methods, calculate the remaining

reserves using an economic limit water cut of 95%. Make the reserves

calculations for linear oil cut, linear water cut, log oil cut, log water cut

and log of water oil ratio versus cumulative oil production Np. Investigate

the relationship between the various methods using a random generation

of more than 20,000 sets of cumulatives and water cuts.

Figure 6-1 shows the comparison of the results between some of the methods. As shown

in the upper left graph, there is a one-to-one correlation when calculating reserves using

linear oil cut versus Np and linear water cut versus Np. This makes sense because the

equation for oil cut is f o = 1 − f w is a linear relationship.

Chevron Confidential Use Only Page 109 of 244

 Version 2008.01

Figure 6-1 - Ratio Method Graphs

The upper right graph of Figure 6-1 shows the comparison between log of oil cut and

linear oil cut. The yellow points show the individual data points where the scatter is

cause by the logarithmic values. The linear line is the least squares best fit through the

individual points. The linear oil cut will calculate reserves that are roughly 50% of those

from the log of oil cut.

The lower left graph in Figure 6-1 compares the log of water cut method and linear water

cut method. The linear water cut method calculated roughly 9% more reserves than the

log water cut method. Water cut methods are sometimes preferred by engineers because

Chevron Confidential Use Only Page 110 of 244

 Version 2008.01

the upper limit of water cut is 100% while the log of oil cut method and log of WOR has

no upper mathematical limit.

The lower right graph in Figure 6-1 compares the log oil cut method and the log water cut

method. Although the point scatter is wider, the log of water cut method calculates

slightly less than half the reserves of log oil cut. This demonstrates the effect of taking

the logarithm of oil cut versus water cut.

Figure 6-2 shows the comparison between the two most commonly used methods, the log

of WOR and log of oil cut. As shown, the WOR method calculated reserves which were

10% less than the log of oil cut method. Note that the data fit is quite tight for the

comparison.

Figure 6-2 - Log Oil Cut versus Log WOR Methods

Chevron Confidential Use Only Page 111 of 244

 Version 2008.01

Figure 6-3 shows a bar graph representing the total summation of the reserves produced

for the 20,000 synthetic data cases for each of the methods. This figure is a numerical

summation representation of the graphs shown by Figure 6-1 and Figure 6-2. As shown,

the log oil cut versus Np method produces the highest reserves estimate followed by the

log WOR versus Np method.

Method Comparison

8000
7000 6324
5716
6000
Reserves 5000 3215
4000 2931
Units 3000
2000
1000
0
Log Log Oil Linear Log
WOR Cut Oil Cut Water
Cut

Figure 6-3 - Synthetic Data Comparison of Ratio Methods

The reserves difference between the water ratio methods is very dependent on the final

economic limit ratio. It is important that the engineer understand the effect of a using a

higher economic limit when log oil cut or log WOR methods are use. To demonstrate

this, the following example is given.

Test Method Example - A field had recovered 50 million barrels of oil

when the water cut reached 70%. After a further 15 million barrels of oil

production, the field was producing with a 75% water cut. Show what the

Chevron Confidential Use Only Page 112 of 244

 Version 2008.01

remaining reserves beyond the 65 million barrel oil cumulative will be for

a range of economic limit water cuts from 76% to 99% using the log

WOR, log oil cut, log water cut and linear water cut methods.

Figure 6-4 shows the results. The effects of the water cut economic limit can be

dramatic, especially where water cut economic limits exceed 90%. It is important that

the engineer ensure that water cuts above 90% can be achieved both economically and

mechanically.

WATER RATIO METHODS RESERVES VARIATION WITH CHANGING ECONOMIC LIMIT

300

250

200
Reserves - MMBO

fo or fw
log fo
150
log fw
log WOR

100

50

0
0.75 0.8 0.85 0.9 0.95 1
Economic Water Cut

Figure 6-4 - Effect of Economic Limits on Water Ratio Methods

One of the concerns with the log WOR and log oil cut methods is that these plots do not

have mathematical limits. The water cut methods can not exceed the mathematical limit

Chevron Confidential Use Only Page 113 of 244

 Version 2008.01

of 100%. For WOR and log oil cut, the log cycles allow the curve to continue

indefinitely. The practical engineer realizes that the physical limit in waterflooding is the

residual oil saturation. Regardless of how many hydrocarbon pore volumes of water are

cycled through the reservoir, the Sor limits the theoretical maximum recovery unless some

miscible or chemical process is used. This means that at some point the log WOR curve

must turn up and the log oil cut must turn down.

Before the widespread use of personal computers, this slope change used to be handled

by using a ships curve to forecast the log oil cut curve. Some engineers also used to

calculate reserves using the best fit for the log oil cut method and the linear oil cut

method and then averaging. Figure 6-4 gives an idea of what the averaging would

calculate.

The 2001 SPE Reserves Manual11 discusses the turn down oil cut phenomenon stating

that light oil reservoirs will begin turning down at higher oil cuts than heavy oil

reservoirs. Some of this is due to the mobility ratio driven by the higher oil viscosity in

heavy reservoirs. It also is affected by the reservoir geometry of permeability and

whether it is coarsening upward or downward. Dake in his updated 2001 reservoir

engineering manual provides considerable discussion on the displacement of oil with

water. It is insightful and worthwhile reading.

When faced with making a decision on when a ratio curve will turn up or down, the best

option is to try and find an analog reservoir that has been produced to a high water cut.

One can also observe individual wells in the field which have a high water cut. Figure

Chevron Confidential Use Only Page 114 of 244

 Version 2008.01

6-5 shows an example WOR analogy set. Note that the cumulative production is

replaced with recovery to make the abscissa dimensionless. The RAC also provides

some “Best Practices” guidelines which are listed later in this chapter.

Another guideline that is commonly given for ratio methods is that they should not be

used until the WOR reaches a value of 1.0 or water cut of 50%. The SPE reserves

manual suggests this. Chevron’s experience and guidelines are not to use these methods

until a WOR of 2.0 or water cut of 67% is reached. Figure 6-5 shows a distinct

difference for the Indonesia field extrapolations between a WOR of 1.0 and 2.0.

Another consideration is whether to perform reserves by individual wells or on a field or

field-reservoir basis. Cobb in his waterflooding course strongly advocates the using a by

well analysis. This subject is covered in depth in other sections in this manual.

Chevron Confidential Use Only Page 115 of 244

 Version 2008.01

WOR History vs. RF (%OOIP)

100

10
WOR

1
BALAM SO BANGKO BEKASAP
KOTABATAK PETANI PETAPAHAN
Minas

0.1
0. 0%

5. 0%

%
10 .0

15 .0

20 .0

25 .0

30 .0

35 .0

40 .0

45 .0

50 .0

55 .0
RF (% OOIP)

Figure 6-5 - WOR Analogy Comparison for Indonesia

6.2 Water Oil Ratio Curves

The most common water ratio method used throughout the industry by oil producers and

consulting companies is the log WOR method. The Chevron Reserves Advisory

Committee (RAC), after considerable study, standardized on the log WOR versus Np

method in 2006. This standard was instituted because of a wide variation in water ratio

reserves estimate methods and confusion that existed throughout the company. The RAC

guidelines are listed later in this chapter. Because of this standardization, only the log

WOR method will be discussed further in this manual.

Chevron Confidential Use Only Page 116 of 244

 Version 2008.01

The Water Oil Ratio (WOR) is calculated using Equation 6-1. It is simply the water

production divided by oil production in surface units. Some engineers prefer to use

reservoir units as shown by Equation 6-2. The plot would also use reservoir barrels (Np

x Bo) for the cumulative production. Either is acceptable.

BWPD BFPD − BOPD

WOR = =
BOPD BOPD
Equation 6-1 - Water Oil Ratio

BWPD × Bw
WORRe servoir =
BOPD × Bo

Equation 6-2 - WOR in Reservoir Units

Equation 6-3 shows an easy way of calculating oil rate when the BFPD and WOR are

known. This equation can be derived by rearranging Equation 6-1.

BFPD
BOPD =
(1 + WOR)

Equation 6-3 - BOPD from WOR

The reserves using the log WOR method are calculated by determining the equation of

the best fit line using the familiar y = mx + b format. The slope is calculated using

Equation 6-4. The y intercept is then calculated using Equation 6-5 which substitutes one

of the WOR Np data pairs.

Chevron Confidential Use Only Page 117 of 244

 Version 2008.01

⎛ WOR 2 ⎞
ln⎜⎜ ⎟
(ln( WOR 2 ) − ln( WOR 1 )) ⎝ WOR 1 ⎟⎠
Slope = =
Np 2 − Np1 Np 2 − Np1

Equation 6-4 - Slope of WOR Curve

Yint ercept = ln( WOR 1 ) − slope × Np1

Equation 6-5 - WOR Curve Y Intercept

The estimated ultimate recovery (EUR) at the abandonment WOR, or Np at any WOR is

calculated using Equation 6-6. Once the EUR has been calculated, the reserves can be

calculated by subtracting the reserves date cumulative oil from the EUR using Equation

6-7.

ln( WOR EL ) − Yint ercept

EUR =
slope

Equation 6-6 - EUR from WOR

Reserves = EUR − Np AsOf

Equation 6-7 - Reserves from WOR

A simpler form of the reserves equation is shown by Equation 6-8. It is derived by

equating the slopes between a historical WOR and Np, the reserves date WOR and Np

and the economic limit WOR and Np as shown in Figure 6-6.

Chevron Confidential Use Only Page 118 of 244

 Version 2008.01

⎛ WOR EL ⎞
ln⎜⎜ ⎟
⎝ WOR AsOf ⎟⎠
Reserves = (Np AsOf − Np Historical ) ×
⎛ WOR AsOf ⎞
ln⎜⎜ ⎟⎟
⎝ WOR Historical ⎠

Equation 6-8 - Simple Form of WOR Reserves Equation

(NpEUR,WOREL)

(NpAsOf,WORsOf)

Reserves
(NpHistorical,WORHistorical)

Figure 6-6 - Equal Slopes Method

Figure 6-7 shows an example of the WOR method for a strong water drive field located

in the Big Horn Basin of Wyoming. The best fit straight line extrapolation forecasts the

cumulative production to a final WOR of 90. The field will have a final cumulative

production of slightly larger than 21 million barrels of oil.

Chevron Confidential Use Only Page 119 of 244

 Version 2008.01

Figure 6-7 - WOR Method Example

Figure 6-8 - Big Horn Basin Field Rate Time

Chevron Confidential Use Only Page 120 of 244

 Version 2008.01

The Big Horn Basin example has a cumulative production of 17.8 MMSTBO with a

WOR of 50.2. In 1992 the field had a WOR of 31.6 and a cumulative of 15.1 MMSTBO.

Calculate the remaining reserves assuming an economic limit WOR of 90.

⎛ 90 ⎞
ln⎜ ⎟
⎝ 50.2 ⎠
Reserves = (17.8 − 15.1) × = 3.4 MMSTBO
⎛ 50.2 ⎞
ln⎜ ⎟
⎝ 31.6 ⎠

6.3 Rate Time Forecasts from Water Oil Ratio Curves

The WOR ratio method calculates the remaining reserves in a waterflood or water drive

reservoir but does not generate the oil rate versus time forecast. The rate time forecasts

are necessary for economic calculations. They are also required for reserves on fields

with a constraint such as a contract life or facility life. It is important that the engineer

ensures that the WOR economic limit will be reached before any constraint is reached. If

not, the reserves must be adjusted downward accordingly. The rate time forecast is also a

reasonableness check on the WOR reserves. The engineer should ask the question “Does

the forecast look reasonable based on the historical rate versus time or rate versus

cumulative?”

Surprisingly most petroleum engineering texts have not addressed the issue of how to

generate a rate time forecast from a WOR evaluation. Two simple methods will be

Chevron Confidential Use Only Page 121 of 244

 Version 2008.01

presented here. The first method is a simple empirical mathematical solution. The

second uses a “spreadsheet” time step concept.

The Empirical Method - The first method makes a simple assumption that the reservoir

voidage is in balance, the reservoir pressure is not increasing or decreasing and that the

barrels of produced fluid will remain constant for the length of the forecast. This is a

reasonable engineering assumption for many waterfloods or water drive reserves that

have been in operation for an extended length of time. Of course every engineer realizes

that fluid rate will change with changing well count, increased injection or changing

mobility ratio. Reservoirs with a mobility ratio less than 1.0 may show decreased water

production since the water viscosity is higher than the oil viscosity. Reservoirs with a

mobility ratio greater than 1.0 may show increased water production as the more viscous

oil is depleted. A pragmatic analysis is required when making the assumption that fluid

will remain constant. Note that BFPD is used instead of BWIPD which may be higher

because all of the injection may not be effective injection. Most waterfloods have some

losses to the aquifer or to out-of-zone injection.

Once the BFPD rate is determined, the oil rate at any WOR can easily be calculated using

Equation 6-3 for any WOR.

BFPD
BOPD =
(1 + WOR)

The cumulative production Np can also be determined at any WOR using Equation 6-6.

Chevron Confidential Use Only Page 122 of 244

 Version 2008.01

ln( WOR ) − Yint ercept

Np =
slope

By substituting WOR with Equation 6-1, Equation 6-6 can be modified into Equation

6-9.

⎛ BFPD − BOPD ⎞
ln⎜ ⎟ − Yintercept
⎝ BOPD ⎠
Np =
slope

Equation 6-9

Equation 6-10 is generated by solving Equation 6-9 for BOPD.

BFPD
BOPD = slope×Np + Yintercept
e +1

Equation 6-10 - BOPD from WOR

To generate time from BOPD and Np, one can to integrate the reciprocal of Equation

6-10 over a range of cumulative production. For the time to the economic limit WOR,

the integration is shown by Equation 6-11.

Np EL

∫
e slope× Np + Yintercept + 1
n

Time = × dNp n
BFPD
Np AsOf

Equation 6-11 - Time to WOR Economic Limit

Chevron Confidential Use Only Page 123 of 244

 Version 2008.01

Since the rates are daily, the time units are days. After performing the mathematics, the

solution for Equation 6-11 is given by Equation 6-12.

× slope + Yintercept
e Np el
+ Np el × slope − e Np ×slope + Yintercept − Np AsOf × slope
AsOf

Time EL =
slope × BFPD

Equation 6-12 - Time to Economic Limit WOR

Note that the exponential values conveniently become the WOREL and WORAsOf to

generate a simplified form of Equation 6-12 and Equation 6-14.

(WOR EL − WOR AsOf ) + ( Np el − Np AsOf ) × slope

Time EL =
slope × BFPD

Equation 6-13 - Time to EL WOR Simplified

By substituting Equation 6-4 for the slope, Equation 6-13 can be further simplified into
an easier to use form without having to calculate a slope as shown by Equation 6-14.

⎛ WOR EL ⎞
(WOR EL − WOR AsOf ) + ln⎜⎜ ⎟⎟
⎝ WOR AsOf ⎠
Time EL =
⎛ WOR EL ⎞
ln⎜⎜ ⎟⎟
⎝ WOR AsOf ⎠
× BFPD
(Np EL − Np AsOf )

Equation 6-14 - Time to EL WOR Further Simplified

Note that the cumulative production should have units of STBO. As one can see by the

nature of the equation, if the BFPD being constant is not assumed, the empirical

equations will quickly become unwieldy and difficult to use.

Chevron Confidential Use Only Page 124 of 244

 Version 2008.01

The Time-Step Method - The second method makes a simple use of the power of

modern computers, especially spreadsheets. This method divides the WOR range and Np

range into a number of small increments. At each WOR, the oil rate can be calculated

using Equation 6-3 and the cumulative production can be calculated using Equation 6-6.

