Vous êtes sur la page 1sur 54

Water Saturation Modelling

From petrophysics to volumes and


simulation

FORCE seminar 30-31 March 2004

Jan C. Rivenæs
J.I. Kristiansen, Tor Barkve, Morten Fejerskov,
Stephen Rodgers, Julio Marre, Geir Håskjold
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 2
Philosofical background

Philosephor Thales from Miletus ( ~ 500 B.C ):


Everything is water.

Petrophysiscist Julio Marre ( ~ 2000 A.C )


Much is water.

Hydro Boss Torgeir Kydland ( ~ 2000 A.C )


Too much is water!

T&K
Page: 3
In the beginning, there was water…

Migration of oil:
A drainage process

T&K
Page: 4
An integrated work process
CORE
ANALYSIS LAB

PETROPHYSICS CAN WE FIND


A common language? RESERVOIR
A common understanding?
A common model?

GEOLOGY

T&K
Page: 5
The Sw modelling 2003 project
JAN INGOLF KRISTIANSEN
CORE
ANALYSIS
GEIR HÅSKJOLD

STEPHEN RODGERS

PETROPHYSICS
RESERVOIR

MORTEN FEJERSKOV

TOR BARKVE

JULIO MARRE GEOLOGY

T&K
JAN CHR RIVENÆS Page: 6
Historic view on Sw modelling

 1935: Realisation of existence of capillary water


 Garrison – 1935
 Schilthuis – 1938
 1940: Methods for predicting water saturation
 Leverett – 1941
 Archie – 1942
 70’ties: The first reservoir simulator
 Beta II – 1975
 Eclipse - 1982
 80’ties: Geological modelling using computers
 Mapping of zone averages
 Integrate Sw function over zones thickness - 1985
??
 90’ties: 3D modelling and visualisation
 2003: Sw modelling is still a "black box"…

T&K
Page: 7
Main deliverables from Sw modelling

 STOOIP/GIP
 Function of bulk rock volume, NTG, PHI, Sw and PVT

 HCPV distribution
 In which segments are the oil/gas accumulated?

 Input to reservoir simulation


 Initial hydrocarbon distribution prior to production
 Integrated part of SCAL model

T&K
Page: 8
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 10
Water saturation - Sw
Fraction of water in the pore volume

Matrix (sand, silt)


Sw

Dry clay Sh = 1 - Sw
Clay bound water
Free Water
Free water Hydrocarbon

Sw = 1 Sw < 1
Sh = 0 Sh > 0
T&K
Page: 11
Vertical equilibrium

The initial fluid distribution is defined by vertical equilibrium


between capillary pressure and gravity forces.

Pc Depth

Sw Sw

T&K
Page: 12
FLUID DISTRIBUTION AND CONTACTS:
ROCK QUALITY DEPENDENCE OF WATER SATURATION- HEIGHT PROFILE

Height above FWL

k = 50 md

500 md

10 md

200 md

1D

k = 100 md

Water saturation

 Saturation depends on rock quality (average pore


throat radius) as well as height above the FWL.

T&K
Page: 13
Water saturation definitions
SWIR
Formation Pressure (FMP)

Drainage

FOL Imbibition

FWL

Depth
SWIR Sw
SWCR
T&K
Page: 15
Is OWC precise?

OWC

FWL

No, refer to fluid contacts based on pressures instead

Good sand: OWC  FWL


T&K
Poor sand: OWC FWL
Page: 16
FOL and appearant FWL for gas

GAS

FOL
OIL
FWLG

FWL
WATER

T&K
Page: 17
Wettability and capillary pressure
WATER
The contact angle can be used
to distinguish between a water wet,
OIL mixed wet, and oil wet system.
ROCK

Drainage: Non-wetting phase saturation is increasing.


Imbibition: Wetting phase saturation is increasing.

The wettability is a function of rock and fluid properties,


but also on history.

T&K
Page: 18
Sources for Sw data
Pressure Sw logs (CPI)
measurements

Sw

Production and
Routine core analysis production
Special core analysis logging
T&K
Page: 19
Sw and the other properties…
Porosity
Apparent, non-apparent, absolute, Permeability etc
effective, total, residual, Appearant, effective, relative,
intergranular, intermatrix, primary, Klinkenberg, air, horisontal,
secondary,... vertical, syntetic, well-test,
mobility, wettability
?

