You are on page 1of 41

SDR TRAINING COURSE 2013

History Matching Workflow


Sophie Verdiere, Lisette Quettier

© Laurent Pascal/Total
AGENDA

Definition and a business case

History Match workflow

HM criteria and uncertain parameters

General Strategy for History Matching

And the adapted tools


To be used during the training

SDR advanced 2013 - History matching -2-


HISTORY MATCHING DEFINITION

●3
According to you?

Why is so difficult?

• Adjustment of uncertain parameters in the reservoir model(s) until the


historical data (the observations) are matched with the simulated
What is HM? ones
• Observations: Pressure, RFT, PLT, 4D, etc…
• These adjustments have to be geologically consistent .
• Impose the observed rate (oil/gas/voidage) = history and try to match
the observed pressures, Wcut and GOR, salinities, PLT .

Iterative process
• solving an optimisation problem allowing to constrain the “static”
(trial and error or more model of the reservoir with “dynamic” observations. The solution is not
sophisticated) unique

• Classically seek one matched model


1 versus
• Nowadays could be several matched models, in order to assess
multirealization uncertainty
SDR advanced 2013 - History matching -3-
HISTORY MATCHING DEFINITION

According to you?
●4

Why is so difficult?

• Adjustment of uncertain parameters in the reservoir model(s) until the


simulated data match the historical ones (the observations)
What is HM? • Observations: Pressure, RFT, PLT, 4D, etc…
• These adjustments have to be geologically consistent .
• Impose the observed rate (oil/gas/voidage) = history and try to match
the observed pressures, Wcut and GOR, salinities, PLT .

Iterative process
• solving an optimisation problem allowing to constrain the “static”
(trial and error or more model of the reservoir with “dynamic” observations. The solution is not
sophisticated) unique

• Classically seek one matched model


1 versus
• Nowadays could be several matched models, in order to assess
multirealization uncertainty
SDR advanced 2013 - History matching -4-
Jafra/ Dynamic study: before and after HM
Well Jafra-102 Initial Model Current Model Observed data

Oil Rate (WOPR) Water cut (WWCT)

GOR (WGOR) Static Pressure (often WBP9)

By convention, simulation in line and measured data in markers


SDR advanced 2013 - History matching -5-
EXAMPLE - HM WORKFLOW EXAMPLE
K1-A EXAMPLE - 2007 STUDY
● Dry gas field - 4 Development
wells
● 1 Self killed well K1-A3
● Match the K1-A field with the
experimental design methodology
In pressure
Water breakthrough of K1-A3
● Study the interest of a new well
By evaluating the water risk
By considering the different
scenarios of GIIP in south (panels
7-8) and depletion (honouring the Possible infill Locations
match) Existing Development wells
Explo wells
Development Well shut-in

SDR advanced 2013 - History matching -6-


HM WORKFLOW EXAMPLE - K1-A EXAMPLE - 2007
1
K1A2 2

K1A4
4
● 31 parameters identified to impact the K1A1 3

9 5
history match (perm, pore volume, etc..) 4bis K1A3
6
10
- Ranges are defined for each parameters 7
8

- The most important parameters are


identified using an experimental design
approach (in house ‘EST’ methodology)
- No acceptable match possible with the
geological model !
● Update of the geological model
● New screening with EST showing
possible matches
● Assisted history match with EST

Experimental Design in EST


Thursday
SDR advanced 2013 - History matching -7-
IDENTIFICATION OF RELEVANT
PARAMETERS WITH EXPERIMENTAL K1A2
1
2

DESIGN K1A4

K1A1 3 4

●8 4bis
9 5

K1A3
6
10
7
8

Each blue line shows a run


with a combination of
“match parameters “ most influent parameters
are identified with a Pareto
plot: only a few have a
significant impact

SDR advanced 2013 - History matching -8-


ASSISTED HISTORY MATCH WITH “EXPERIMENTAL DESIGN”
USING ONLY THE 9 MOST RELEVANT PARAMETERS

EST AHM : Thursday

SDR advanced 2013 - History matching -9-


PRESSURE MATCHING WITH DIFFERENT SCENARII

● 10 K1A2
1
2

K1A4

3 4
K1A1

9 5
4bis K1A3
6
10
7
8

Satisfactory history match


K1A1
is achieved with 5 different K1A2
combination of parameters !
There is no unique solution
K1A3 K1A4

panel region in Gm3


2 Total
Nom run 1 3 4 6 7 8 9 GIIP
5 South zone (SMm3)
RUN_BASE
4 527 1 750 880 618 627 311 482 9 195
not matched
CALAGE1 4 167 3 879 472 716 1 339 655 482 11 711
CALAGE2 3 967 3 879 1 293 677 637 321 482 11 257
CALAGE3 4 127 3 879 472 689 1 429 700 482 11 779
CALAGE4
SDR advanced 2013 - 3History
649 matching
3 879 2 076 697 259 128 482 11 171 - 10 -
CALAGE5 3 887 3 765 1 385 683 610 301 482 11 113
HM: BASIC SCHEMATIC WORKFLOW - SUMMARY
FIRST GEOLOGICAL Model DYNAMIC SYNTHESIS