BFPD
BOPD =
(1 + WOR)

ln( WOR ) − Yint ercept

Np =
slope

The volume of production can be determined by subtracting the Npn for WORn and Npn+1

for WORn+1. The average rate during can be calculated by averaging BOPDn and

BOPDn+1. The length of the time step is calculated using Equation 6-15.

Np n+1 − Np n
dT =
BOPD n+1 + BOPD n
2

Equation 6-15 - Time Step Equation

The engineer can calculate the oil rate at a large number of WORs and the time steps

summed to get the remaining life. A relationship between BFPD and Np can also be

entered into the calculations of the spreadsheet if the fluid rate is expected to change over

the remaining life.

Chevron Confidential Use Only Page 125 of 244

 Version 2008.01

Time to Economic Limit Example – A large field is producing a constant 50,000 BFPD

with a WOR of 5. The latest cumulative production is 35 MMSTBO. The WOR method

estimates that the field will be abandoned when the WOR is 10 and the EUR is 72

MMSTBO. Calculate the remaining life assuming the fluid production remains constant.

Use both the Empirical Method and the Time Step Method using 10 time steps.

Empirical Method:

⎛ 10 ⎞
(10 − 5) + ln⎜ ⎟
Time EL = ⎝5⎠ = 6078 days = 16.7 years
⎛ 10 ⎞
ln⎜ ⎟
⎝5⎠ × 50,000
(72,000,000 − 35,000,000)

Time Step Method:

WOR BOPD Np Δ Np Average BOPD Δ Days Σ Days

5.00 8,333 35,000,000 0 0 0
5.50 7,692 40,087,630 5,087,630 8,013 635 635
6.00 7,143 44,732,273 4,644,643 7,418 626 1,261
6.50 6,667 49,004,930 4,272,657 6,905 619 1,880
7.00 6,250 52,960,793 3,955,863 6,458 613 2,492
7.50 5,882 56,643,613 3,682,820 6,066 607 3,100
8.00 5,556 60,088,660 3,445,048 5,719 602 3,702
8.50 5,263 63,324,786 3,236,125 5,409 598 4,300
9.00 5,000 66,375,886 3,051,100 5,132 595 4,895
9.50 4,762 69,261,978 2,886,093 4,881 591 5,486
10.00 4,545 72,000,000 2,738,022 4,654 588 6,074

As demonstrated, both methods produce an answer with identical engineering accuracy.

Chevron Confidential Use Only Page 126 of 244

 Version 2008.01

WOR Rate Time Forecast

10,000

Hyperbolic Best Fit

b = 0.85
di = 0.047
BOPD

1,000
0 2 4 6 8 10 12 14 16 18
Years

Figure 6-9 - Plot of Rate Time WOR Forecast Example

Figure 6-9 shows the semi-log rate time plot for the previous example. Note that the

solution produces a forecast with hyperbolic decline characteristics. This is generally the

case where the fluid rate remains constant. The initial decline is quite low, around 4 to

5% and the hyperbolic constant is approximately 0.85 which is not unusual. The decline

is directly affected by the reservoir processing rate. If the BFPD was doubled to 100,000

BFPD, the initial decline would have increased to slightly over 9%. For both processing

rates, the hyperbolic constant remains constant at approximately 0.85. Although this is

not initially intuitive, it can be explained because all the rates will be shifted with the

change in total fluid production while the cumulatives remain the same.

The factors most affecting the hyperbolic b factor are the WORAsOf and WOREL. Figure

6-10 shows how the initial decline and hyperbolic constant b varies with a changing

WOREL for the previous example. The BFPD, NpAsOf, NpEL and WORAsOf were kept

Chevron Confidential Use Only Page 127 of 244

 Version 2008.01

constant. Only the WOREL was varied. Note how both ai and b increased as the WOREL

was increased.

WOR Hyperbolic Constants and Initial Decline

Assuming AsOf WOR = 5.0

0.94 21%

b Factor
Initial Decline
0.92 18%

0.9 15%
Hyperbolic b Factor

Initial Decline - %
0.88 12%

0.86 9%

0.84 6%

0.82 3%

0.8 0%
0 10 20 30 40 50 60
WOR Economic Limit

Figure 6-10 - Hyperbolic Constants Variation

The process of determining the hyperbolic constants were discussed in Section 5.2. The

“Two Rate Method”, “Three Point Method” and “Initial Rate and Decline Method”

were both presented. All are acceptable to solve the hyperbolic constants for the WOR

solution. The solution sometimes works best when one first solves for the ai before

solving for b. Equation 6-3 can be used to calculate both the initial and economic limit

oil rates. Equation 6-6 gives Np for any WOR. Equation 6-14 calculates the remaining

life. The only variable that is unknown is the initial decline Di or nominal decline ai.

Chevron Confidential Use Only Page 128 of 244

 Version 2008.01

Conveniently, the value of ai can be calculated directly using the WOR data. To solve for

ai, the engineer needs to solve for dq/dNp where WOR is equal to WORAsOf . This can be

seen from the reserves Equation 6-16.

365 × dq
dNp =
ai

Equation 6-16

By rearranging, the equation and then solving for ai, the equation becomes.

365 × dq
ai =
dNp

Equation 6-17

The derivative of dq with respect to WOR is given by Equation 6-18.

d ⎛ BFPD ⎞ BFPD
dq = ⎜ ⎟=
⎜ 1 + WOR AsOf ⎟ (1 + WOR 2
dWOR AsOf ⎝ ⎠ AsOf )

Equation 6-18

The derivative of dNp with respect to WOR is given by Equation 6-19.

d ⎛ Y int ercept − Ln(WOR AsOf ) ⎞ 1

dNp = ⎜ ⎟=
⎜ ⎟ WOR
dWOR AsOf ⎝ slope ⎠ AsOf × slope

Equation 6-19

Chevron Confidential Use Only Page 129 of 244

 Version 2008.01

Substituting Equation 6-18, Equation 6-19 and Equation 6-4 into Equation 6-17 produces

Equation 6-20 which calculates ai directly.

⎛ WOR EL ⎞
ln⎜⎜ ⎟
BFPD ⎝ WOR AsOf ⎟⎠
a i = 365 × × WOR AsOf ×
(1 + WOR AsOf )2 (NpEL − Np AsOf )

Equation 6-20 - Solution for ai from WOR Forecast

For our previous example, the value of ai would be calculated.

⎛ 10 ⎞
ln⎜ ⎟
50,000 ⎝5⎠
ai = 365 × ×5× = 0.04748
(1 + 5)2 (72,000,000 − 35,000,000)

The initial exponential decline can then be calculated using Equation 5-4.

D i = 1 − e −0.04748 = 0.046

With variable ai known, the value of hyperbolic life can be calculated over a range of b factors

using Equation 5-15 and plotted against life calculated using Equation 6-14. Figure 6-11 shows

the life in days calculated as the b factor is varied. The intersection of the b factor life against the

previously calculated life of 6,078 days occurs at a hyperbolic b factor of 0.841.

Chevron Confidential Use Only Page 130 of 244

 Version 2008.01

The value for ai in the above example could have also been by calculating the change in rate

between a WOR = 5.0 and a WOR = 5.0001 and the cumulative production between those two

WORs. The oil rates at WOR5.0 and WOR5.0001 are 8333.333 BOPD and 8333.1945 BOPD

respectively. The production between WOR5.0 and WOR5.0001 is 1067.58 STBO.

(8,333.3333 − 8,333.1945)
ai = × 365 = 0.04748
1,067.58

Example - Solution for b Factor

6100

time
6095
life-days

6090

6085

6080
Life - Days

6075

6070

6065

6060

6055

6050
0.82 0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86
b Factor

Figure 6-11 - Example Solution for b Factor Matching Life

A similar plot comparing the reserves versus hyperbolic b factor is shown by Figure 6-12. The b

factor solutions for the life and the reserves will not be exactly equal since the production forecast

generated by the WOR solution is only approximated by a hyperbolic decline. Both will be

within acceptable reservoir engineering accuracy.

Chevron Confidential Use Only Page 131 of 244

 Version 2008.01

Example - Solution for b Factor Matching Reserves

37.30

37.25
Hyperbolic Reserves WOR Reserves

37.20

37.15

37.10
Reserves - MMSTBO .

37.05

37.00

36.95

36.90

36.85

36.80

36.75
0.82 0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86
b Factor

Figure 6-12 - Example b Factor Solving for Reserves

6.4 RAC Guidance for DCA in Waterfloods and Water Drive

Reservoirs

In 2006 the RAC formed a technical committee from its membership to study the use of ratio

methods for calculating reserves on waterflood and water drive reservoirs. The main decision

reached was that Chevron would standardize on the log WOR versus cumulative oil production

plot for ratio reserves. The RAC also requires that log of oil rate versus time and linear oil rate

versus cumulative oil production plots be made when the WOR method is used.

The following guidelines are issued by the RAC for use when making estimates on waterflood

and water drive reserves:

Chevron Confidential Use Only Page 132 of 244

 Version 2008.01

1. Plots of log of WOR versus cumulative oil production, log of oil rate versus time and oil

rate versus cumulative oil production should be used as primary methods for determining

reserves in water drive and waterflooded fields. The use of one single method is not

recommended as this approach can result in erroneous reserves calculations in many

cases.

2. A forecast of total fluid production (honoring fluid in and out targets) should be made

using the extrapolated behavior of oil rate and WOR performance. This forecast should

be checked to ensure that total fluid production and injection facilities capacity are not

exceeded or are reasonable. In addition, when using the log of WOR versus cumulative

oil production or rate versus cumulative oil production, an estimate of reasonable time

frame for forecasting must be considered using the rate versus time plot.

3. In fields under active development for drilling or workovers, the QRE should ensure the

established line to be used for extrapolation in all plots is not being affected by the

activity level. It should be noted that all of the above extrapolation techniques assume the

future conditions for operating the field will stay the same as the past (used as a basis for

extrapolation).

4. Extrapolation techniques should also be used cautiously when early life well/reservoir

drainage can differ dramatically with late life drainage. Examples of this are fractured

reservoirs and coning in bottom water drive reservoirs.

5. An estimate of proved reserves should be calculated on an extrapolation of the WOR plot

to an endpoint of 90% water cut. Extrapolation (for proved reserves) beyond a 90% water

cut will need to be justified by appropriate analogs to ensure over booking does not occur

because of late-life changes. Note that in higher mobility ratio waterfloods, significant

amounts of oil are produced at higher WOR values.

Chevron Confidential Use Only Page 133 of 244

 Version 2008.01

6. A significant history of WOR greater than 2 will need to exist before using the log of

WOR versus cumulative oil production plot for proved reserves estimation. Too much

scatter exists at low WORs to establish a trend.

Chevron Confidential Use Only Page 134 of 244

 Version 2008.01

7 Decline Curve Data and Assumptions

Exponential declines and most of the other declines require a linear prediction to

calculate the reserves. For the exponential decline, the natural logarithm of the

production rate plotted on the y axis versus linear time plots a straight line. The main

exceptions to straight lines are hyperbolic and harmonic declines which curve.

Because of the convenience of modern computers, some engineers have lost the ability to

perform best fit regression through a data set. These calculations can be useful,

especially if some data points are invalid or violate boundary conditions and need to be

eliminated from the best fit. Some, but not all, DCA software packages allow data points

to be easily culled.

7.1 Data Regression - Least Squares Best Fit

The most common regression technique is called the Least Squares Best Fit (LSBF). It

gets its name from the optimization technique that minimizes the summation of the

squared differences between the data point y axis value and the predicted y axis value.

The Least Squares formula used is shown by Equation 7-1. The squared function is used

to ensure that the difference is always positive and that large negative values are not

Chevron Confidential Use Only Page 135 of 244

 Version 2008.01

being offset by large positive values. The best fit regression is achieved when the sum of

the squared differences is minimized.

Sum of Squared Differences = ∑1 ( y predicted − y actual ) 2

n

Equation 7-1 - Sum of Squared Differences

Let’s look at an example. An oil well is has the following annual production for 10

years.

Year BOPD Ln (BOPD)

0 1000 6.907755
1 867 6.765039
2 752 6.622323
3 652 6.479606
4 565 6.33689
5 490 6.194174
6 425 6.051457
7 368 5.908741
8 319 5.766025
9 277 5.623309
10 240 5.480592

The initial production rate is 1000 BOPD. Assuming that the initial production rate is

correct, calculate the Sum of Squared Differences value assuming a 10% annual

exponential decline. Also construct a plot showing the Sum of Squared Differences for

exponential declines varying from 10% to 16% and make an estimate of the actual

exponential decline.

Chevron Confidential Use Only Page 136 of 244

 Version 2008.01

Year BOPDactual Ln(BOPDactual) BOPD10% Ln(BOPD10%) Square Value

0 1000 6.907755 1000 6.907755 0.0000

1 867 6.765039 900 6.802395 0.0014
2 752 6.622323 810 6.697034 0.0056
3 652 6.479606 729 6.591674 0.0126
4 565 6.33689 656 6.486313 0.0223
5 490 6.194174 590 6.380953 0.0349
6 425 6.051457 531 6.275592 0.0502
7 368 5.908741 478 6.170232 0.0684
8 319 5.766025 430 6.064871 0.0893
9 277 5.623309 387 5.959511 0.1130
10 240 5.480592 349 5.85415 0.1395
Total 0.5373
Table 7-1 - Sum of Squared Differences Calculation

Decline versus Sum of Squares

0.6

0.5

0.4
Sum of Squares

0.3

0.2

0.1

0
10.0% 10.5% 11.0% 11.5% 12.0% 12.5% 13.0% 13.5% 14.0% 14.5% 15.0% 15.5% 16.0%
Exponential Decline - %

Figure 7-1 - Sum of Square Minimum

Table 7-1 shows the calculation for the Sum of Squared Differences between the actual

decline and a 10% declines. Note that the summation is approximately 0.53. Figure 7-1

shows the Sum of Squared Differences for declines between 10% and 16%. On this

Chevron Confidential Use Only Page 137 of 244

 Version 2008.01

graph notice that at a 10% decline, the graph also matches the value 0.53. The minimum

Sum of Squared Differences occurs between a decline of 13% and 13.5% indicating that

the best fit decline is approximately 13%.

This example could have also been solved using the classical equations for LSBF. The

regression equations for LSBF contain two equations. Equation 7-2 calculates the slope.

Equation 7-3 calculates the y axis intercept.

1
∑ xy − n (∑ x )(∑ y)
Slope =
1
∑x 2
−
n
(∑ x)2

Equation 7-2 - Least Square Best Fit Slope

1 1
Y int =
n
∑ y − Slope × ∑ x
n
Equation 7-3 - Least Squares Best Fit Y Intercept

For our previous example, calculate the slope, y intercept and decline.

Count X Y X2 XY

1 0 6.907755 0 0
2 1 6.765039 1 6.765039
3 2 6.622323 4 13.24465
4 3 6.479606 9 19.43882
5 4 6.33689 16 25.34756
6 5 6.194174 25 30.97087
7 6 6.051457 36 36.30874
8 7 5.908741 49 41.36119
9 8 5.766025 64 46.1282
10 9 5.623309 81 50.60978
11 10 5.480592 100 54.80592
Total 55 68.13591 385 324.9808

Chevron Confidential Use Only Page 138 of 244

 Version 2008.01

1 1
∑ xy − n (∑ x)(∑ y ) 324.9808 −
11
(55 × 68.13591)
Slope = = = −0.1427
1 1
∑ x − n (∑ x)2
2
385 − (55)
11
2

1 1 1 1
Y int =
n
∑ y − Slope × ∑ x = × 68.13591 − ( −0.1427) × × 55 = 6.9077
n 11 11

Y int ercept = e 6.9077 = 1000 BOPD

Decline = 1 − e −0.1427 = 0.133

Let’s also confirm this 13.3% decline with an Excel spreadsheet trend by creating Figure

7-2. Note the y = mx + b equation best fit trend matches the slope and y intercept

calculated above.

