You are on page 1of 73

Modelos de Simulación

Material Balance and


Production Behavior

Ing. Nelson Cabrera Maráz, Msc


roneven@cotas.com.bo
N@Plus 2019
Water-Drive Gas
Reservoirs:
Material Balance and
Production Behavior

N@Plus 2019
2
Goals

 Identify the factors that control rate and reserves


in Water-Drive Gas Reservoirs
 Introduce the methods used to estimate reserves
and predict behavior for these systems
 Provide resources to locate additional solutions
or data if needed

N@Plus 2019
3
Motivation

 Water drive gas reservoirs are affected by many


factors
 permeability, net-to-gross, residual saturations,
aquifer size, rock compressibility, well rates, tubing
strings…
 material balance methods are useful for surveillance,
for estimating reserves, and prior to simulation

N@Plus 2019
4
Challenges
 Compressibility and water drive effects can be
confused
 The recovery efficiency in water drive reservoirs
depends on
 reservoir properties, including k and Sgr
 aquifer properties, including k and size
 well placement
 well rate

N@Plus 2019
5
Approach
 Derive material balance expressions that include
important mechanisms
 Estimate gas in place and reserves without water
encroachment
 Examine effects of water encroachment on
reserves and well behavior
 Briefly discuss how all these components can be
coupled

N@Plus 2019
6
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
7
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
8
The Material Balance Equations

 Material Balance Concept


 Basic Equations
 Physical Interpretations
 Drive Indices
 Dry gas
 Havlena-Odeh Forms

N@Plus 2019
9
Material Balance Concept

With all quantities in reservoir volumes


Production = Expansion of fluids
+ Influx into reservoir
+ Injection
+ Pore volume reduction
or
F = E + Influx; F = net withdrawal; E =
reservoir energy or “expansions”.
N@Plus 2019
10
Material Balance Terms : F

Production - Injection = Net(Gas + Water + Oil)


Neglect oil production for this course. No injection. Then
Production = F = GpBg + WpBw [rb]
= the cumulative reservoir volume produced.

Gp = cumulative gas produced, Mscf


Bg = gas volume factor, rb/Mscf
Wp = cumulative water produces, stb
Bw = water volume factor, rb/stb (sometimes = 1)

N@Plus 2019
11
Short Course Outline

 The Material Balance Equations


 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
12
Drive Mechanisms

 Gas expansion
 treated via the Bg - Bgi term in material balance
 Evolution of gas from brine
 Pore compressibility
 Shale dewatering
 Water influx

N@Plus 2019
13
Expansion of Gas

Initial standard volume of gas is G


Its initial volume was GBgi
Its current volume is GBg (includes the gas
produced)
Thus the energy due to gas expansion is
Eg = G(Bg-Bgi)
Bg always evaluated at average reservoir pressure!!!

N@Plus 2019
14
Expansion of Water

Reservoir pore volume is GBgi /(1-Swc)


Initial reservoir volume of water is GBgi Swc /(1-Swc)
Usually write expansion using water compressibility
rather than volume factors,
Ew = cw(pi-p)GBgi Swc /(1-Swc)
Always in terms of average reservoir pressure!!!

N@Plus 2019
15
Reduction in Pore Volume

As noted for water expansion term, reservoir pore


volume is GBgi /(1-Swc)
Usually write pore volume reduction using pore
compressibility rather than volume factors,
Ef = cf (pi-p)GBgi /(1-Swc)
Always in terms of average reservoir pressure!!!
Often combine with the water expansion
Efw = (cf + Swc cw)(pi-p)GBgi /(1-Swc)

N@Plus 2019
16
Water Influx Term
 Always expressed in reservoir barrels
 We will discuss how to calculate later
 Just write as We for now.

N@Plus 2019
17
Material Balance Equation - General Form

 S wc cw  c f We 
G p Bg  W p Bw  G Bg  Bgi    pi  p Bgi  
 1  S wc G

•Usually know B’s, Gp, Wp, cw, pi


•Average pressure estimated from surveys
•sometimes problematic
•We must be estimated - function of time, not constant
•cf is often unknown
•Wish to estimate G
N@Plus 2019
18
Drive Indices

 S wc cw  c f 
GBg  Bgi  G 1  S  pi  p Bgi   We  Wp Bw 
1  wc 
G p Bg
•Each term in brackets is a drive, scaled on the gas voidage.
•First term is gas expansion, second is water-pore collapse
drive, and the third is water drive.
•They must sum to one.
•Useful indices to compare reservoirs.

N@Plus 2019
19
Drive Indices -- How much gas comes from each source.