Scale
Core plugs, log scale, geomodel,
Net-to-gross simulation model, upscaling,
Cut-off, reservoir, non-reservoir modelling approach,…
net sand, net pay, Vshale,…

T&K
Page: 20
The petrophysical porosities

Matrix (sand, silt)

Dry clay
Clay bound water

Ft Free Water
Free water Fe Fc Hydrocarbon

Log analysis: Fe = PHIE and Ft = PHIT


Fc (low temperature humidity drying) ~ Fe
Fc (high temperature oven drying) ~ Ft
T&K
F Use Fe (PHIE) and proper Fc Page: 21
Should we use NTG and cut-off?
If you model effective
props (PHIE and Keff),
cut-off in a 3D model is
not needed

NTG should not be


modelled as a property…

However, for up-scaling


No contribution? Sw, post-model cutoff
may be needed in some
cases

No contribution?

T&K
Page: 22
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 23
Sw … An integrated work process

CORE
ANALYSIS LAB

COOPERATION
PETROPHYSIC
S EXTERNAL CONDITIONS
RESERVOIR
WORKFLOW

ORGANISATION

GEOLOGY

T&K
Page: 24
A question of scale…
Core scale

Log scale Geo model scale

T&K
Flow model scale
Page: 25
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 26
Workflow for petrophysical Sw-model

CORE
ANALYSIS LAB

CORE DATA:
k, w, Pc
PETROPHYSICS
RAW LOGS

FLOW ZONES
FMP RESERVOIR
CPI’s
Sw MODELS ZONATION
FLUID LEVELS FACIES

GEOLOGY

T&K
Page: 27
Traditional Leverett model building
Sw  aJ b
K g
Using core data: J H
  cos
Pc(Sw)+Fc+KJ-modeltest

J (Sw)-model

Using log data:


SWE+PHIE+KHLOG+HJ-modeltest
T&K
Page: 28
Sw-modelling preferences

T&K
Page: 29
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 30
Workflow for geological Sw-model

CORE
ANALYSIS LAB

PETROPHYSICS CORE DATA:


k, w RESERVOIR

CPI’s
Sw MODELS
FLUID LEVELS ZONATION & REGIONS
GEO MODEL Sw
VOLUMES

GEOLOGY

T&K
Page: 31
3 steps for the geomodeller

K
J  A H

where A  3.141
 w  o g
?  cos
S w  aJ b

Receive function(s) Implement in model Deliver products

Drainage part
T&K
Page: 32
Handling input from petrophysicists
Until now
Sw formulas on mystical ??
forms, in mystical units

SWE, SWT, PHIE, PHIT, PHIcore, Kair, Ksynth


(Use whatever you want)

2004 …
Sw formulas on standard !
forms, in NH standard units

SWE, PHIE, K

T&K
Page: 33
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 35
Work flow for the simulation Sw-model

CORE
ANALYSIS LAB

SCAL DATA:
PC AND REL.PERM
PETROPHYSICS
J FUNCTIONS RESERVOIR
FLUID LEVELS

GEO MODEL SW
VOLUMES

$$$$$

GEOLOGY

T&K
Page: 36
Upscaling - prone for confusion…

I can perhaps use your Talk together:


upscaled Sw, but only The way R.T. choose
partially! to incorporate Sw in
Eclipse may
influence the
RT upscaling

Partially ...? Don’t you like


me?
The upscaling is a
multidisciplinary task,
but in practice R.T. is
in charge!
GEO

T&K
Page: 37
The world as seen by a reservoir dog

 The geo model alone is not sufficient to


define all necessary Sw data for the
simulation model.

 The upscaled Geo Sw cannot be used


directly in the simulation model.

 Sw data related to mobility must be


taken from SCAL data.

 SCAL data are not always sufficiently


available.

 SCAL and Geo data may seem


T&K inconsistent. Page: 38
Requirements for the simulation Sw model

 The initial volumes should be (approximately) stable.


 Capillary pressure/gravity equilibrium.
 “Correct” dynamic response.
 Imbibition or drainage data?
 Initial volumes should match geo model.
 Drainage Pc data.

STABILITY
Use NOW
Give us LESS
Pc! SWCR SW!

T&K
Page: 39
Vertical equilibrium in the simulator
To achieve true vertical equilibrium:
Depth values must be taken at grid cell centers.

Pc

FWL

Cell refinement can be used in Eclipse initialization


but leads to an initially instable model.

ALWAYS CHECK INITIAL MODEL STABILITY.

T&K
Page: 40
Summary

 The reservoir challenge is the combination of


the SCAL data and the geo model data.
 The petrophysicist has already handled this
problem, but on a different scale.
 A Sw implementation in the simulation model
should preferably be based on a J function
approach.
 Sw modelling for reservoir is time consuming.
Plan for this!

T&K
Page: 43
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 44
Upscaling

 Best practice:
 Upscale porosity arithmetic, weighted on bulk cell
volume
 Compute PoreVolume for each cell
 Upscale Sw weighted on pore-volume, arithmetic
 Consider use of "cut-off" in Geomodel before
upscaling
 Challenge
 What input does the res.eng want?
– Sw
– Swir? (a rock parameter, ~independent of contacts)
– Swcr? (a rock parameter, ~independent of contacts)
 Be precise on delivery to R.T.