Chevron Confidential Use Only Page 139 of 244

 Version 2008.01

Least Squares Best Fit Decline Example

y = -0.1427x + 6.9078
5
Ln(BOPD) .

0
0 1 2 3 4 5 6 7 8 9 10
Years

Figure 7-2 - Least Squares Best Fit Excel Example

7.2 Boundary Conditions

Decline curve analysis using the Arp’s Equations is based on the fundamental assumption

that past operating conditions will remain unchanged. The following is a list of inherent

boundary condition assumptions implied when performing rate-time decline curve

analysis:

• The well or field is produced at or near capacity.

• The well or field drainage radius remains constant.

Chevron Confidential Use Only Page 140 of 244

 Version 2008.01

• The well or field is produced at a constant bottom-hole pressure.

There are many conditions in hydrocarbon production operations where these

assumptions are not valid. They include:

• prorated production

• curtailed production

• workovers

• drilling

• equipment changes

• equipment downtime

• fluid or gas injection changes

• scale formation

• paraffin formation

• fluid handling capacities

• operating practices

• personnel changes

A key QRE responsibility when performing decline curve analysis is to ensure that the

production periods used for analysis are indeed stable.

Figure 7-3 shows a good example where changing boundary conditions can cause a large

negative reserves revision. It shows a gas well with 40 md permeability, 1200 acres

drainage and a 5000 psia initial reservoir pressure. The gas well sells into a line requiring

Chevron Confidential Use Only Page 141 of 244

 Version 2008.01

a minimum of 1250 psia FBHP. However, because of sand production concerns, the well

is operated with no more than a 1000 psia sand face draw down. During the first four

years of production the well is choked because of the 1000 psia sand face drawdown

restriction. After year four, the tubing pressure is riding the 1250 psia sales line pressure

with the well no longer choked. As can be seen, a decline curve analysis performed on

the first four year of data would yield an effective exponential decline of approximately

15%. Once the well reached sales line pressure, the decline would increase to over 40%.

Any forecast which ignored the sales line restriction would cause a large negative

downward revision in reserves. For this reason, a best practice is to include tubing

pressure on all gas well DCA.

Gas Well with Changing Boundary Conditions

100 6000

5000

4000
MMCFPD

PSI
10 3000

2000

1000

1 0
0 1 2 3 4 5 6 7 8 9 10
Years

Rate Reservoir Pressure Flowing BHP

Figure 7-3 - Decline Curve with Changing Boundary Conditions

Chevron Confidential Use Only Page 142 of 244

 Version 2008.01

7.3 Data Accuracy and Uncertainty

Another practical assumption is that the production and pressure data available are

accurate. Accurate oil production is usually available where it is measured for sale or

collected in a central production gathering facility. It becomes much less accurate where

well tests are used to allocate production back to each well and sometimes to commingled

reservoirs in a well. In many fields the well testing frequency and accuracy are not

sufficient to accurately allocate monthly production on a well basis. It is not uncommon

for the sum total of the well tests into a gathering facility to differ considerably from the

measured sales production. Ideally, the allocation factor should be between 0.95 and

1.05. It is also not uncommon for well test frequency to be very low in some operations

where well testing is difficult, expensive or requires other production be curtailed. It is a

QRE responsibility to understand the limitation of the available production data when

using that data to calculate reserves.

Data on gas wells is usually more accurate, mainly because gas flow rates are easier and

cheaper to measure when orifice meters are used. Most gas wells will have their own

metering equipment including surface pressure and temperature recording. Also, many

regulatory agencies require annual gas well deliverability testing.

Chevron Confidential Use Only Page 143 of 244

 Version 2008.01

All data produced by systems has natural variability. When some data points vary

significantly from the established trend because of major operational events, the least

squares best fit can change significantly. Good DCA software should allow the engineer

to eliminate data points that obviously are caused by something other than natural

decline. Many times these bad data points are caused by accounting or allocation

mistakes. Figure 7-4 shows a best fit trend through a data set with reasonably tight data.

The exponential best fit through the data has a 27% exponential decline. Figure 7-5

shows the same data set except that the numbers 2 and 17 data points are bad. These two

data points increased the exponential decline to 32% which would have significantly

reduced the reserves forecast.

Semi-log Rate Time Data with Accurate Data

1000

-0.3125x
y = 408.66e
BOPD

100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Year

Figure 7-4 - Best Fit Example with All Good Data

Chevron Confidential Use Only Page 144 of 244

 Version 2008.01

Semi-log Rate Time with 2 Bad Data Points

1000

y = 445.64e-0.4156x
BOPD

100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Years

Figure 7-5 - Best Fit Data with 2 Bad Data Points

7.4 Well Groupings versus Individual Well DCA

One decision that the QRE needs to make is at what level decline curve analysis should

be performed. Although it is still more labor intensive, modern DCA software tools

allow easy analysis and summation all the way down to the individual well level.

Individual well analysis, however, is not always the best choice especially if well testing

frequency and accuracy is causing the allocated well production to be “noisy” thus

making production forecast difficult and less certain. Some reservoirs also have

Chevron Confidential Use Only Page 145 of 244

 Version 2008.01

changing flow patterns where the drainage area and flow patterns to each well vary

substantially with time causing “noisy” data.

The production history on a field basis is generally less “noisy” because any allocation

errors are corrected by the sales facility. However, decline curve analysis on a field level

also has short comings. On many large fields different geographic areas will produce

with significantly different characteristics. Many of the differences between areas are

controlled by the geology of the field. In one area, fractures or vugular porosity might

cause excessive water break through and production. Other areas might be at a different

stage of depletion because that area was developed later or have lower permeability or be

experiencing edge water encroachment. Many times the producing characteristics of

different areas mask each other. For example, the northern area of a large field may have

a high decline while the southern area is responding to recent injection changes. The

composite production curve will not adequately allow the QRE to forecast the reserves.

7.5 Equipment Operating Conditions

Production data will also appear “noisy” if the producing environment does not keep the

working bottom-hole pressure (WBHP) constant. This is particularly the case when

surface flow lines vary in pressure because of equipment downtime, temperature changes,

fluid shipping schedules or compressor downtime. Changing WBHP also occurs when

rod pumped wells do not have adequate downhole gas separation or have a history of

Chevron Confidential Use Only Page 146 of 244

 Version 2008.01

frequent pump changes. Gas lift environments, especially intermittent gas lift operations,

have widely varying producing bottom-hole pressures.

Figure 7-6 shows the effects of changing working bottom-hole pressure on a well with 30

feet of net pay, a permeability of 5 md and an oil viscosity of 0.6 cp. The plot was

created using a well simulator. The blue line shows the flow rate when bottom-hole

pressure is held constant at 300 psia. The red line shows the flow rate when WBHP is

randomly varied over a range of ± 150 psia. The variation for the WBHP used a normal

distribution with a standard deviation of 90 psia. Figure 7-7 shows the pressures used by

the simulator. Figure 7-8 shows the same plot as Figure 7-6 but with semi-log

coordinates. Note the visual effects of the semi-log scale, especially in the later years.

The logarithmic scale gives the effect that the data variation is larger in the later time

period when in reality variation remained constant.

Chevron Confidential Use Only Page 147 of 244

 Version 2008.01

Example Showing Effects of Varying WBHP

800
300 PSI WBHP
300 PSI WBHP ± 150 psi (σ 90)
700

600

500
BOPD

400

300

200

100

0
2000 2001 2002 2004 2005 2006 2008 2009 2010
Year

Figure 7-6 - Effects of Varying WBHP

Chevron Confidential Use Only Page 148 of 244

 Version 2008.01

Example Showing Effects of Varying WBHP

500
300 PSI WBHP
300 PSI WBHP ± 150 psi (σ 90)
450

400

350
WBHP - psi

300

250

200

150

100
2000 2001 2002 2004 2005 2006 2008 2009 2010

Figure 7-7 - WBHP Variation

Example Showing Effects of Varying WBHP

1000
300 PSI WBHP
300 PSI WBHP ± 150 psi (σ 90)
BOPD

100
2000 2001 2002 2004 2005 2006 2008 2009 2010
Year

Figure 7-8 - Semi-log Plot of Varying WBHP

Chevron Confidential Use Only Page 149 of 244

 Version 2008.01

7.6 DCA Mapping

In many fields several large producers will dominate the field decline. This is especially

true with water production where many times a couple of producers with high volume lift

equipment produce over 75% of the field’s water production. When these wells

eventually become marginally economic and are shut-in, the field production curve

changes dramatically.

Sometimes the best practice is to group the field into several different areas and perform

decline curve analysis by different areas. These areas can be determined using maps that

show producing rates or changes in production (percentage or volume). Several of

today’s DCA software tools allow these maps to be made with ease. Figure 7-9 shows an

example of a decline curve contour map. It shows annual exponential declines across a

field. Notice that the declines on the northern edge of the field are large while several

areas in the center have varying degrees of decline and are lower.

Another technique that has proven very useful is “Location Plots.” These plots were first

used in the REXDCA12 decline curve analysis software. The technique uses a montage

format which plots individual well declines at their latitude and longitude position on a

map. Figure 7-10 shows an example showing semi-log rate versus time for the producing

Chevron Confidential Use Only Page 150 of 244

 Version 2008.01

wells and semi-log injection rate and tubing pressure for the injection wells. They are

very useful for monitoring individual well declines, especially in fields responding to

waterflood. In Halliburton’s DSS package, the option is call “Chart Montage Format.”

Figure 7-9 - Decline Contour Map using DSSTM

Chevron Confidential Use Only Page 151 of 244

 Version 2008.01

Figure 7-10 - Location Plots

Chevron Confidential Use Only Page 152 of 244

 Version 2008.01

8 Infill and Extension Well Drilling Effects

Field level or reservoir level decline curve analysis is difficult when fields are under

continuous drilling programs. Over-booking reserves frequently occurs where an active

workover or drilling development program is underway. The development program

causes the field to have an artificially low decline or sometimes an incline. This causes

reserves to be over-predicted and results in a negative revision once the drilling program

is completed and the field returns to its normal higher base decline. This effect was listed

as a common major reserves error in the 2004 SPE paper “Oil and Gas Reserves

Estimates: Recurring Mistakes and Errors” by Ron Harrell, Ryder Scott.32

8.1 Drilling Wedge Method for Rate Time or Rate Cumulative

When a continuous drilling situation exists, it is recommended that a drilling wedge

(sometimes called a Vintage Well Analysis) be performed. This technique compares the

decline from a group of older wells with the current total field or reservoir. The

technique constructs semi-log rate time curves for the base wells that existed in a field on

key dates. For example, Base 2003 wells are all of the wells that exist in a reservoir prior

to January 1, 2003. When this base decline is compared to the total field production, the

base decline of the field can be calculated. Assuming negligible well drainage

Chevron Confidential Use Only Page 153 of 244

 Version 2008.01

interference, the base decline is a good estimate of what the field decline will be when

drilling activities cease or reduce. A similar plot using rate versus cumulative production

could also be constructed. Both plots are especially useful in determining the

effectiveness of drilling programs by providing an estimate of how much reserves are

being increased and how much production is only being accelerated.

Figure 8-1 shows an example where a long term continuous drilling program in a

waterflood has arrested the decline. However, when an analysis plot is made for the wells

that existed in the field at the end of years 2000, 2002, 2003, 2004 and 2005, one can see

that the base decline is quite substantial. This base decline should be used to book the

Proved Developed Producing Reserves. Reserves recovered above the base decline

should be classified as Proved Undeveloped or Proved Developed Non-producing

Reserves provided the business certainty of the continued drilling exists. The BOPD

versus cumulative oil is shown in Figure 8-2. It demonstrates the additional recovery

occurring because of infill drilling.

Chevron Confidential Use Only Page 154 of 244

 Version 2008.01

Active Infill Drilling Field

10000 300

250
Base 2001
200 Base 2003

Well Count
BOPD

Base 2004
1000 150
Base 2005
100 Total Field
Well Count
50

100 0 Base 2003 indicates

wells drilled prior to
1999 2001 2004 2007 Jan 1st of that year

Figure 8-1: Example of semi-log Rate versus time for Active Drilling Program

Chevron Confidential Use Only Page 155 of 244

 Version 2008.01

Active Infill Drilling Field

5000

4000
Base 2001
3000 Base 2003
BOPD

Base 2004
2000 Base 2005
1000 Total Field

0
0 5 10 15 Base 2003
indicates wells
Cumulative Oil - MMBO drilled prior to Jan

Figure 8-2: Example of Rate versus Cum Oil for Active Drilling Program

Figure 8-3 shows another good example where a drilling program makes calculation of reserves

difficult. It shows the total gas production and well count history for the Grand Valley field, a

tight gas reservoir in the Piceance Basin, Garfield County, Colorado. Almost 800 wells have

been drilled in this field which has increased production from 20 MMCFPD to over 200 MCFPD

in a decade. Production has been on a continuous incline. Base curves can be used to

approximate the reserves being added. Figure 8-4 shows the production for all of the wells that

existed in the field at the beginning of 2002. As shown, the wells declined 32% during year 1 and

14% for the years after. Figure 8-5 shows the base plot for the wells in the field in at the

beginning of 2004. The decline for the first year was 30% followed again by a decline of 14% for

the remaining years. Since the field is showing hyperbolic tendencies, these declines could

probably be applied to the current production to get a reasonable estimate for the developed

reserves as of 1/1/2007.

Chevron Confidential Use Only Page 156 of 244

 Version 2008.01

Figure 8-3 - Grand Valley Field Gas Production

Chevron Confidential Use Only Page 157 of 244

 Version 2008.01

Figure 8-4 - Grand Valley Field - Base 2002 Production

Chevron Confidential Use Only Page 158 of 244

 Version 2008.01

Figure 8-5 - Grand Valley Field Base 2004 Production

8.2 Drilling Wedge Method for Waterflood Fields

An infill drilling program will also affect the WOR versus cumulative oil plots. This occurs

because infill drilling will change the reservoir flow lines and increase the horizontal sweep

efficiency. The new infill wells generally have lower water oil ratios than the older wells. Figure

8-6 shows this effect for a field under continuous drilling.

Chevron Confidential Use Only Page 159 of 244

 Version 2008.01

Infill Drilling Example

Log Water Oil Ratio vs Cummulative Oil

10
WOR

0.1
0 5 10 15 20 25 30 35 40 45
Oil Cum [m3 x 10^6]
Base 2003 indicates
wells drilled prior to
Base 2003 Base 2004 Total Field Jan 1st of that year

Figure 8-6: Effect of semi-log WOR versus Cum Oil for Active Drilling Program

Figure 8-7, Figure 8-8 and Figure 8-9 shows the performance rate time, rate cumulative

and WOR cumulative plots for a North Sea field that produces with high rates, heavy oil

and is rate sensitive. As shown, all of the plots show the validity of the prediction

techniques.

Chevron Confidential Use Only Page 160 of 244

 Version 2008.01

UCS Main (Area A)

LogRATE v Time
Oil Rate (bbls/d) 1,000,000

10,000

100
01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01- 01-
Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan- Jan-
97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20
Cumulative Oil (MMbbls)
1998 1999 2000 2001 2002 2003 2004 2005 Forecast

Figure 8-7 - North Sea Field Rate versus Time

Chevron Confidential Use Only Page 161 of 244

 Version 2008.01

Figure 8-8 - North Sea Field Rate versus Cumulative

Chevron Confidential Use Only Page 162 of 244

 Version 2008.01

Figure 8-9 - North Sea Field - WOR versus Cumulative

8.3 Type Well Method for Drilling or Workover Programs

One very useful DCA evaluation techniques is called the normalized type curve. Several

DCA software packages generate these automatically including PHDWINTM and

PEEPTM. The technique is especially useful in evaluating recently drilled wells or

recompletions. With this technique, the selected wells are time shifted so their dates of

first production are in alignment, the dates are converted to a time that starts with 0.0 and

Chevron Confidential Use Only Page 163 of 244

 Version 2008.01

the total production for each time period is divided by the number of active wells for that

time period. The usual time period increment is months but it can also be days or years

depending on the data available.