G Bg  Bgi 
Gas Expansion Index
G p Bg

S wc cw  c f
G  pi  p Bgi
1  S wc “Compressibility” Drive
G p Bg
• Only water drive is not
We  W p Bw proportional to G.
Water Drive • Compressibility effective only for
G p Bg large Dp.
N@Plus 2019
20
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
21
“Classic” p over z Equation

 For zero influx, and rock/water compressibility


G p Bg  G Bg  Bgi ; Bg 
psc zT
pz scTsc
psc zT  psc zT psc ziTi 
Gp  G   ; T  Ti
pz scTsc  pz scTsc pi z scTsc 
z  z zi  pi pi p
 
G p  G  ; G p  G  G
p  p pi  zi zi z
p pi  G p 
 1  
z zi  G 
N@Plus 2019
22
Classic p/z
p pi  G p 
 1  
z zi  G 
 Plot p/z versus Gp
 For volumetric reservoir,
straight line with x-
intercept of G

N@Plus 2019
23
Modified p/z
p  S wc cw  c f  pi  G p 
1   pi  p   1  
z 1  S wc  zi  G 
 Similar to modified
Cole Plot
 Must know cf
 Deviation now
caused by We only

N@Plus 2019
24
Modified p/z
Example of Modified p/z
7000

6500

6000
Modified p/z

5500

5000

4500 Modified
Classic
4000 Fit-Modified

3500

3000
0 5000 10000 15000 20000 25000 30000 35000 40000
Gp
N@Plus 2019
25
Comments on Modified p/z
 Estimate is 79 BCF
Modified p/z
 Classic p/z overestimates 7000

by > 10% -- 90 BCF 6500

6000

 Using only half the data, 5500

Modified p/z
classic p/z estimates 110 5000

4500 Modified

BCF -- almost 40% error! 4000


Classic
Fit-Modified

3500

3000
0 5000 10000 15000 20000 25000 30000 35000 40000
Gp

•Used an estimate of cf = 10E-6

N@Plus 2019
26
Models Without Aquifers
 Use p/z if water and rock compressibility are
negligible
 Use modified p/z if rock and water
compressibility are significant and known
 Use Roach plot or regression if rock and water
compressibility may be significant and are
unknown.

N@Plus 2019
27
Shale Dewatering as a Drive Force
 Low net-to-gross reservoirs may have a
significant water-filled pore volume interleaved
with reservoir rock
 At production timescales, pressure change may
diffuse into these rocks.

N@Plus 2019
28
Shale Dewatering as a Drive Force
 The time for pressure to diffuse into the shales is
approximately

 k h , ss kv , sh 
t  t adj  
   c A   c x2 
 ss g t , ss sh w t , sh sh 

where t adj is the time for the sandstone to adjust  t pss


A is the well drainage area
xsh is the thickness of the shales
N@Plus 2019
29
Shale Dewatering as a Drive Force
 The amount of water expelled is
 GBgi hgross  hnet 
Vsh shct ,sh ( p  pi )    shct , sh ( p  pi )
 ss 1  S wc  hnet 
GBgi  hgross  hnet   sh
   ct , sh ( p  pi )
1  S wc   hnet  ss
 GEsh
 This mechanism is simple to add to the Material
Balance
N@Plus 2019
30
Shale Dewatering as a Drive Force
 Or, an overall compressibility term is
GBgi  hgross  hnet  sh 
E fw, sh   S wc cw  c f  ct , sh ( p  pi )
1  S wc   hnet  ss 

E fw, sh 
GBgi
S c  c f  csh ( p  pi )
1  S wc  wc w

hgross  hnet  sh
csh  ct , sh
hnet  ss
 The shale effect can be comparable to cf
N@Plus 2019
31
Shale Dewatering
 Can be analyzed in same ways as cfw
 Cannot be distinguished from cfw
 If neglected, may cause errors
 overestimate compressibility
 anticipate well failures unnecessarily
 underestimate permeability
 overestimate aquifer size
 complicates facilities planning

N@Plus 2019
32
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
33
Material Balances with Aquifers
 Diagnostic Plots
 Aquifer Models
 Coupling Aquifers to Reservoirs

N@Plus 2019
34
Diagnostic: Modified Cole Form

F We
G where F  G p Bg  W p Bw
Et Et
S wc cw  c f
and Et  Bg  Bgi    pi  p Bgi
1  S wc

Plot F/Et versus Gp


• L is horizontal if We = 0
• Different aquifers types have different shapes
N@Plus 2019
35
Modified Cole Plot

 Excellent way to detect


water influx
 Gives some idea of
aquifer strength
 Not a good method for
estimating G in
presence of We
 “Modified” to include
rock & water
After Pletcher, SPE 62882, 2000 compressibility
N@Plus 2019
36
Outline
 The Material Balance Equations
 Drive Mechanisms in the
 Models for Water Influx
 Reserves Estimation
 The Material Balance as a Reservoir Simulator
 Production Strategies