GEOLOGICAL MODEL UPDATE


POSSIBLY DST MATCH PRIOR
OR
INITIAL RUN(S) TO QC NEW REALISATION

SIMULATION RUNS MODIFICATION OF


PRESSURE then SATURATION PARAMETERS
YES

NO
NO
Sufficient match in
“static” pressure?

YES Geological
modifications?
Sufficient match in
saturation? YES
NO
YES

Sufficient match in
FORECAST RUNS
well productivity?

SDR advanced 2013 - History matching


HM is an integrated process - 11 -
Main difficulties in history matching

long and “painful” task

incomplete reservoir picture made from local measurements

observations reliability

parameters interaction

Integrated task

Bias due to the numerical model (grid orientation, numerical dispersion,


cell size ….)
• there are assisted HM tools but no “push-button” techniques
In spite of its difficulties, HM becomes exciting
if the methodology is respected
and if it is performed in collaboration with all the actors

SDR advanced 2013 - History matching - 12 -


THE STEPS OF HM WORKFLOW J : Jafra exemple
• Mandatory for understand the limitations and problems
0 – Previous studies (if any) encoutered in the past J

Before Simulation
I - Gathering and checking • Rates, P, salinity, DSTs
production data • QC, consistency

• Identifying the production mechanisms


II - Dynamic synthesis • Flow units and regions J
• Initial guess at volumes, aquifer parameters

• In place
III- Geological building • Parametrization of geological uncertainties J
IV - Base case, Defining • List of uncertain parameters
criteria and possible • Base case building J
workflows • Match quality
V- DST Matching • An opportunity to QC (between Step II and IV)

VI - “Reservoir Pressure • RFT data, static (shut-in) BHP


matching” Energy • The model acts as a material balance problem J
VII - “Saturation matching” • Act on the parameters of the transport equation.
Transport • Water cut and GOR, contact levels, PLT data J

VIII- Productivity matching • BHP, THP and network pressures


● 1
3

SDR advanced 2013 - History matching - 13 -


STEP I - GATHERING OBSERVATIONS: LIST ALL
AVAILABLE DATA

pressure measurements

• RFT, DSTs, flowing BHP, THP, networks P

saturation measurements

• TDT/RST logs, 4D seismic

production data

• oil, water and gas rates, BT dates, PLT, salinities, tracers

schedule of completions

• work-overs, shut-in, problems (cementation, plug …)

This step is usually performed at the affiliate


SDR advanced 2013 - History matching - 14 -
STEP I - PREPARING THE PRODUCTION DATA

• well position in the grid


• measurement frequency
• measurement conditions (P,T)
Quality Check • allocation (per well, layers)
• well operation data
• Consistency (ex: THP and GOR)

• Smooth on 1 month period


• PETREL RE to build the production data
Prepare the History • Make measured pressures comparable to simulated pressures

Prepare synthetic • Perforation, equipment, wells interventions history


diagram of well & • Use PETREL-RE or X-cel
completion status Petrel RE: Tuesday

Depending on the study, the data gathering can be already performed


To be done before to the geological building by RE
● SDR1 advanced 2013 - History matching
5
- 15 -
STEP II - DYNAMIC SYNTHESIS TO CAPTURE THE MAIN
HETEROGENEITY

Before the geomodel


perform the 2G&R Synthesis

Initial conditions / Key dynamic


Main static units heterogeneities /
General workflow K field

•Definition of original conditions

•Heterogeneity status in
production phase

•Productivity / Status of K field Definition of rules Volumetrics &


of modeling aquifer activity
/ HM anticipation /Energies
•Field behaviour

•HM anticipation

•Material Balance

SDR advanced 2013 - History matching - 16 -


Step II - Field Review Mbal
fully documented on intranet
• Importance to test the different systems of energy per dynamical sector