Figure 8-10 shows an example of a normalized type curve from the Grand Valley field in

Garfield County, Colorado that was shown in Figure 8-3. A total of 439 completions in

the Williams Fork and Williams Fork-Cameo zones occurred during the years 2004 to

2006. As shown, the average initial gas production for month 2 was 1000 MCFPD but by

the end of 12 months the average production had dropped to an average of 426 MCFPD.

The lower producing rate for month one is because of wells not producing a full month.

This could be adjusted upwards by a factor of 2 assuming that the start dates of the wells

are distributed evenly across the first month. The well count curve drops to 315 by

month 12 and to 143 by month 24. Some of the decrease is due to wells being shut in but

most of the decline occurs because of new wells without a production history greater than

12 to 24 months in length. Note the hyperbolic nature of the curve which has a

hyperbolic exponent “b” equal to 0.73.

Chevron Confidential Use Only Page 164 of 244

 Version 2008.01

Normal Type Curve - Grand Valley Field

10000 500

MCFPD
Wells 450

400

350

300

Wells Count .
MCFPD .

1000 250

200

150

100

50

100 0
0 5 10 15 20 25 30 35 40
Months

Figure 8-10 - Normalized Type Curve

Figure 8-10 is very useful for determining the average or “type” well. A further

technique can generate the P10, P50 and P90 curves for the data. To do this, the data

needs to be arranged into spreadsheet column showing months across the top and well

identifier down the rows. For the Grand Valley field, the upper left portion of the

spreadsheet is shown by Table 8-1 where the data is MCFPM.

Chevron Confidential Use Only Page 165 of 244

 Version 2008.01

Gas Normalized Month

WELL ID 1 2 3 4 5 6
2050010450781000WFCM 7759 38868 31293 32985 33505 30150
2050010450781100WMFK 13022 16169 8513 10775 13715 12361
2050010450812000WMFK 2734 13453 3121 1865 842 2667
2050010450821400WFCM 1024 27395 19361 13756 23938 17893
2050010450892700WFCM 1953 26391 21235 25581 11749 25494
2050010450893000WFCM 9057 26266 14703 22044 28807 24150
2050010450893100WFCM 36500 29812 27219 20998 16792 19035
2050010450899600WMFK 4146 14551 13563 10757 9600 8310
2050010450899701WMFK 4624 17637 18509 13127 11897 10775
2050010450901600WFCM 1994 6949 6459 6210 4917 5472
2050010450901700WFCM 5428 13074 15838 11823 9519 9910
2050010450901800WFCM 3429 6390 4939 4078 3147 2689
2050010450901900WMFK 3477 2876 19595 19132 12894 12124
2050010450902000WMFK 12324 12673 11664 9362 7641 6920
2050010450902100WFCM 2161 1542 12358 16149 12457 13060
2050010450929200WFCM 17653 38792 34404 26870 22697 19583
2050010450929300WFCM 42681 40867 38445 33744 29110 25424
2050010450929400WFCM 4261 22647 35057 32077 26007 20655

Table 8-1 - Grand Valley Normalized Data

This table can be created using the following steps.

1) Downloading the monthly well production data into a spreadsheet with

columns including well ID, date, oil volume, gas volumes and days on

production.

2) If the production database does not have a date of first production, create a

second worksheet containing the well ID and date of first production using

a pivot table with well ID in the left column and min(date) as the value.

3) Add a column to the first table in insert the date of first production using

the vlookup() function.

4) Calculate the months on production using the formula

months = (year × 12 ) + month − (yearstart × 12) − month start + 1

Chevron Confidential Use Only Page 166 of 244

 Version 2008.01

5) Create a Table 8-1 using a pivot table with well ID in the left column,

months in the top row, and sum(production) as the value.

With all of the production in a spreadsheet identical to Table 8-1, the mean, P10, P50

and P90 monthly production can be calculated using the Excel spreadsheet formulas

average(range) and percentile(range, probability) formulas where probability is 0.1

for P10 and 0.5 for P50. The values then may need to be converted into daily rate by

dividing by 30.4, the average days per month. Important: Remember that these

formulas ignore cells which are null but will include cells with a 0 entry. If

drilled wells are shut-in because of performance, a zero production value needs

to be included in the data.

Applying these formulas to the Grand Valley data, we can generate probabilistic type

curves as shown by Figure 8-11. As shown, the P90 well will have a peak rate of

1400 MCFPD while the P10 well will peak at 450 MCFPD. Also notice that the EV

or mean well and the P50 well overlay each other quite closely.

Chevron Confidential Use Only Page 167 of 244

 Version 2008.01

Probabilistic Normal Type Curve - Grand Valley Field

1600

Mean
P10
1400
P30
P50
P70
1200
P90

1000
MCFPD

800

600

400

200

0
0 5 10 15 20 25 30 35 40
Months

Figure 8-11 - Probabilistic Type Curve

Caution: It is important to realize that this is just one technique which takes into

account averaging and ignores dependencies between wells. The Central Limit

Theorem will cause well results to approach the mean if many wells are drilled.

Another equally valid technique would be to do well-by-well DCA and perform statistics

on the reserves distribution. This is more time consuming but eliminates well averaging

errors and is necessary if a considerable amount of money is being invested in future

wells.

Chevron Confidential Use Only Page 168 of 244

 Version 2008.01

9 Field versus Individual Well Decline Curve Analysis

One of the key DCA decisions that the reserve estimator needs to make is whether to

perform the analysis on a field, reservoir, well or other grouping basis. Historically

analyses were done on a field or reservoir basis because the effort required for individual

well analyses were beyond the resources of the typical reserves estimator. The power of

personal computers and the productivity of DCA software packages make it possible to

calculate reserves analysis on a well-by-well basis including economics and group

forecasting.

Calculating reserves on a well-by-well basis has the benefit of eliminating many of the

activity related effects to a grouped production curve which tend to mask the true proved

developed base decline. These activities are discussed at length in starting on page 153 in

the section titled Infill and Extension Well Drilling Effects.

Many industry experts have recognized the difficulties in performing DCA on groups of

wells. Chapman Cronquist1 highlights the following difficulties:

• For grouped wells, the high-rate wells dominate the aggregate decline.

For pressure depletion situations, these wells are declining more rapidly

than low-rate wells and the aggregate production decline might lead to

pessimistic reserve estimates. Conversely for water influx reservoirs, the

high-rate wells may decline less rapidly than the low rate wells and result

Chevron Confidential Use Only Page 169 of 244

 Version 2008.01

in overly optimistic reserves estimates especially if wells tend to rapidly

water out due to a gravity stable rising water-oil-contact.

• Once wells in a reservoir reach semi-steady-state with well interference

boundaries established, shutting in one or more such wells might result in

the expansion of the drainage volumes of surrounding wells. Such an

expansion typically results in a slight increase in reserves for the affected

wells, a phenomenon that might not be apparent in the aggregate decline.

• Another difficult aspect of multiwell aggregates is determination of the

number of producing wells in future years and the minimum number of

wells at the economic limit. For operations not requiring central facilities,

a single well may be the economic limit. For operations such as offshore

platforms, the well count at economic limit might be a significant number

of wells.

• In estimating wells for large multiwell fully developed fields, engineers

typically assume that the well count will decrease uniformly with time. It

might be more appropriate to decrease the number of producing wells so

the remaining lives have a log-normal distribution.

• Curtailment for reasons such as pipeline capacity, market restrictions, and

water handling capacity or treating limits may cause artificially low

declines. The engineer should understand the field operations and be

aware of any curtailment when calculating reserves using multiwell

aggregate DCA analysis.

Chevron Confidential Use Only Page 170 of 244

 Version 2008.01

Bill Cobb10 discusses the effects of multiwell aggregate versus individual well DCA

analysis when using WORs. He makes the following comments:

• The WOR versus Np graph is most dependable when applied on a well-

by-well basis so that the good, bad and ugly (low, medium, and high)

WOR wells are not combined.

• Oil and water production rates are influenced by water injection rate.

Significantly, when applied on a well-by-well basis, the WOR plot is

independent of injection rates.

• WOR plots on multiwell graphs are usually more difficult to analyze than

single wells due to different conditions in the various wells.

9.1 DCA by Field Groupings

In spite of the above listed comments, it is not always practical to calculate reserves using

well-by-well declines. Some fields are too large and have too many wells, or reliable and

accurate well tests are not available. Sometimes a good compromise is to perform DCA

on subsets of the total field data provided foresight is used to select the best subset

groupings. Sometimes groups should be selected by area to fit geologic, drive

mechanism or reservoir fluid trends in the field. Groups can also be based on the

maturity of field development so the areas developed early are evaluated separately from

the areas developed later.

Chevron Confidential Use Only Page 171 of 244

 Version 2008.01

Groups can also be based on the mechanical equipment configuration. In most fields, oil,

water, and gas production are recorded accurately by separate tank batteries or production

gathering stations serving all of the wells in a certain areas. A good compromise

technique for large fields is to perform DCA on the production at the gathering station

level. This eliminates well allocation errors assuming production is allocated to the wells

from the metered total production.

Groupings also have the advantage of minimizing the work requirements for those areas

of the field with stable well counts or low development activity levels. Areas with

drilling or workover activity should still be reviewed on a well basis or by using one of

the other recommended techniques for separating out these effects.

9.2 Decline and Rate Statistics for a Field

Another technique is to perform individual DCA on the higher rate wells and on groups

for the lower rate wells. This requires that statistical analysis be performed on the

individual wells in a field. The distribution of well production rates at any point in time

is log normal3. Figure 9-1 shows the log normal nature of the distribution for the Grand

Valley Field in Colorado. This rate was calculated using the last three months of gas

production data in 2006. This field was previously presented in Figure 8-3, Figure 8-4

and Figure 8-5. In this field, one might be able to calculate reserves by well for wells

Chevron Confidential Use Only Page 172 of 244

 Version 2008.01

with a current rate higher than 400 MCFPD and group the remaining wells in one or more

groups.

Grand Valley Field - Average Rate Distribution

140

120

100

80
Well Count

60

40

20

0
0 200 400 600 800 1000 1200 1400
Oct-Dec, 2006 Average MCFPD

Figure 9-1 - Grand Valley Well Rate Distribution

Besides showing the producing rate distribution, it can also be very useful to review the

distribution of the declines for the latest period of time. This can be done by loading the

data into an Excel spreadsheet and using the logest() function to calculate the

mathematical hard fit of the declines for the last 12, 18 or 24 months. Figure 9-2 shows

the distribution of declines for the Grand Valley field. This decline is the mathematical

best fit exponential decline for the 12 months of 2006 gas rate data by well. Shut-in

Chevron Confidential Use Only Page 173 of 244

 Version 2008.01

months were treated as null data points. As shown, the bulk of the declines are between

10% and 60% per year. The negative decline portion is where production is inclining.

This may be occurring due to well clean up, fracture treatments, artificial lift installation

or gas compression installation. Some other technique such as volumetric estimates or

analogies with offset wells would need to be applied to the wells with inclining

production.

Grand Valley Field - Exponential Decline Distribution

70

60

50

40
Well Count

30

20

10

0
-100% -80% -60% -40% -20% 0% 20% 40% 60% 80% 100%
Exponential Decline - %

Figure 9-2 - Grand Valley 12 Month 2006 Exponential Decline

A third step in field wide statistics is to generate a scatter plot of average producing rate

versus decline. Figure 9-3 shows this plot for Grand Valley and includes the rate data

from Figure 9-1 and the decline data Figure 9-2. The plot also contains lines of constant

Chevron Confidential Use Only Page 174 of 244

 Version 2008.01

remaining reserves of 0.1 BCF up to 2.0 BCF. These lines were generated for the IP and

decline rate required to equal the remaining reserves for each line. They assume an

abandonment rate of 10 MCFPD. From this plot one can get an eyeball estimate as to

what a typical well might recover in remaining reserves.

Grand Valley Field - Decline Versus Rate

100%

50%
Exponential Decline - %

2006 Decline
2 BCF
1 BCF
0%
0.5 BCF
0.25 BCF
0.1 BCF

-50%

-100%
0 200 400 600 800
Gas Rate - MCFPD

Figure 9-3 - Grand Valley Rate versus Decline Scatter Plot

9.3 Advantages of Calculating Reserves by Well

If reserves are done on a well-by-well basis across multiple years, the engineer can

document very clearly any performance reserves changes in a field or reservoir through

Chevron Confidential Use Only Page 175 of 244

 Version 2008.01

the uses of a table showing remaining reserves and estimated ultimate recovery for each

well by year. Table 9-1 shows a possible format for such a table.

Reserves by Well for the ABC Field – XYZ Formation

2005 2006 Δ EUR

Well Reserves - MBO EUR - MBO Reserves - MBO EUR - MBO MBO

1 230 400 230 400 0

2 221 434 188 450 16

3 45 125 23 120 -5

4 66 99 20 99 0

5 87 90 0 45 -45

6 124 155 99 155 0

7 111 133 85 144 11

Total 884 1436 645 1413 -23

Table 9-1 - Ultimate Recovery Changes

Another advantage provided by a well-by-well analysis is that if each well is forecasted

to its economic limit, the summation of the individual forecasts will give a combined

production forecast and well count that can be used for field budgeting purposes. If

grouped forecasts are made, the engineer needs to provide justification for how the well

Chevron Confidential Use Only Page 176 of 244

 Version 2008.01

counts and production curtails late in the life of the field as wells are depleted or watered

out.

Figure 9-4 shows an example of a coal bed methane field with 480 wells where the field

grouped forecast is compared to the summation of the individual well forecasts. Coal bed

methane wells have the unique characteristic of increasing in production as the field is

dewatered. Once the field is dewatered, the wells began producing with declining

production. In this example, the heavy red line to the right of the reserves date is the

summation of the individual forecasts for the 480 wells. The dark green line shows the

well count with time. This count drops as wells reach their economic or physical

production limit. The light red forecast is an exponential best fit analysis for the history

of the field before the reserves date when the well count is relatively stable. As shown,

the field analysis has a lower decline (11.4%) than the summation of the individual wells

decline (14.8%). In this case the field grouped estimates would probably over predict the

future reserves.

Chevron Confidential Use Only Page 177 of 244

 Version 2008.01

Figure 9-4 - Gas Field Forecast Example for Individual Wells

The latest DCA software packages will generally also print individual well curves and

forecasts to either a printer or an electronic format such as jpeg, pdf or bmp files. Some

packages do it better than others. These should be generated using a batch printing

routine with the results sorted from largest to smallest reserves. The field should then be

archived to the reserves electronic folder for documentation. The wells contributing the

most proved reserves can be quickly and conveniently presented during the RAC meeting

for sign off.

Chevron Confidential Use Only Page 178 of 244

 Version 2008.01

9.4 Aggregating Wells or Reservoirs to a Hyperbolic Decline

Although not recognized by many reservoir engineers, hyperbolic declines will occur

when DCA is performed on aggregated wells or reservoirs that are performing with

exponential declines of different values. Brons25 recognized this as early as 1963. It is

important that the QRE recognize this effect when forecasting field totals generated from

a number of exponential declines.

To demonstrate this effect, let’s look at an example.