N@Plus 2019
37
Models for Water Influx
 All models predict water influx based on the
history of average reservoir pressure
 lower reservoir pressure causes greater influx
 higher permeability has faster and greater influx
 larger aquifers have greater influx
 Many models available

N@Plus 2019
38
Mathematical Models for Aquifers
 Hurst and van Everdingen
 quite rigorous, but difficult to use because it requires
superposition
 Carter and Tracy
 somewhat simpler solution, but superposition still
needed
 Fetkovitch
 simple to use and often adequate
 Pot aquifer
 very simple; okay for small, high k aquifers

N@Plus 2019
39
Computing We with the Fetkovich Model

 Must specify aquifer properties (can adjust to fit)


 k, h, L or re and ro, , ct, 
 Fetkovich assumed a very simple exponential
relationship to predict influx
 has correct shape
 conserves mass
 accurate enough, given uncertain assumptions

N@Plus 2019
40
Computing We with the Fetkovich Model

 Should not use Fetkovich when


 aquifer diffusivity is very low

k

c
 aquifer is very large

N@Plus 2019
41
Equations
 The total amount of water that could encroach is
 A influx index is similar to a well PI,
Wei  pi ct Ah
 x
0.00708 fkh
J for radial flow; f  aquifer angle/360 0 ; ro  aquifer radius
  re  3 
 ln    
  ro  4 
0.003351khW
J for linear flow; W  aquifer wi dth, L  length
L
Both formulas are in field units.
N@Plus 2019
42
Equations (continued)
 The amount that encroaches in a single timestep
Dt is

  Jpi Dt 
DWe t  Dt  
Wei
 paq (t )  p(t  Dt )1  exp 

pi
Finally, aquifer pressure from   Wei 

 n 1 
  DWe 
paq (t  Dt )  pi 1  
j 1

Wei 
 
 
N@Plus 2019
43
Computing Procedure
 Choose a fixed timestep if possible
 Compute J, Wei, and other constants
 Beginning with the second step, compute DWei
using the current aquifer pressure and previous
step’s reservoir pressure.
 Compute the cumulative We
 Update the aquifer pressure
 Repeat Finally, aquifer pressure from

N@Plus 2019
44
Inferring Reserves and Aquifer Models

 Use a modified Cole plot to assess possibility of


water drive
 If possible, get best estimates of aquifer
geometry and properties from maps, seismic, or
analogs
 Then begin the iterative process of plotting,
adjusting, and replotting to refine model

N@Plus 2019
45
History-Match or Havlena-Odeh Form of
Material Balance Eqn
 The Havlena-Odeh “trick” is to divide by the
known drive energy, Eg

F We
G A plot of F/Et versus We/Et will have
Et Et
•intercept equal to the gas in place
F  G p Bg  W p Bw
•slope of 1 if the aquifer size is correct
S wc cw  c f
Et  Bg  Bgi    pi  p Bgi
 1  Scwc (or assume = 0)
Must know cf and w

N@Plus 2019
46
Adjust Aquifer Parameters to Fit

N@Plus 2019
47
Example Data Set
450000

400000

G = 202 BCF
Production Terms

350000

300000
y = 0.989x + 202329
R2 = 0.957

250000

200000
0 50000 100000 150000 200000 250000
Encroachment Terms

N@Plus 2019
48
25,000.0
Using Regression
Cumulative Gas, MMscf 20,000.0

15,000.0

10,000.0

5,000.0
G = 200 BCF

0.0
0 500 1000 1500 2000 2500
Time, days

 Compute Gp from the model and minimize our


error of that prediction.
N@Plus 2019
49
Drive Index Chart
1

0.9 Water-Rock
0.8 Gas
Influx
0.7
Drive Index

0.6

0.5

0.4

0.3

0.2

0.1

0
0 5000 10000 15000 20000 25000
Cumulative Production, MMscf

 Indices sum to 1
 Water increases in importance with time
 Water-Rock Compressibility is small
N@Plus 2019
50
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
51
Gas Trapping
 As water advances, gas is trapped as residual
phase
 Amount trapped depends on
 Sgr
 p
 We Gtrapped  We S gr Bg

N@Plus 2019
52
Residual Gas Saturation
 Not subject to much control
 some variation with rate of saturation change
 important for big blowdown projects like Brent
 Commonly estimated using Land’s formula
 Estimates typically range from 20 to 50%
 Must be obtained from an imbibition curve
 not drainage, like mercury capillary pressure

N@Plus 2019
53
Trapped Gas Pressure
 Can be lowered by depleting the reservoir more
rapidly
 Causes p to be lower as gas is trapped

N@Plus 2019
54
Trapped Gas Bulk Volume
 The volume trapped is closely related to We
 We increases with time for any level of pressure
change
 By decreasing depletion time, we can decrease
influx and trapped gas
 Produce at the highest possible rate!