• « quick » history match on dynamic data with material balance


• On the full field, for each sector
+1
• Use Experimental design to screen parameters quickly
-1 -1 +1
• Identify and classify the influential parameters for history matching +1
• Consider the potential alternative scenarii -1

Tank Pressure • 16 uncertain parameters screened Pareto on Region 7


Region 7 • Aquifer is the most important parameter Pressure
Region 7

Test the different systems of energy as soon as possible


to reduce / confirm the range of uncertainty (in place, dynamic parameters)

SDR advanced 2013 - History matching - 17 -


SDR advanced 2013 - History matching - 18 -
WELL TEST MATCH EXAMPLE – MOHO (CONGO) - 1
2008/10/16-0359 : OIL
Clean Up PFA
& Flow 1 BU-1 BU-2 Main Final
• DST example from deep offshore turbiditic well
210.

BU BU
• Consistent data
200.

Flow 2 Main Flow


190.
180.

Max
170.

Flo
0. 50. 100. 150. 200. 250. 300.
w
300.
100.
-100.

0. 50. 100. 150. 200. 250. 300.


Time (hours)

Layer 2
2008/10/16-0359 : OIL
-1.5

Layer 3
10

BU-1 BU-2
-2.5

Final-BU Main-BU
10

Layer 4

Drainage radius 18 m 65 m 220 m


-3.5
10

10-3 10-2 10-1 100 101


Delta-T (hr)

MOHO BILONDO MARINE3 inter pvt b21339

SDR advanced 2013 - History matching - 19 -


WELL TEST MATCH EXAMPLE – MOHO (CONGO) - 2

• Simulate the well test data on Reservoir simulation by building Local Grid Refinement

• 4 sources of uncertainties:
• Horizontal Permeability
• Kv/Kh ~ % injectites
• Erosion surface disconnection
• Skin

• Consider the different geological scenarii through an Experimental Design to cover the uncertain
domain

 Screening through experimental design showed quickly that no HM is possible

SDR advanced 2013 - History matching - 20 -


WELL TEST MATCH EXAMPLE – MOHO (CONGO) - 3

• Come back to geological context  other uncertain geological parameters have been
● 21
added through analog (injectites content in another channel facies)

• New screening performed


• 4 Different alternative geologically
History Match were found

• Well test match enable


• To calibrate the model
• To provide reserves distribution
on the possible geological scenarii

SDR advanced 2013 - History matching - 21 -


AGENDA

Definition and a business case

History Match workflow

HM criteria and uncertain parameters

General Strategy for History Matching

SDR advanced 2013 - History matching - 22 -


QUALITATIVE CRITERIA FOR HM VALIDITY - 1

A match may be considered as good if it gives reliable forecast but


• the quality of the predictions can be checked only later
• the experience shows that a good history matching does not always mean good prediction

If the mechanisms are different between the history and prediction


periods, then different parameters affect the field behaviour. In such
cases, forecast reliability may be low

If production history is short with many uncertainties, try to get


different solutions allowing to estimate the range of possible future
performance

SDR advanced 2013 - History matching - 23 -


QUALITATIVE CRITERIA FOR HM VALIDITY - 2

If it is not possible to achieve a good match in the full history


period, retain the model which gives a good late agreement

Do not worry about badly matched wells if they are “end of life” or
not situated in a strategic area for prediction

Give preference to trends/derivatives; keep physical


understanding

Remember when to stop : not necessary when the budget runs


out  balance between time consuming and quality improvement

Don’t neglect the parameters which have no impacts in the


History match but could impact the forecast

SDR advanced 2013 - History matching - 24 -


QUANTITATIVE CRITERIA FOR HM VALIDITY- 1

No universal criteria; they have to be defined with respect to the


study objectives and specificities

Precision varies depending on the quality; how the observations


are representative (repartition at well level) and frequent

It is pointless to try and get all simulated values lying within the
error bars associated with the measurements! The pressure
gauge may be accurate within 1 bar, but …
• extrapolation of depth with uncertain gradient
• measurement during a transient period
• well penetrating several layers with different pressures
• take care to compare apples to apples …

SDR advanced 2013 - History matching - 25 -


QUANTITATIVE CRITERIA FOR HM VALIDITY - 2

Pressure
• one to several bars for individual wells depending on total DP value

Production of main phase


• oil (or gas) production total should be matched within 1-2% at field and
group/region level
• oil (or gas) production rate should be matched at well level within 5-10%
at the end of history period; and also the gradient dQ/dt during the last 6
months
Injection phase
• water (or gas) injection total should be matched within 10% at field and
group/region level. Depends on the uncertainty about the injected
quantities