Example – A field has four oil reservoirs all producing with exponential declines. The

reservoirs rates and exponential declines are:

Reservoir BOPD Decline - %

1st Sand 500 6%
2nd Sand 700 10%
3rd Sand 800 16%
4th Sand 900 22%
Total 2900

Table 9-2 – Multiple Reservoir Declines

Let’s generate a 20 year field decline, plot the results, calculate the hyperbolic exponents

and compare the reserves:

Chevron Confidential Use Only Page 179 of 244

 Version 2008.01

BOPD Annual
Year 1st Sand 2nd Sand 3rd Sand 4th Sand Total Decline
0 500 700 800 900 2900 NA
1 470 630 672 702 2,474 14.7%
2 442 567 564 548 2,121 14.3%
3 415 510 474 427 1,827 13.9%
4 390 459 398 333 1,581 13.5%
5 367 413 335 260 1,375 13.1%
6 345 372 281 203 1,201 12.7%
7 324 335 236 158 1,053 12.3%
8 305 301 198 123 928 11.9%
9 286 271 167 96 820 11.6%
10 269 244 140 75 728 11.2%
11 253 220 118 59 649 10.9%
12 238 198 99 46 580 10.6%
13 224 178 83 36 520 10.3%
14 210 160 70 28 468 10.1%
15 198 144 59 22 422 9.8%
16 186 130 49 17 382 9.6%
17 175 117 41 13 346 9.4%
18 164 105 35 10 314 9.2%
19 154 95 29 8 286 9.0%
20 145 85 24 6 261 8.8%
Reserves - 2,094 2,130 1,624 1,313 7,160
MBO

Table 9-3 – Exponential Declines b y Reservoir

Chevron Confidential Use Only Page 180 of 244

 Version 2008.01

A plot of the results of each reservoir and the field aggregation from Table 9-3 are shown

by Figure 9-5. Figure 9-6 shows how the aggregated field deviates from the exponential

decline.

10000

1st Sand
1000
2nd Sand
BOPD

3rd Sand

100 4th Sand

Field

10
0 5 10 15 20
Years

Figure 9-5 – Reservoirs and Field Total Decline

10000
b = 0.25
di = 15.0%
BOPD

1000 Field

100
0 5 10 15 20
Years

Figure 9-6 – Field Total Decline Compared to Exponential

Chevron Confidential Use Only Page 181 of 244

 Version 2008.01

q20 = 2900 × (1 + 0.25 × (− ln(1 − 0.150)) × 20) −1/ .25 = 269 BOPD

2900 0.25 × ( 2900 (1−0.25) − 269 (1−0.25) )

Np Hyperbolic = × 365 = 7,224 MSTBO
− ln(1 − .15) × (1 − 0.25)

Np Exponential = 2,094 + 2,130 + 1,624 + 1,313 = 7,160 MSTBO

As shown, the difference between exponential and hyperbolic is roughly 1%. Sometime

the difference can be considerably larger.

Chevron Confidential Use Only Page 182 of 244

 Version 2008.01

10 Advanced Decline Curve Analysis – Rate Transient

Analysis

When J. J. Arps published his decline curve equations for exponential, hyperbolic and

harmonic decline curves, there was no engineering basis given for the equations. The

industry (discussed earlier in this manual) has since discovered that the exponential

decline occurs with volumetric depletion of a fluid with a constant compressibility.

There is no engineering basis for the hyperbolic curves other than they have been shown

to work in general petroleum engineering practice.

A large improvement in the technology of decline curve analysis occurred in the 1980

through the remarkable work of M. J. Fetkovich. He developed a set of type curves to

analyze wells in both the transient and pseudosteady-state flow periods of the producing

lives of wells. Figure 2-3 shows an example of a Fetkovich type curve.

It plots dimensionless rate versus dimensionless time. The fan curves to the left side of

the graph shows the transient flow period for a variety of re/rw reservoir values. The right

side shows the pseudosteady-state depletion portion with the fan curves representing the

hyperbolic exponent b.

Chevron Confidential Use Only Page 183 of 244

 Version 2008.01

10.1 Blasingame Plot Type Curve

The next large improvement in the advanced decline curve analysis came from work by

Tom A. Blasingame, a petroleum engineering professor from Texas A&M University in

the late 1980’s and early 1990’s. Blasingame developed an improved type-curve analysis

that incorporates reservoir conditions, material balance, rate information and pressure

information to analyze wells. Provided sufficient rate and pressure data is available and

the well meets certain flow boundary conditions, the Blasingame analysis can provide

OOIP, OGIP, permeability, drainage area, skin factor and initial pressure (assuming it is

not known).

The Blasingame plot contains three curves plotted on log versus log scales. The x axis is

material balance time with hours as the units. The calculation for material balance time is

shown by Equation 10-1 and is essentially the cumulative production divided by the

current rate.

Q( t )
te =
q( t )

Equation 10-1 - Material Balance Time

The first Blasingame curve is the normalized rate curve which is defined by Equation

10-2.

Chevron Confidential Use Only Page 184 of 244

 Version 2008.01

q( t )
PI( t ) =
pi − p w (t)

Equation 10-2 - Normalized Rate

The second Blasingame curve is the normalized rate integral. It is shown by Equation

10-3.

t te
1 e 1 q( τ)
PI Int. = ∫ PI(τ)dτ = ∫p dτ
te 0 te o i − p w ( τ)

Equation 10-3 - Normalized Rate Integral

The third Blasingame curve is the normalized rate integral derivative. It is shown by

Equation 10-4.

∂(PI Int )
PI Int. Derivative =
∂ ln( t e )

Equation 10-4 - Normalized Rate Integral Derivative

Figure 10-1 shows an example of a Blasingame plot generated using the Kappa

Engineering TopazTM software. The type curves are manually shifted to best fit the data.

Chevron Confidential Use Only Page 185 of 244

 Version 2008.01

Figure 10-1 – TopazeTM Blasingame Plot

10.2 Log-log Plot Type Curve

The next method using in advanced decline curve analysis is also derived from the work

of Blasingame. It is called the Log-log plot. It also uses material balance time for the x

axis but only plots two curves. The first curve is the integral of normalized pressure as

shown by Equation 10-5.

te
1 p i − p w ( τ)
Integral of Normalized Pr essure =
te ∫
0
q( τ)
dτ

Equation 10-5 - Integral of Normalized Pressure

Chevron Confidential Use Only Page 186 of 244

 Version 2008.01

The second curve is the Integral of normalized pressure derivative shown by Equation

10-6.

∂ (Integral of normalized pressure)

∂ ln( t e )

Equation 10-6 - Integral of Normalized Pressure Derivative

Figure 10-2 shows the log-log plot as generated by the Kappa Engineering TopazeTM

software. Like the Blasingame plot, the type curve is shifted to best fit the data.

Figure 10-2 - TopazeTM Log-log Plot

Chevron Confidential Use Only Page 187 of 244

 Version 2008.01

10.3 Advanced Decline Curve Analysis Software

The two main software packages used in Chevron are Kappa Engineering’s TopazeTM

software and Fekete [Link] RTA software. Like all software packages that are

compared, each has things that it does better than the other. The TopazeTM software is

part of the EcrinTM package which includes both TopazeTM and the pressure build up

software module SaphirTM. This software is available from the Chevron GIL Options

panel and uses a world-wide licensing system. The Fekete software is also available

from the Gil Options panel but is mainly supported and licensed by the MCA SBU.

This advanced DCA software is only possible because of the power of today’s modern

desktop computers. For best applicability the wells analyzed should have the following

boundary and reservoir characteristics:

• Boundaries equi-distant from the wellbore (or infinite acting)

• Constant reservoir properties (porosity, thickness, permeability)

• Constant saturations and relative permeability

• A single producing zone or layer

• No influences from offset production or water influx

TopazeTM also has a modeling feature that allows more detailed reservoir characteristics

including faulting, multiple wells, etc.

Chevron Confidential Use Only Page 188 of 244

 Version 2008.01

To analyze a well it is required that one has an understanding of the reservoir

characteristics for the well along with the rate and producing bottom-hole pressure over

the history of the well. The software packages do have flowing tubing/casing gradient

capability to calculate flowing bottom-hole pressure from surface pressures.

The subject of advanced decline curve analysis is some what beyond the scope of this

manual. It requires a two day to a five day course to become grounded in the techniques

and the software packages. The author highly recommends that any petroleum engineer

who is wants to improve his engineering capabilities include this subject into his personal

training program.

To give a brief introduction into the capabilities of advanced DCA, let’s work an example

using the TopazeTM software.

Example – An oil well has been on flowing production water free for five years. The

initial month production rate was 700 BOPD. The well has produced with a 25%

exponential decline. The flowing wellhead pressure has been held constant at 400 psia.

The reservoir has the following conditions.

Porosity = 24%

Net Pay = 50 feet

So = 0.65

Sw = 0.35

Chevron Confidential Use Only Page 189 of 244

 Version 2008.01

Initial Reservoir Pressure = 4,000 psia

Depth = 8,500 feet

Reservoir Temperature = 190 degrees F

Water Salt = 100,000 ppm

Formation Compressibility = 3E-6 psi-1

GOR = 700 SCF/STBO

Oil Formation Volume Factor = 1.374

Oil Viscosity = 0.515 cp

Total Compressibility – Ct = 1.182E-5 psi-1

Well Radius - rw = 0.3 feet

Tubing Size – ID = 2.441 inches

Drainage Geometry = Circular Finite Radius

The 400 psia flowing tubing pressure should be converted to a flowing bottom-hole

pressure using the tubing inflow correlations and TopazeTM used to model the drainage

radius, effective permeability and original oil-in-place using TopazeTM. Assume that the

skin factor is 0 and that the initial reservoir pressure is known. At the end of five years,

artificial lift will be installed. Artificial lift will allow the well to produce with a 500 psia

bottom-hole pressure. Model the production rate under artificial lift for the next five

years and calculate the exponential decline.

Chevron Confidential Use Only Page 190 of 244

 Version 2008.01

The flowing tubing pressure of 400 psia is entered into TopazeTM using the screen shown

in Figure 10-3. The production history is entered using the screen shown in Figure 10-4.

The data tables for both pressure and production can be constructed in a spreadsheet and

pasted into the software.

Figure 10-3 - Pressure History

Chevron Confidential Use Only Page 191 of 244

 Version 2008.01

Figure 10-4 - Production History

Figure 10-5 shows the model match using the Blasingame plot method. Figure 10-6

shows the model match using the Loglog plot method. As shown for this simple case,

very good model matches can be achieved for both methods.

Chevron Confidential Use Only Page 192 of 244

 Version 2008.01

Figure 10-5 - Blasingame Plot Match

Figure 10-6 - Loglog Plot Match

Chevron Confidential Use Only Page 193 of 244

 Version 2008.01

Figure 10-7 shows the production and pressure history matches. The upper plot shows

the historical production data in yellow with the TopazeTM model match in red. The plot

with the negative slope is the production versus time. The plot with the positive slope

shows the cumulative production Np. As shown a very good production history match

was achieved.

The lower plot in Figure 10-7 shows the pressure history. The solid yellow line is the

average reservoir pressure. It declined from the initial pressure of 4000 psia to

approximately 2300 psia at the end of 5 years. The green points are the flowing bottom-

hole pressure calculated by TopazeTM using the flow tubing curve gradients. The red line

shows the history match for the flowing bottom-hole pressure.

Chevron Confidential Use Only Page 194 of 244

 Version 2008.01

Figure 10-7 - Model Production and Pressure Match

Figure 10-8 shows the model results reservoir parameters necessary for the history match

assuming a skin factor of 0 and an initial reservoir pressure of 4000 psia. The key

parameters are a permeability of 5.65 md, a drainage radius of 2600 feet (drainage area of

488 acres) and an initial oil-in-place of 21.4 MMSTBO. Because of the large spacing

and low permeability, the recovery factor at the end of 10 years was 5.7%.

Chevron Confidential Use Only Page 195 of 244

 Version 2008.01

Figure 10-8 - Model Results

Figure 10-9 shows the TopazeTM forecast for the artificial lift case where the working

bottom-hole pressure is lowered to 500 psia. The production was increased from 170

BOPD to almost 591 BOPD by the artificial lift installation. Figure 10-10 shows the

production plotted on semi-log scale. The effective decline for the flowing period was

25% per year which increased to 27% per years for the artificial lift period.

Chevron Confidential Use Only Page 196 of 244

 Version 2008.01

Figure 10-9 - Artificial Lift Forecast

Topaze Example - Artificial Lift

10000

Natural Flow
Artificial Lift
Expon. (Natural Flow)
Expon. (Artificial Lift)

1000

y = 2830.2e-0.3156x
BOPD .

y = 699.95e-0.288x

100

10
0 2 4 6 8 10 12
Years

Figure 10-10 - TopazeTM Semi-log Artificial Lift Declines

Chevron Confidential Use Only Page 197 of 244

 Version 2008.01

11 Probabilistic Decline Curve Analysis

Considerable effort has been expended throughout Chevron applying probabilistic

methods to decline curve analysis. After six years of attending RAC meetings, the

author’s opinion is that this effort has been less than successful in generating value

through better reserves estimates and production forecasts. There are several reasons

why probabilistic DCA has not produced the value anticipated. Each will be discussed

separately.

11.1 Reasons for Poor Success of Probabilistic DCA

One reason for the poor success is that declines do not fit well with probabilistic

calculations. Thus the most common probabilistic DCA method is really not

probabilistic. The usual technique involves three steps. The first is a review of the

history of a well or field. The second is selecting a time period with a high decline and

calling that decline a P10 decline. The third involves selecting a low decline time period

and calling that decline a P90 decline. These declines are then put into a Monte Carlo

simulator to generate a probability exponential decline distribution.

There are two major flaws with this method. One is there is really no way to calculate

the probability of the declines selected. They are really just a high and a low estimate.

The second flaw is that most of the reservoir and operating conditions for the time

periods of the declines are no longer relative and valid for the latest DCA estimate. As

Chevron Confidential Use Only Page 198 of 244

 Version 2008.01

discussed in Section 5.4 “Comparison of Exponential, Hyperbolic and Harmonic

Declines,” declines are governed mostly by reservoir and operating boundary conditions.

The uncertainty in reservoir or boundary conditions such as WBHP or well testing

accuracy are probably the more valid probabilistic variables.

Another reason for the poor success is that declines from semi-log rate time plots tend to

become lower with time. As shown and discussed in Table 5-4 - Hyperbolic Exponent

Frequency – Arps (1944), most wells have a hyperbolic exponent greater than 0.0. Thus

the decline will become lower with time. Where the common practice is to use

exponential declines to calculate reserves, the chances are high that reserves will have to

be added in the future due to hyperbolic nature of oil and gas wells.

Some publications such as the SPE reserves manual1 state that a least squares best fit

(LSBF) is a P50 estimate. This means that the exponential forecast line has an equal

chance of being higher or lower than the P50 estimate assuming a normal distribution.

This is in conflict with hyperbolic nature of many producing wells. Where declines

increase with time, the reason is usually mechanical and is caused by skin damage,

abandoned wells, equipment problems, etc.

Another difficulty in probabilistic DCA comes from the Central Limit Theorem. A

detailed discussion on the Central Limit Theorem is available in the Chevron Reserves

Manual. When probabilistic DCA is performed on an individual well basis for a group of

wells and then summed probabilistically treating each well as a probabilistic variable, the

Chevron Confidential Use Only Page 199 of 244

 Version 2008.01

distribution of the total quickly collapses to a very narrow decline range. This range is

much tighter than the range used on most field probabilistic DCA evaluations.

Let’s use an example to better explain the Central Limit Theorem effect. A typical well

in a field has a P10 to P90 exponential decline rate from 0.15 to 0.30. What would the

average decline be for 50 wells averaged probabilistically? Figure 11-1 shows a

comparison between single well distribution and the 50 well distribution. Table 11-1

shows the percentile values. As shown, the P10 to P90 values range from 0.21 to 0.24.