N@Plus 2019
55
Water Drive Changes p/z Curve

N@Plus 2019
56
Effect of Rate and Aquifer Size

After Al Hashim and Bass, SPE 13233


N@Plus 2019
57
Modified p/z for influx

 p  p z i  G p ,aband 
1  
  
 z  aband  E p S gr 
 1  E p 
G 

 S gi 
E p  volumetri c sweep efficiency
aband  at abandonmen t
 Denominator adjusts for trapped gas at paband
 This gives the equation of the cutoff line.
After Agarwal et al., JPT, 1965
N@Plus 2019
58
Effect of Sweep Efficiency
 Field abandoned at
intersection of trend
and cutoff curves
 Remaining gas is
trapped gas at Sgr
 Dependence on
sweep Ep is not very
strong

After Agarwal et al., JPT, 1965

N@Plus 2019
59
Effect of Residual Gas
 Field abandoned at
intersection of trend
and cutoff curves
 Additional gas
trapped in bypassed
areas,1 - Ep
 Dependence on
saturation Sgr is
rather strong
After Agarwal et al., JPT, 1965

N@Plus 2019
60
Water Drive Material Balance
 Agarwal et al. assumed trapping pressure was
constant
 Hower and Jones (1991) included varying
trapping pressure
 Use of aquifer models, gas PVT, and rate
dependency make simulation a reasonable
option

N@Plus 2019
61
Outline
 The Material Balance Equations
 Drive Mechanisms
 Material Balance without Influx
 Material Balance with Influx
 Gas Trapping in Water Drive Reservoirs
 Production Strategies

N@Plus 2019
62
Production Strategies - High Q
 Aquifer requires time to react
 Rapid depletion “outruns” the aquifer
 Lowering pressure rapidly decreases the mass of
gas in Sgr
 But may not be able to produce fast enough
 sanding, skins, low permeability, no compression, too
small tubing
 Gp can be increased by up to 1/(1-Sgr/Sg)

N@Plus 2019
63
Production Strategies - Lift
 Artificial lift can be used to keep gas wells
flowing at high water rates
 continues gas production
 High water rates may lower reservoir pressure
 “overpower” the aquifer
 lower pressure increases Sg, mobilizing gas

N@Plus 2019
64
Production Strategies -- Coproduction

 May not be able to “outrun” aquifer


 Extremely high water rates from watered out or
aquifer wells can reduce reservoir pressure, rate
of encroachment, and mass of trapped gas
 Continue to produce gas updip
 May increase recovery up to one-third or so.

After Acaro and Bassiouni, 1987


N@Plus 2019
65
Production Strategies -- “Secondary Recovery”

 Idea: increase gas saturation of watered-out


reservoir by lowering reservoir pressure.
Mobilized gas will flow to wells.
 Similar to previous cases, but gas production has
(nearly) ceased.

N@Plus 2019
66
Production Strategies -- “Secondary Recovery”

Requirements
 Relatively high pressure and Sgr

 more trapped gas


 Inexpensive lift
 Inexpensive water disposal
 Favorable relative permeability curves

N@Plus 2019
67
Production Strategies -- Coning
 Gas is much more mobile than water
 Water is much denser than gas
 Gas wells often “benefit” from reverse coning
 The main focus should usually be on rate rather
than on limiting water production
 Wells may “lose column” rather than “load up”

N@Plus 2019
68
Getting a Time Scale in Material Balances

 We can relate pavg to Gp using Mat Bal


 We can relate qg to pavg using inflow
performance curves (if pwf is known)
 We can estimate time from Dt = DGp/qg
 We can even include the effects of tubing
pressure drops if we really want to easy to get
matches without XY plots

N@Plus 2019
69
Getting a Time Scale in Material Balances
Pressure History

8000
7000 Tubing head pressure

6000 Bottomhole pressure


Pressure, psia

5000 Average Reservoir Pressure

4000
3000
2000
1000
0
0 500 1000 1500 2000 2500 3000
Time, days

Example of coupled material balance, inflow, and tubing curves.


N@Plus 2019
70
Conclusions
 Mechanisms for water drive reservoirs include
water influx, compaction, shale dewatering, and
gas expansion.
 These reservoirs can be analyzed using material
balance and inflow performance relationships as
well as reservoir simulation.
 Material balance is an essential element in
reservoir modeling.

N@Plus 2019
71
Conclusions, continued
 Production strategies for water-drive gas
reservoirs should consider trapped gas, water lift
capacities, and water coning.

N@Plus 2019
72

N@Plus 2019
73

You might also like