T&K
Page: 45
Some upscaling of Sw issues

 The contact problem


 Cell below or above contact?

 The resolution problem

 Cut-off (filtering) or not?

 Sampling near faults

 Core-plug to ressim cell problem

T&K
Page: 46
The contact problem
Cell properties refers to
center point; i.e. the whole
cell gets one value

Oil cell? FWL 100% water?

Large (thick) cells


may give problems

T&K
Page: 47
Geomodel: Integrating J over geocell
K
J H

S w  cJ b
c  a  Ab  const
 
H=dZ b
Sw  c H K

Sw  Sw (H )

   H dH
h2 h2
b
 S wdH  c K b

h1 h1
Breath … relax …
   
b 1 b 1
h1 b  h 2
The IPLIB routines do S w h  c K  
 b 1 
this for you!

 b  h1 
b 1 b 1
 h 2
Sw  c  K
 
T&K  (b  1)( h 2  h1)  Page: 48
The resolution problem (tiger problem)

T&K
Page: 50
Having geocell vertical grid 0.5 to 1m is close to Sw log resolution!
The “tiger” problem
??
Heterolithic coarse sand/silt Pc

1m
PHI=0.25, K=3000
Blocked/Upscaled
PHI=0.13, K=10

PHI=0.19, K=1500
Homogenous sandstone

PHI=0.19, K=1500 Same Sw(H) curve?


What about relperm start point?

Should we try to ”back-compute” PHINet SwNet, KNet and NTG (for Sw


calculations only?)
T&K
Page: 51
”Cut-off” (filter) in Sw upscaling
Sw in geo model is often correlated with the permeability k.

In a stochastical model, some grid cells may have low k, but high porosity.

This may lead to artificially high mobile water saturations in the


simulation model.

Cure: Set all cells with K < x and w<y inactive. I.e. a cut off before
upscaling (On Oseberg Øst, x=1mD, y=0.05 p.u.)

This is espacilly important if some kind of end-point scaling is used, but


should also be considered elsewhere

k=0.01 mD
w= 0.2 k=67 mD
Sw = 1
w= 0.2
k=100 mD
w= 0.2 Sw = 0.33
Sw = 0.1
T&K
Page: 52
Water sat upscaling: Sampling crucial!
Down Up Down Up

T&K GeoModel SimModel Page: 53


Quality control on upscaling effects

GEO MODEL SIM MODEL

Ln(SwD)

T&K
Page: 54
The quest for the correct Pc…
How should I define the Pc curve in the simulator to reproduce
the upscaled Geo Sw distribution SWGEO?

How should lithological variation in Pc data be described on the


simulation model scale?

K, w, Sw
J K, w, Sw
Pc?

?
J curves are scale dependent.
Eclipse takes simple form of J curves only. Core plug

T&K
Page: 56
Pc selection for the simulation model
CHARYBDIS
CORE
ANALYSIS LAB
SCYLLA

CAPILLARY PRESSURES

SCAL DATA:
PETROPHYSICS CAPILLARY PRESSURE DATA
J FUNCTIONS RESERVOIR
CORE SCALE
PETROPHYSICAL MODEL:
J-CURVES DRAINAGE OR IMBIBITION?

GEO-MODEL SCALE NO LITHOLOGICAL


GROUPING ON SIM.SCALE
MAY HAVE TO BE SIMPLIFIED

PRIMARY DRAINAGE DATAGEOLOGY

NO LITHOLOGICAL
T&K
GROUPING ON SIM.SCALE Page: 58
Topics

 Introduction

 Definitions

 Overall workflow

 Sw in Petrophysics

 Sw in Geology modelling

 Sw in Reservoir Simulation

 Upscaling issues

 Summary

T&K
Page: 59
Summary (remember this)
 Use effective porosities (PHIE)
 Avoid net-to-gross as a property
 Use FWL instead of OWC
 Use appearant FWL for gas above oil
 Use J-functions (and start simple)
 J-functions should be derived at the proper scale
 … Since permeability is so scale-dependent
 Treat 3D blocks near FWL correctly
 Consider use of appearant properties
 RT have the most difficult task
 Sw modelling is truly interdisciplinary…

 NH work: Best practice document

 Yet, there are many unresolved issues…


T&K
Page: 60
Philosofical epilougue

Petrophysicist Jan Ingolf


Sw means ”Some work”

Geologist Jan
Sw means “Simple wrap”

Res.eng. Tor
Sw means “Sucking worries”

T&K
Page: 61
Thank you for participating

Feedback from you are welcome

T&K
Page: 62

Vous aimerez peut-être aussi