Figure 11-1 - Probabilistic DCA - One Well versus 50 Wells

Chevron Confidential Use Only Page 200 of 244

 Version 2008.01

Percentile Single Well 50 Wells Average

P99 0.00 0.19

P90 0.15 0.21
P80 0.17 0.22
P70 0.19 0.22
P60 0.21 0.22
P50 0.22 0.23
P40 0.24 0.23
P30 0.26 0.23
P20 0.27 0.23
P10 0.30 0.24
P0 0.44 0.26

Table 11-1 - Probabilistic DCA Percentiles

As shown, the overall P10 to P90 distribution of the exponential decline for the field

would be between 0.21 and 0.24. This example assumes no correlation between wells.

An assumed correlation between wells would widen the range slightly but it would still

be quite narrow.

11.2 Improvements to Probabilistic DCA

In most situations, probabilistic DCA does not add sufficient value for the extra effort.

For those situations where business reasons require probabilistic DCA, there are several

approaches that may offer improvements. Some approaches might investigate the

Chevron Confidential Use Only Page 201 of 244

 Version 2008.01

uncertainty in data accuracy, boundary conditions or reservoir drainage or fluid

conditions.

A second improvement approach is to randomly take subsets of the best fit data point and

calculate a series of declines that can be statistically analyzed. This method is called the

“Bootstrap” method and is based on a 1996 SPE paper by Jochem and Spivey37. Care

needs to be taken that data points that are mechanically invalid are eliminated. The

sample size of the random data selection also needs to be carefully chosen as this directly

affects the variation of the decline estimate.

A third and more practical approach might be to focus on the hyperbolic exponent “b”

instead of on the exponential decline. The largest unknown in the Arps equation is the

hyperbolic exponent “b.” Most declines can be represented by the Arps hyperbolic

Equation 5-13 where the hyperbolic exponent varies between 0.0 and 1.0. Exponential

and harmonic declines are really only special case hyperbolic declines where the

exponent “b” is 0.0 and 1.0 respectively. Since rate time data is represented by a time

series and the current data points can be easily determined, the initial production rate and

initial decline can readily be estimated. Statistics from reviewing similar fields can then

be used to determine the most likely distribution of “b.” Remember not to fall into the

common mistake of determining the distribution of the “b” on a well basis and then

applying that distribution to the field as a distribution of the mean.

Chevron Confidential Use Only Page 202 of 244

 Version 2008.01

Let’s look at an example that varies the hyperbolic exponent “b.” A field is under

volumetric depletion. It currently is producing 1000 BOPD and has an established

exponential decline over the last 12 months of 24%. A study of similar fields in the basin

shows that the hyperbolic constant has a P10 value of 0.10 and a P90 value of 0.50 with a

log normal distribution. The economic limit of the field is known to be 100 BOPD.

Calculate the reserves S-curve and the expected value reserves.

The distribution of the hyperbolic exponent is shown by Figure 11-2. It was generated

using a Monte Carlo simulator with the P10 and P90 values in a log normal distribution.

Figure 11-2 - Exponent Distribution

The reserves are then calculated using the following equation assuming the current rate is

the initial rate.

Chevron Confidential Use Only Page 203 of 244

 Version 2008.01

1000 b × (1000 (1−b ) − 100 (1−b ) )

Re serves =
− ln(1 − 0.24) × (1 − b)

The percentile values for the hyperbolic exponent and the associated reserves are shown

by Table 11-2. The reserves S-curve is shown by Figure 11-3.

Probability Hyperbolic Exponent “b” Reserves - MBO

P0 0.02 1,215
P10 0.1 1,292
P20 0.13 1,323
P30 0.16 1,354
P40 0.19 1,388
P50 0.22 1,422
P60 0.26 1,470
P70 0.31 1,534
P80 0.38 1,631
P90 0.5 1,819
P100 0.99 3,027
Table 11-2 - Hyperbolic Distribution and Reserves

Chevron Confidential Use Only Page 204 of 244

 Version 2008.01

Figure 11-3 - Reserves S-curve

The P1 probabilistic reserves are equal to P10 or 1292 MBO. The P2 reserves are P50-

P10 or 1422-1292 = 130 MBO. The P3 reserves are P90-P50 or 1819-1422 = 397 MBO.

Since the distribution is log normal, the EV weighting factors are 1.0, 0.7 and 0.3. The

EV reserves are then calculated as follows:

Reserves ev = 1292 + 0.7 × 130 + 0.3 × 397 = 1,502 MBO

Chevron Confidential Use Only Page 205 of 244

 Version 2008.01

12 Material Balance in Gas Reservoirs

Approximately 30% of the reservoirs in Chevron are classified as gas or associated gas

reservoirs. Material balance methods are used to calculate the reserves on most of these

reservoirs. The equations of state and properties that govern natural gas are considerably

simpler than crude oil. In crude oil systems, the accuracy of the data must be higher than

is available on many of our reservoirs, especially the older reservoirs which had

considerable gas flaring in early life.

12.1 Ideal Gas Law and Gas Equations

The basis for all gas calculations is the ideal gas law given by Equation 12-1.

pV = znRT

Equation 12-1 - Ideal Gas Law

p = pressure, psia,
V = volume, ft3,
z = compressibility factor, dimensionless,
n = No. of pound moles or pounds/molecular weight,
R = Universal gas constant, 10.73 for all gases,
T = Temperature, absolute.

Chevron Confidential Use Only Page 206 of 244

 Version 2008.01

If the pressure is converted to pound per square foot, the universal gas constant is

multiplied by 144 square inches per square foot and becomes the more familiar 1545

value.

In reservoir engineering, we generally do not need to calculate the mass of the natural gas

we are calculating reserves on. What are more common are calculations comparing a

mass of gas at reservoir conditions with the same mass of gas at surface conditions. This

generates Equation 12-2.

p1 V1 p 2 V2
=
T1 z1 T2 z 2

Equation 12-2 - Gas Volume Comparison

When point 2 is surface conditions, P2 is 14.73 psia by Chevron standards unless local

regulations require another pressure, T2 is 520 degrees Rankin (460 + 60 degrees

Fahrenheit) and Z2 is 1.0. After substituting the constants for standard surface conditions,

Equation 12-2 becomes:

p V 14.73 Vscf
= = 0.028327 Vscf
Tz 520 × 1.0

Equation 12-3 – Standard Conditions

Chevron Confidential Use Only Page 207 of 244

 Version 2008.01

Using Equation 12-3, the gas formation volume factor Bg can be calculated for each of

the common conversions. Reservoir barrels are converted to cubic feet using 5.6146

cubic foot per barrel.

0.005045 z T
B g (rb / scf ) =
p

Equation 12-4 – Bg (Reservoir Barrel per SCF)

198.207 p
B g (scf / rb) =
Tz

Equation 12-5 – Bg (SCF per Reservoir Barrel)

0.028327 z T
B g (rcf / scf ) =
p

Equation 12-6 – Bg (Reservoir CF per SCF)

35.302 p
B g (scf / rcf ) =
Tz

Equation 12-7 – Bg (SCF per Reservoir CF)

Chevron Confidential Use Only Page 208 of 244

 Version 2008.01

The original gas-in-place for non-associated gas reservoir or gas caps can be calculated

using Equation 12-8.

43,560 × φ × (1 − S w ) × A × h
OGIP =
B gi

Equation 12-8 – Gas-In-Place

OGIP = free gas initially in place, scf

43,560 = cubic feet in an acre foot
Φ = average porosity, decimal
Sw = water saturation, decimal
A = area, acres
h = net pay, feet
Bgi = initial formation volume factor, rcf/scf

If the Bgi units are rbbl/scf, the constant 7,758 is substituted for 43,560.

Example – A 10,000 foot dry gas reservoir covering 2,500 acres has an initial pressure of

5000 psia. The average reservoir characteristics are 200 degree Fahrenheit temperature,

water saturation of 30%, porosity of 18% and net pay equal to 30 feet. The gas has a

specific gravity of 0.70 (air = 1.0). Calculate the OGIP assuming an initial gas z factor of

0.990.

0.028327 × 0.990 × ( 460 + 200)

B gi = = 0.003702 (rcf / scf )
5000

43,560 × 0.18 × (1 − 0.25) × 2500 × 30

OGIP = = 119.1 × 10 9 scf
0.003702

Chevron Confidential Use Only Page 209 of 244

 Version 2008.01

The gas produced from a reservoir is the gas initially in place minus the current gas in

place. Equation 12-8 can be modified to calculate the cumulative production between the

initial conditions and any pressure. For abandonment pressure this equation becomes

520 ⎡p p ⎤
Gp SCF = 43,560 × A × h × φ × (1 − S w ) × ×⎢ i − a⎥
14.73 × T ⎣ z i z a ⎦

Equation 12-9 - Gas Produced

If one makes the assumption in Equation 12-9 that none of the reservoir conditions

change and that no water influx is occurring, all of the variables except for the reservoir

pressure and the gas compressibility factor remain constant. This means that the

relationship between the gas produced and reservoir pressure adjusted for gas

compressibility is linear. This is the basis for the most common gas material balance

plot, the P/Z versus cumulative gas plot.

For our previous example, generate a table showing the gas produced for pressures of

4000 psia, 3000 psia, 2000 psia and an abandon pressure of 1000 psia. The Z factors for

each of these pressures are 0.920, 0.872, 0.864 and 0.911 respectively. Also construct a

P/Z versus cumulative plot and extrapolate the plot to a zero pressure to check the

previous OGIP calculation.

Pressure (Psia) Z P/Z GIP (BCF) Gp (BCF)

5,000 0.990 5051 119.1 0.0

Chevron Confidential Use Only Page 210 of 244

 Version 2008.01

4,000 0.920 4348 102.6 16.6

3,000 0.872 3440 81.2 38.0
2,000 0.864 2315 54.6 64.5
1,000 0.911 1098 25.9 93.2

Example P/Z Versus Cumulative Gas

6000

5000

4000
P/Z

3000

2000

1000

0
0 20 40 60 80 100 120 140
Gp - BCF

Figure 12-1 - P/Z versus Cumulative Plot Example

As shown, Figure 12-1 plots a linear line which extrapolates to the OGIP of 119 BCF

calculated in the example.

12.2 Gas Reservoir Data Acquisition

When gas material balance methods are used to calculate reserves, it is extremely

important that good pressure and production data be collected. An initial reservoir

Chevron Confidential Use Only Page 211 of 244

 Version 2008.01

pressure should be taken when the field is discovered using an MDT or similar tool.

Once a reservoir is on production, a sufficient number of reservoir pressures should be

acquired annually so that the average reservoir pressure can be calculated. A data

collection program should also be in place where pressures are recorded in the event of

any planned or unplanned downtime.

Accurate gas, water and condensate production should also be available. Fortunately gas

rates on individual wells are generally available since most wells are equipped with a

continuous recording sales meter. Extra equipment is required if condensate or water

needs to be measured. Recording accurate water production is especially important if

water influx is suspected.

There are several ways that pressure data can be collected. Each will be discussed

separately.

MDT Logs - As mentioned above, the initial pressure of any reservoir should be

recorded at discovery. As additional wells are drilled, unless economics prevent it, MDT

pressure sampling should be taken on each well. When all the depth versus pressure data

is plotted on a single graph, one can quickly visualize the amount of depletion that is

occurring in the reservoir and in what zones. Equally important, one can see which wells

appear not to be in pressure communication.

Chevron Confidential Use Only Page 212 of 244

 Version 2008.01

Continuous Downhole Gauges – Many new large gas reservoirs are being equipped

with high accuracy downhole gauges that transmit data back to surface in real-time. The

data available enable the wells to be monitored continuously. Pressure buildup and

drawdown analysis can be analyzed looking for performance degradation and level of

depletion. This data enables very accurate history matches and analysis using reservoir

simulation. They provide excellent data for material balance calculations.

Retrievable Downhole Pressure Gauges – One of the most common methods to get

annual pressure information is using a wireline retrievable pressure recording gauge.

They can be run in a well to record the static bottomhole pressure after the well has been

shut-in and stabilized. They can also be run in the well while the well is producing and

left on bottom. The well can then be shut in and allowed to build up over a period of

time so the SBHP build up can be used to analyze the well. The pressure gauge is then

wireline retrieved and the data recovered.

Surface Pressure Recordings – Another very common technique is to simply record the

surface wellhead pressure after a well has been shut-in and stabilized. The surface

pressure should be recorded with an accurate pressure gauge, preferably one that has been

calibrated using a dead weight tester. Some engineers will construct the P/Z plot using

surface pressure. Figure 12-2 shows the both the downhole and surface plot for our

previous example. Although the surface plot will have sufficient accuracy in the later life

of the field for reserves purposes, in the early life the method may under predict the GIP.

Chevron Confidential Use Only Page 213 of 244

 Version 2008.01

Comparison of BHP/Z versus WHP/Z

6000

SBHP/Z
WHP/Z
5000 Linear (WHP/Z)
Linear (SBHP/Z)

4000
PSIA/Z

3000

2000

1000

0
0 20 40 60 80 100 120 140
Gp - BCF

Figure 12-2 - Surface P/Z Versus SBHP/Z

A better technique is to calculate the SBHP from the surface WHP. This calculation is

performed using Equation 12-10.

0.01875×SG gas ×D
z avg ×Tavg
SBHP = WHP × e

Equation 12-10 - SBHP from WHP in Gas Well

SGgas = specific gravity of gas (air = 1.0)

D = true vertical depth, feet

Zavg = average compressibility factor for gas column

Tavg = average temperature in well column, Rankin

Chevron Confidential Use Only Page 214 of 244

 Version 2008.01

There are several ways to use Equation 12-10 to calculate the SBHP. One method uses

the following iterative procedure.

1) Calculate the average temperature by averaging the surface and reservoir

temperature.

2) Guess at a SBHP and calculate the average column pressure by averaging

the SBHP and WHP.

3) Using the average pressure, average temperature and gas SG, calculate the

average compressibility factor.

4) Use the calculated compressibility factor to calculate an improved SBHP

and new average pressure.

5) Iterate the calculations in steps 3 and 4 until the SBHP and average

compressibility factor converges. This generally occurs in 2 to 4

iterations.

Example – A 10,000 foot gas well has been shut in for 7 days. The surface pressure has

stabilized at 3750 psia. The surface temperature is 60 degrees Fahrenheit. The reservoir

temperature is 200 degrees Fahrenheit. The gas SG is 0.7 and is free of nitrogen, carbon

dioxide and hydrogen sulfide. Construct a table showing the convergence of the SBHP

assuming an initial guess of 6000 psia.

(60 + 200)
Tavg = + 460 = 590
2

Chevron Confidential Use Only Page 215 of 244

 Version 2008.01

Run SBHP WHP Pavg Tavg Zavg SBHP new

1 6000 3750 4875 590 1.088 4601
2 4601 3750 4175 590 1.060 4626
3 4626 3750 4188 590 1.061 4625
4 4625 3750 4188 590 1.061 4625
5 4625 3750 4188 590 1.061 4625

Another way to use Equation 12-10 is to divide the gas column into increments and then

calculate the SBHP by changing the temperature and gas gravity with each depth

increment. For the previous example, calculate the SBHP using 500 foot increments and

display in a table. Table 12-1 shows pressures calculated.

Depth Pressure Tdepth Tavg Zavg Calculated Pressure

0 3,750 520
500 3,796 527 523.5 1.035310 3,796
1,000 3,841 534 530.5 1.038078 3,841
1,500 3,887 541 537.5 1.040809 3,887
2,000 3,932 548 544.5 1.043503 3,932
2,500 3,977 555 551.5 1.046161 3,977
3,000 4,021 562 558.5 1.048782 4,021
3,500 4,066 569 565.5 1.051368 4,066
4,000 4,111 576 572.5 1.053919 4,111
4,500 4,155 583 579.5 1.056434 4,155
5,000 4,199 590 586.5 1.058915 4,199
5,500 4,243 597 593.5 1.061362 4,243
6,000 4,287 604 600.5 1.063775 4,287
6,500 4,330 611 607.5 1.066154 4,330
7,000 4,374 618 614.5 1.068500 4,374
7,500 4,417 625 621.5 1.070814 4,417
8,000 4,460 632 628.5 1.073095 4,460
8,500 4,503 639 635.5 1.075345 4,503
9,000 4,546 646 642.5 1.077563 4,546
9,500 4,589 653 649.5 1.079751 4,589
10,000 4,632 660 656.5 1.081908 4,632
Table 12-1 - Static Tubing Gradient Calculations

As shown by the Table 12-1, the SBHP is 4632 psia versus 4625 psia in the first method,

both within engineering accuracy of each other. An interesting observation that many

engineers do not recognize is that the gradient of the gas column gets lighter in the lower

Chevron Confidential Use Only Page 216 of 244

 Version 2008.01

portion of the wellbore. This is not initially intuitive because one assumes that as the

pressure increases, the gradient increases because more moles of gas are pushed into the

incremental depth. Just the opposite occurs because of the higher temperatures effect.

Looking at Table 12-1, one can calculate that at the top of the gas column the pressure

increases 46 psia for each 500 foot increment and at the bottom of the well the increment

is 43 psia.

A third method which is probably the best method is to calibrate Equation 12-10 from a

pressure gradient run. If a gradient run measures both the WHP and the SBHP, the gas

condition variables in Equation 12-10 can be treated as a single constant J. By

rearranging Equation 12-10, J can be calculated using Equation 12-11. Once J is known,

it can be used on other wells in the field assuming that the gas conditions are identical. J

should be updated with an occasional pressure gauge run as the reservoir is depleted.

0.01875 × SGgas × Depth

Tavg × Zavg
WHP = SBHP × e = SBHP × e J × Depth

⎛ SBHP ⎞
ln⎜ ⎟
⎝ WBHP ⎠
J=
Depth

Equation 12-11 - Equation of Gas Gradient Constant

In our example, assume that a pressure gauge was run in the well. The surface pressure

was 3750 psia. At a depth of 10000 feet, the pressure recorded was 4640 psia. Calculate

the J constant.

Chevron Confidential Use Only Page 217 of 244

 Version 2008.01

⎛ 4640 ⎞
ln⎜ ⎟
J= ⎝ 3750 ⎠
= 2.130 × 10 −5
10000

All three of these methods assume that no fluid level is found in the tubing or casing.

Fortunately for many wells which produce small amount of condensate or water, any

liquid falls back to bottom when the well is shut-in and is pushed back into the formation

as the gas pressure builds up in the tubing. There are some wells, especially low rate or

low pressure wells, which will have a fluid level in either the tubing or casing. This

introduces inaccuracy to the pressure data. The QRE should always confirm that wells

are clear of fluid before surface extrapolation methods are used to determine downhole

pressures.

12.3 Determining Average Reservoir Pressure

The most essential step in the preparation in constructing a P/Z plot is determining the

average reservoir pressure. In moderate to high permeability reservoirs, there is generally

good communication between wells such that the reservoir is acting “tank-like.” In these

reservoirs, P/Z and cumulative production should be constructed using average static

bottomhole pressures. If possible, the pressures should be measured during a field shut-

in such that a single cumulative production point can be used.

Chevron Confidential Use Only Page 218 of 244

 Version 2008.01

If there are permeability barriers in the field that can be identified geologically or from

the observation of the pressure data, the field or reservoir can be sub-divided. If the

pressures recorded from wells show significant differences across the reservoir, an annual

SBHP contour map can be constructed and used to calculate the average pressure.

Some times it is useful to construct a P/Z plot containing all of the data points. Each

should be plotted using its individual P/Z value and the cumulative production for the

field for the time when the pressure measurement was made. The points can then be

observed to determine how “tank-like” the field behaves. One can also investigate points

which deviate significantly from the field trend to eliminate data errors or pressures that

are less than static.

In low permeability reservoirs, obtaining a true static BHP is much more difficult because

the time required to reach static conditions is generally longer than is acceptable to

operations personnel. Thus any pressures recorded represent something less than a static

pressure. This can be demonstrated with an example using a well simulator.

Example – A 10,000 foot deep gas reservoir with a single well is producing with a

constant 1000 psia WBHP. The initial reservoir pressure is 5000 psia. The well has an

average drainage area of 675 acres. The average permeability for the 30 feet of net pay is

0.238 md. The well has been fracture treated and exhibits a skin factor of -4.1. The gas

contract requires that the well be annually shut in for 96 hours for a SBHP measurement.

The following table gives all of the reservoir parameters.

Chevron Confidential Use Only Page 219 of 244

 Version 2008.01

Data Value
Initial pressure 5000 psia
Reservoir Temperature 212 Degrees F
Net pay 30 feet
Porosity 10%
Water Saturation 40%
Permeability 0.238 md
Gas SG 0.65
Drainage Radius 3060 feet
Well Radius 0.3 feet
Skin Factor -4.1
Z initial 1.00915
OGIP 13.8 BCF

The initial production from the well was 2.5 MMCFPD. The 10 year production history

is show by Figure 12-3. As shown, the tight gas well has a considerable period of

transient flow.

Chevron Confidential Use Only Page 220 of 244

 Version 2008.01

Tight Gas Reservoir Example

10000
MCFPD

1000

100
0 1 2 3 4 5 6 7 8 9 10
Years

Figure 12-3 - Rate Time for Tight Gas Well Example

Figure 12-4 shows the output from the history match of the well simulator with the 96

hour shut-in periods shown for the WBHP. The top section of the output shows the rate

in MCFPD and the cumulative gas production. The bottom section shows the SBHP and

the WBHP from the simulator. The top line is the SBHP. The red line shows the WBHP

which is 1000 psia except for the initial 5000 psia pressure and the annual 96 hour shut-in

periods. One can quickly see the difficulty that will be encountered using the P/Z method

on this well as none of the shut-in pressures measure close to the simulator average

reservoir pressure.

Chevron Confidential Use Only Page 221 of 244

 Version 2008.01

Figure 12-4 - Well Simulator Results for Tight Gas Example

A closer look at the third shut-in pressure which occurs around 26,000 hours is shown by

Figure 12-5. At the end of the four day shut-in period, the well has increased in pressure

from 1000 psia to slightly over 3000 psia and the pressure is building at a rate of

approximately 4 psia per hour. Unfortunately, the SBHP for the reservoir at this time is

still over 4000 psia so a considerably longer shut-in period would be necessary to reach

the SBHP. Some form of pressure build-up analysis could be used also but most shut-in

data tests do not have the necessary level of detail and are plagued by after-flow effects.

Chevron Confidential Use Only Page 222 of 244

 Version 2008.01

Tight Gas Example - Shut-in Number 3

3500

3000

2500

2000
PSIA

1500

1000

500

0
26000 26050 26100 26150 26200 26250 26300 26350 26400 26450
Hours

Figure 12-5 - Pressure Shut-in Number 3

A construction of the P/Z curve for the tight gas well example using the 96 hours shut-in

pressures along with the known simulator SBHP is shown by Figure 12-6. One of the

first things noted is that the 96 hours pressure build-up pressures generate a P/Z plot that

is reasonably close at predicting the OGIP provided the first several points including the

initial point are excluded. The straight line is generated because of the uniform shut-in

periods. If some of the shut-in pressures had been 48 hours pressures, there probably

would have been much more noise in the data. Also note that back extrapolating the

curve gives an initial reservoir pressure around 4000 psia. Some engineers may

misinterpret the transient effect and flattening of the P/Z curve as water influx.

Chevron Confidential Use Only Page 223 of 244

 Version 2008.01

P/Z Versus Gas Cum Example

6000
SBHP/Z
675 Acre 0.24 md Single Gas Well PSTAR/Z
SBHP/Z LSQBF
5000 PSTAR/Z LSQBF

Well Simulator Reservoir Pressure

4000

96 Hour Annual SIBHP Tests

P/Z

3000

2000

1000

0
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000
Cum Gas MMCF

Figure 12-6 - Tight Gas Well P/Z Plot

12.4 Water Influx in Gas Reservoirs

The examples and equations so far are based on having complete volumetric depletion

because of gas compressibility. They assume that no additional reservoir energy is being

obtained from water influx, expansion of connate water or from the expansion of the

formation rock as depletion occurs. All of these conditions must be recognized and

corrected for where their effect is significant. Water influx can have the most effect

Chevron Confidential Use Only Page 224 of 244

 Version 2008.01

because large volumes of aquifer water can invade the gas pore space. The additional

reservoir energy obtained from connate water expansion or formation compressibility is

usually not significant. This is because both have compressibilities one to two orders of

magnitude smaller than the gas compressibility. Thus the effect of the gas

compressibility dominates the equations. The exception to this generalization is over-

pressured reservoirs, especially when they have initial pressures in excess of 10,000 psia.

Here the formation compressibility is approximately equal to the gas compressibility.

Ignoring the formation effect for over-pressured reservoirs may cause reserves to be

greatly over estimated.

Figure 12-7 shows the P/Z curves for a depletion reservoir, a steady-state water drive

reservoir and an unsteady-state water drive reservoir. It is difficult to generalize water

drive types because the many plumbing and PVT configurations can cause a multitude of

P/Z curves. The Craft and Hawkins9 “Applied Petroleum Engineering” text does a good

job showing the difference. The steady-state model shows its influence earlier than the

unsteady-state model. An example of a steady-state water drive might be a bottom water

drive where there is very short time period before the effects of the water influx occurs.

Unsteady-state water drives might be radial edge drives where it may take a considerable

time period before pressure effects are felt out into the aquifer because of the transient

flow period. We will work examples for both later in this chapter.

Chevron Confidential Use Only Page 225 of 244

 Version 2008.01

Typical Waterdrive Effects

6000
Volumetric Depletion
Steady State Waterdrive
Unsteady State Waterdrive
5000

4000
PSIA/Z

3000

2000

1000

0
0 100 200 300 400 500 600 700 800 900 1000
Gas Cumulative - MMCF

Figure 12-7 – Water Drives versus Volumetric Depletion

Equation 3-16 showed the basic material balance equation for gas depletion. This

equation, however, did not take into account any water influx or water production. For

simplicity we will assume that no condensate or oil is being produced from the reservoir.

OGIP × B gi = (OGIP − G p ) × B g

To keep the reservoir voidage in balance, we need to account for water influx and water

production using Equation 12-12 where

Chevron Confidential Use Only Page 226 of 244

 Version 2008.01

V = Vi − We + BwW p

Equation 12-12 - Material Balance by Volume

V Volume of reservoir space in ft3 that is in the reservoir and filled with gas at some

time of depletion,

Vi Initial volume of reservoir space filled with gas

We Volume of water influx in ft3,

Bw Water formation volume factor (units either rcf/bbl or rcf/scf),

Wp Water produced in either barrels or ft3 depending on Bw units.

The cumulative production in reservoir units is the initial volumes minus the current

volume. This can be stated using the ideal gas law equations as Equation 12-13.

p sc × G p pi × Vi p × (Vi − We + Bw × W p )
= −
Tsc zi× T ztime× T

Equation 12-13 - Gas Material Balance

This equation can be solved for the cumulative gas production as shown by Equation

12-14. The equation can also be solved for the pressure at any cumulative gas Equation

12-15.

Chevron Confidential Use Only Page 227 of 244

 Version 2008.01

Tsc ⎛ p i × Vi p × ( Vi − We + B w Wp ) ⎞
Gp = ×⎜ − ⎟
p sc ⎜⎝ z i × T z× T ⎟
⎠
Equation 12-14 – Solution for Cumulative Gas

⎛ Gp × p sc p i × Vi ⎞
T × z × ⎜⎜ − ⎟⎟
⎝ Tsc T × z i ⎠
p=
Vi − We + B w × Wp
Equation 12-15 - Solution for Pressure

One of the fundamental equations for calculating fluid volume changes is expansion

through the use of compressibilities. Table 3-1 gives some of the typical ranges

encountered for oil, water and gas fluids in the reservoir. The compressibility for water

ranges from 3 x 10-6 to 5 x 10-6 psi-1. The equation to calculate the compressibility of

water (or any other fluid) is Equation 12-16 using the change in volume for a known

change in pressure.

⎛ ΔV ⎞
⎜ ⎟
⎝ V ⎠ 1 ΔV
Cw = = ×
Δp V Δp

Equation 12-16 - Compressibility of a Fluid

Equation 12-17 rearranges and solves for the change in volume for a change in pressure.

ΔV = C w × V × Δp

Equation 12-17 - Volume Change due to Pressure Change

Chevron Confidential Use Only Page 228 of 244

 Version 2008.01

The use of this equation can best be demonstrated using an example. A water aquifer

connected to a gas field is depleted from 5000 psia to 4500 psia. The initial aquifer

volume was 100 million reservoir barrels. The water in the aquifer has a compressibility

of 4 x 10-6 psi-1. Ignoring formation compressibility, how much water will have to be

removed from the aquifer to lower the pressure?

ΔV = 4 × 10 −6 × 100 × 10 6 × (5000 − 4500) = 200,000 barrels

Sometimes it is necessary to use a total compressibility if the effects of formation, oil,

water and gas need to be accounted for. Equation 12-18 is the equation for total

compressibility. Ct is the total compressibility, sometimes called effective

compressibility Ce. Cf is the formation compressibility.

Ct = (Co × so ) + (C g × s g ) + (C w × s w ) + C f

Equation 12-18 - Total Compressibility

When calculating compressibility for water filled aquifers, Equation 12-18 becomes

Equation 12-19.

C t = (C w × 1.0 w ) + C f = C w + C f

Equation 12-19 - Total Compressibility for Water Filled Aquifer

Chevron Confidential Use Only Page 229 of 244

 Version 2008.01

Let’s investigate another example to better demonstrate the effects of water influx on P/Z

plots.

Example – A large gas field covering 10,000 acres has a GWC at -10,000 foot MSL. The

reservoir has an average of 50 foot of net pay. No wells have penetrated completely the

through the aquifer reservoir section below the GWC but it is believed that the aquifer

section could be as thick as 1,000 feet. There is good vertical permeability and the

aquifer will supply a bottom water drive which will contribute water influx with no time

lag as the gas field is depleted. No water influx from outside the productive area is

expected. The wells are completed horizontally near the top of the reservoir so no water

production is expected down to a pressure of 1,000 psia. Calculate the P/Z cumulative

curves assuming no water influx, water influx from a 500 foot thick aquifer and water

influx from a 1000 foot thick aquifer. Also calculate the movement of the GWC

movement. The following reservoir parameters exist for the gas reservoir and the

aquifer.

Parameter Gas Reservoir Aquifer Units

A 10,000 10,000 Acres
H 50 500 or 1000 Feet
Ф 0.2 0.2 Decimal
Sw 0.2 1 Decimal
GWC -10,000 NA MSL
T 660 660 Absolute
Tsc 520 520 Absolute
psc 14.73 14.73 psia

Chevron Confidential Use Only Page 230 of 244

 Version 2008.01

pi 5000 5000 psia

SGgas 0.7 NA Air = 1
Cw NA 2.9 x 10-6 Psia-1
Cf NA 3.0 x 10-6 Psia-1
Bw NA 1.02 SCF/RCF

Vi = 43,560 × 10,000 × 50 × 0.2 × (1 − 0.2) = 3.485 × 10 9 ft 3

Vaquifer (500 thick ) = 43,560 × 10,000 × 500 × 0.2 = 43.56 × 10 9 ft 3

Vaquifer (1000 thick ) = 43,560 × 10,000 × 1000 × 0.2 = 87.12 × 10 9 ft 3

The total compressibility in the aquifer including the formation is calculated assuming a

gas saturation of 1.0.

Ct = Cw × 1.0 + Cf = 5.9 × 10 −6 psia −1

Volumetric 500’ Aquifer 1000’ Aquifer

We We
Psia Z P/Z Gp (BCF) Gp (BCF) Gp (BCF)
(mmrcf) (mmrcf)

5000 0.990 5050 0.0 0.0 0.0 0.0 0.0

4800 0.975 4923 23.7 51.4 37.2 102.8 50.7
4600 0.960 4790 48.5 102.8 74.9 205.6 101.2
4400 0.946 4650 74.7 154.2 113.0 308.4 151.4
4200 0.933 4502 102.2 205.6 151.7 411.2 201.2
4000 0.920 4347 131.2 257.0 190.9 514.0 250.7
3800 0.908 4183 161.7 308.4 230.7 616.8 299.7
3600 0.898 4011 193.8 359.8 271.0 719.6 348.2
3400 0.888 3830 227.6 411.2 311.8 822.4 396.0
3200 0.879 3639 263.1 462.6 353.1 925.2 443.2
3000 0.872 3439 300.3 514.0 394.9 1,028.0 489.4
2800 0.867 3230 339.3 565.4 436.9 1,130.8 534.6
2600 0.863 3012 379.9 616.8 479.3 1,233.6 578.6
2400 0.861 2786 422.0 668.2 521.6 1,336.4 621.2
2200 0.862 2553 465.5 719.6 563.8 1,439.2 662.0
2000 0.864 2314 510.1 771.0 605.5 1,542.0 700.9
1800 0.869 2070 555.4 822.4 646.5 1,644.8 737.6

Chevron Confidential Use Only Page 231 of 244

 Version 2008.01

1600 0.877 1825 601.2 873.8 686.5 1,747.6 771.8

1400 0.886 1580 646.9 925.2 725.1 1,850.4 803.3
1200 0.898 1337 692.3 976.6 762.1 1,953.2 831.9
1000 0.911 1097 736.9 1,028.0 797.2 2,056.0 857.5

The aquifer constant U in RB/ΔPsi for the 500 foot thick aquifer is calculated as follows:

We 1028 * 10 6 * 5.6146
U 500 = = = 1,443,000 bbls / Δpsi
Pi = P 5000 − 1000

We 2056 * 10 6 * 5.6146
U1000 = = = 2,886,000 bbls / Δpsi
Pi = P 5000 − 1000

Figure 12-8 shows the P/Z plots for each of the three cases. Note the effects of the water

influx on the P/Z versus Gp curves.

Chevron Confidential Use Only Page 232 of 244

 Version 2008.01

Bottom Waterdrive Example

6000

Volumetric
500' Aquifer
5000
1000' Aquifer
Linear (Volumetric)

4000
PSIA/Z

3000

2000

1000

0
0 100 200 300 400 500 600 700 800 900 1,000
Gp - BCF

Figure 12-8 - Bottom Water Drive P/Z Plot

The GWC movement is calculated by dividing the water influx by the gas pore volume

on a per foot basis. The GWC movement for the 1000 foot thick aquifer is calculated as

follows. Note that for this example, no trapped gas or sweep efficiency are used in the

calculations. In practice, most GWC movement calculations would include trapped gas

and sweep effects, especially if the aquifer is an edge drive. A bottom water drive may

have less trapped gas because of gravity segregation effects.

We 2056 × 10 6
ΔGWC = = = 29.5 feet
43560 × A × Φ × (1 − S w ) 43560 × 10000 × 0.20 × (1 − 0.2)

Figure 12-9 shows the GWC movement for each of the pressure depletion points.

Chevron Confidential Use Only Page 233 of 244

 Version 2008.01

GWC Movement

50

45
Volumetric
40 500' Aquifer
1000' Aquifer

35
GWC Movement Feet

30

25

20

15

10

0
0 100 200 300 400 500 600 700 800 900 1,000
Gp - BCF

Figure 12-9 - GWC Movement Example

Another challenge with P/Z plots is to interpret them to determine if a water drive effect

is occurring. This can be difficult, especially in the early years of production. Looking at

Figure 12-8 with its depletion down to a P/Z of 1000 psia, one can easily see that water

influx is occurring for both the 500 foot and 1000 foot thick aquifer cases. However, if a

plot is made of just the first 200 BCF of production as shown in Figure 12-10, most

engineers could easily interpret each of the curves as linear, especially if the data contains

some measurement variation.

Chevron Confidential Use Only Page 234 of 244

 Version 2008.01

Bottom Waterdrive Example for Early Time

5500

Volumetric
500' Aquifer
5000 1000' Aquifer

4500
PSIA/Z

4000

3500

3000
0 20 40 60 80 100 120 140 160 180 200
Gp - BCF

Figure 12-10 - P/Z Example - Early Time

One technique that is frequently used because it is more sensitive to early data is the Cole

Plot. It plots cumulative gas production versus Bg × Gp ( Bg − Bgi ) on a linear

coordinate system. For volumetric depletion, the plot will be a horizontal line where the

ordinate value is the OGIP. For water influx situations, the line will have a slope which

if back extrapolated to the ordinate intercept will give the apparent OGIP if no water

influx is assumed. Figure 12-11 shows the Cole Plot for the water influx example. It is

easy to see how useful this plot is to determine if water influx is occurring. The Cole Plot

should be made for each P/Z analysis as a best practice.

Chevron Confidential Use Only Page 235 of 244

 Version 2008.01

Cole Plots

2,500
Volumetric
500' Aquifer
1000' Aquifer

2,000
Bg x Gp/(Bg-Bgi) .

1,500

1,000

500

0
0 100 200 300 400 500 600 700 800 900 1,000
Gp - BCF

Figure 12-11 - Cole Plot Example

The previous example was for a bottom water drive reservoir. With a bottom water drive

the time lag until water influx begins is rather short because the distance for the water

influx to travel is small.

An edge water drive, however, can take considerably longer to react to the lowering of

the reservoir pressure through production. This is because the aquifer begins producing

in unsteady-state transient flow and can remain in that mode for a considerable time

period, especially if the radius of the aquifer to the radius of the gas reservoir is large,

such as greater than 8.

Chevron Confidential Use Only Page 236 of 244

 Version 2008.01

Let’s look at an example from Dake’s Revised Reservoir Engineering Text7 (Page 490)

as it shows how a strong water drive can produce a P/Z curve which is very linear thus

masking the water influx effects. A gas reservoir has 823 BCF OGIP. The reservoir is

120 feet thick with permeability of 120 md. The aquifer radius is 8350 feet. The aquifer

has identical rock properties, 360 degrees of radial encroachment and a radius that is 10

times larger than the gas reservoir. The reservoir is being analyzed using the material

balance software GasWatTM.

Figure 12-12 shows the best fit analysis for the P/Z data. Note that the curve has a good

linear fit but that the OGIP calculated is 1225 BCF instead of the known 823 BCF.

Figure 12-13 shows the Cole Plot for the example with some of the early time transient

points removed. The increasing slope of the Cole Plot shows very early that water influx

is occurring but it still predicts the OGIP at 872 BCF which is within engineering

accuracy of the official 823 BCF OGIP.

Figure 12-14 shows the model prediction fit for the OGIP assuming unsteady-state radial

aquifer flow. It also has some of the early transient time pressure points removed. Note

that a very good fit is achieved. The OGIP is 872 BCF, identical to the Cole plot. Figure

12-15 shows the predicted water influx. Over 400 million barrels of water influx occurs

in the reservoir.

Chevron Confidential Use Only Page 237 of 244

 Version 2008.01

Figure 12-12 – Water Drive Example P/Z Best Fit Analysis

Figure 12-13 – Water Drive Example Cole Plot

Chevron Confidential Use Only Page 238 of 244

 Version 2008.01

Figure 12-14 – Water Drive Example P/Z Model Fit

Figure 12-15 – Water Drive Example Water Influx

Chevron Confidential Use Only Page 239 of 244

 Version 2008.01

The green dashed line in Figure 12-16 shows a P/Z prediction most engineers would

quickly recognize as characteristic of a water drive reservoir because the slope of the P/Z

curve changes with time. This prediction was made using the same reservoir conditions

and cumulative withdrawal volumes as before except that the time to reach each pressure

point was doubled. This has the effect of allowing more water influx to supplement the

gas depletion. Comparing Figure 12-14 and Figure 12-16 shows that the shape of a P/Z

curve in predicting water influx is very subjective. A very linear P/Z curve can exist with

a strong water drive. It is recommended that a full analysis including Cole Plots be

undertaken to ensure that no strong water influx is occurring before a material balance

volumetric reserves estimate is made on a gas reservoir.

Figure 12-16 – Water Drive Example P/Z with Reduced Withdrawal Rates

Chevron Confidential Use Only Page 240 of 244

 Version 2008.01

13 Technical References

# Author(s) Title Year

Published
1 Cronquist, Chapman Estimations and Classification of 2001
Reserves of Crude Oil, Natural Gas, and
Condensate – Society of Petroleum
Engineers
2 Petroleum Society Determination of Oil and Gas Reserves – 2004
Petroleum Society Monograph No. 1
Second Edition – Canadian Institute of
Mining, Metallurgy and Petroleum
3 Petroleum Society Determination of Oil and Gas Reserves – 1994
Petroleum Society Monograph No. 1 First
Edition – Canadian Institute of Mining,
Metallurgy and Petroleum
4 Towler, Brian F. Fundamental Principles of Reservoir 2002
Engineering – SPE Textbook Series Vol.
8
5 Thompson, Robert S., Oil Property Evaluation 1985
Wright, John D.
6 Dake, L. P. The Practice of Reservoir Engineering 2004
(Revised Edition)
7 Dake, L. P. The Practice of Reservoir Engineering 1994
(First Edition)
8 Slider, H. C. (Slip) Practical Petroleum Reservoir 1976
Engineering Methods
9 Craft, B. C., Hawkins, Applied Petroleum Engineering 1959
M. F.
10 Cobb, William M., Waterflooding 2005
Smith, James T,
11 Society of Petroleum Petroleum Engineering Handbook 1987
Engineers
12 Reep, G. J. REXDCA – Decline Curve Analysis 1990
13 Williams, D. B. Production and Injection Performance 2000
Analysis DCA and Specialty Curves
14 Reep, G. J. Waterflood Management Course 1991
Indonesia, Section VI.8 Decline Curves
15 The Texas Company Reservoir Engineering Handbook 1959
16 Willhite, G. Paul Waterflooding – SPE Textbook Series 1986
Vol. 3
17 Levorsen, A. I. Geology of Petroleum (Second Edition) 1967
18 Jensen, Jerry L., Lake, Statistics for Petroleum Engineers and 2003
Larry W., Corbett, Geoscientists (Second Edition)

Chevron Confidential Use Only Page 241 of 244

 Version 2008.01

Patrick W.M., Goggin,

David J.
19 Arps, J. J. Analysis of Decline Curves – AIME 1944-1945
20 Huffman, C. H., Probability Ranges for Reserves 1994
Thompson, R. S. Estimates from Decline Curve Analysis –
SPE 28333
21 Jochen, V. A., Spivey, Probabilistic Reserves Estimation Using 1996
J. P. Decline Curve Analysis with the
Bootstrap Method – SPE 36633
22 Mattar, L., Anderson, A systematic and Comprehensive 2003
D. M. Methodology for Advanced Analysis of
Production Data – SPE 84472
23 Blasingame, T. A., A Production-Based Method for Direct 2005
Rushing, J. A. Estimation of Gas-In-Place and Reserves
- SPE 98042
24 Spivey, J. P. A New Algorithm for Hyperbolic Decline 1986
Curve Fitting – SPE 15293
25 Brons, Folkert On the Use and Misuse of Production 1963
Decline Curves – Producers Monthly
26 Kewen, Li, Horne, A Decline Curve Analysis Model Based 2003
Roland N. on Fluid Flow Mechanisms – SPE 83470
27 Cummings, C. A., A Method for Predicting Solution Gas- 1986
Gentry, R. W. Drive Production Decline – SPE 15021
28 Gentry, Robert W. Decline-Curve Analysis – SPE 3356 1972
29 Fetkovich, M. J. Decline Curve Analysis Using Type 1980
Curves – SPE 4629
30 Thompson, R. S., The Error in Estimating Reserves Using 1987
Wright, J. D. Decline Curves – SPE 16295
31 Pletcher, J. L. Improvements to Reservoir Material- 2002
Balance Methods – SPE 75354
32 Harrell, D. R., Hodgin, Oil and Gas Reserves Estimates: 2004
J. E., Wagenhofer, T. – Recurring Mistakes and Errors – SPE
Ryder Scott Company 91069
33 Blasingame, T. A., Decline Curve Analysis for Variable 1991
McCray, T. L., Lee, W. Pressure Drop/Variable Flowrate Systems
J.
34 Kappa Engineering Advanced Decline Curve Analysis Course 2005
35 IHS Energy Group OilWat/GasWat Material Balance 2001
Technical Reference Manual
36 Vogel, J. V. Inflow Performance Relationships for 1968
Solution-Gas Drive Wells – SPE 1476-
PA
37 Jochen, V.A., Spivey, Probabilistic Reserves Estimation Using 1996
J.P., S.A. Holditch & Decline Curve Analysis with the
Associates, Inc. Bootstrap Method – SPE 36633-MS

Chevron Confidential Use Only Page 242 of 244

 Version 2008.01

14 Index

A G
allowables · 9 GasWat software · 237
Arps, J. J. · 9, 35 grouped wells decline · 169, 172
average reservoir pressure · 218

H
B
harmonic declines · 35, 78
Blasingame · 11, 34, 184 hyperbolic constant · 55
boundary conditions · 140 hyperbolic decline
minimum decline · 63
hyperbolic declines · 35, 54
C cubic solution · 70
Excel Solver method · 75
from aggregated wells · 179
Cole plot · 235
gas wells · 94
constant fluid compressibility · 85
Inital Rate and Decline method · 69
constant percentage exponential decline · 37
Three Point method · 68
constant pressure solution · 22
transient flow · 66
constant rate solution · 22
hyperbolic exponent · 55

D I
Darcy's law · 13
ideal gas law · 206
gas wells · 10, 94
infill drilling · 44
oil wells · 13, 82
DCA mapping · 150
decline scatter plot · 175
downhole gauges · 213 L
downtime · 45
Drilling Wedge Method · 153 least squares best fit · 135
WOR plots · 159 location plots · 152

E M
effective exponential declineA · 37 material balance
Excel Solver · 75 gas reservoirs · 206, 210, 226
exponential declines · 35 material balance equation · 13, 25, 95
MDT · 212

F
N
Fekete RTA software · 12, 188
Fetkovich · 10 nominal exponential decline · 37
fluid property ranges · 28

Chevron Confidential Use Only Page 243 of 244

 Version 2008.01

P T
P/Z plot · 210, 219 tight gas wells · 25, 220
probabilistic DCA · 198 Topaze software · 12, 185, 187, 188, 194, 196
production data accuracy · 143 transient flow · 66
Productivity Index - PI · 15 turbulent flow · 17

R U
radial diffusivity equation · 13, 20, 22, 33 upper boundary analysis · 47
radius of investigation · 21
rate cumulative plot · 44
recovery efficiency · 28 V
reserve errors · 63
reserves by well · 175
Vintage Well Analysis method · 153
Rule-of-thumb tables
Vogel PI tailoff · 15
current rate life · 50
rate versus cumulative slopes · 53
remaining life · 51
W
S water displacement · 34
water drives · 107
water influx
secondary product streams · 99
gas reservoirs · 224
shut in periods · 45
waterfloods · 107
skin factor - S · 33
WOR method · 111, 116
solution gas reservoirs · 99
Empirical method · 122
statistical analysis · 172
RAC guidelines · 132
superposition · 22
rate time forecast · 121
surface pressure recorders · 213
remaining life · 124
Time Step method · 125

Chevron Confidential Use Only Page 244 of 244