SDR advanced 2013 - History matching - 26 -


QUANTITATIVE CRITERIA FOR HM VALIDITY- 3

Wcut, GOR
• GOR less precise than Wcut
• if no evolution Þ prediction reliability will be limited
• breakthrough dates: ± 3 months
• some % for Wcut
• it is more important to achieve a good match for the final months
• in case of coning, use of a LGR is essential
• a good match of Wcut at well level is compulsory in case of:
• Gas-lift
• infill wells location
• Debottlenecking; size of facilities for water treatment, re-injection
• Work-over performance

SDR advanced 2013 - History matching - 27 -


LIST OF UNCERTAIN PARAMETERS
• aquifers (extension, strength)
• transmissivities (fault, H, V)
• Permeabilities
• In place
usual • Vp and structure outside the wells area
parameters • Kr end-points

• rel perms shape


• cp (possible impact in case of large depletion)
• PVT (co above Pb, mo if not measured)
• position of contacts
less usual • allocation per layer of Injection rates
but possible

List of parameters are case dependent


SDR advanced 2013 - History matching - 28 -
LIST OF PARAMETERS IS CASE DEPENDENT

To be established for each case, at the


beginning of the match:
• Using the tables hereafter (see also methodology
manual)
• with the help of upstream disciplines (geology,
petrophysics, thermodynamics....)
• by setting their range of uncertainty
• after the first runs:
• knowing more about the sensitivity of the model to
the uncertain parameters
• Highlighting some new parameters: eg injection well
induced fracturing

SDR advanced 2013 - History matching - 29 -


STEP III - GEOLOGICAL PARAMETERS EXAMPLE

Geological uncertainty Dynamic Model “Translation”

Vertical disconnection Vertical transmissivity reduced (8 layers)


Flood Events
Erosion
Sand Cementation

Paleozone Porosity reduced by estimated Sgr


Gas residual Permeability reduced (diphasic flow)
Bituminen coated grain (preventing cement.) No cement related permeability reduction

Aquifer cementation Permeability reduced below paleo GWC


Below paleo contact

Dynamic vs Static Kh Overall permeability reduced


Observed on all Elgin-Franklin well
Observed on 4d-4 (no test on G10)

Internal Faults – gravity faults Implicitly modelled by reduction of Kx


Not seen on seismic
Observed on 4d-4

Porosity depth trend and BRV Variation of porosity per vertical zones

North Panel
Partially not mapped
Porosity increased at the boundary
Fault not mapped Reduction of the permeability of the area

North and South West Fault Inter region transmissivity reduced


SDR advanced 2013 - History matching - 30 -
STEP III - DYNAMICAL PARAMETERS

Dynamical uncertainty

Water viscosity

Kr-Pc curves
Sgr
Krwmax

Productivity
bottom perforation

All input parameters have been implemented by


use of regions and areas directly in the simulator

31

SDR advanced 2013 - History matching - 31 -


AGENDA

Definition and a business case

History Match workflow

HM criteria and uncertain parameters

General Strategy for History Matching


• Main recommandations, guidelines
• General methodology

SDR advanced 2013 - History matching - 32 -


GENERAL RULES FOR PARAMETERS UPDATE

make sensitivity tests


using EST software

be “aggressive”
Regional and then
when performing
local changes
sensitivity tests

consult G&G people


especially before Avoid sharp
updating the model discontinuities
drastically

SDR advanced 2013 - History matching - 33 -


RECOMMENDATIONS FOR WORK PROCESS

Optimize run time, eventually performing some simplifications. But


control the consistency with « complete » runs

keep a log of the different runs

RUN n° Modified Data file Observed Result file Graphic files


parameter effect

Use RE studio Project manager


Tuesday afternoon

keep track of the modifications: initial PERMX Include file and MULTIPLY
instead of only the modified PERMX Include file

time spent to fully analyse the results is never lost; integrated tools as
Petrel-RE are plus

SDR advanced 2013 - History matching - 34 -


GUIDELINES AND RECOS FOR RESULTS ANALYSIS

Give more importance to match trends than absolute values

Make groups of wells with similar behaviour based on:


• Flow-units
• Plots of Pres vs time, vs Np
• Evolution of contacts

There is a list of compulsory plots

Dedicated module of PETREL-RE (and EST)

In complicated cases it may be useful to make schemas for


understanding the flows between units

SDR advanced 2013 - History matching - 35 -


STEP III - ANALYSIS OF THE RESULTS: PREPARATION PLOTS

Comparison of simulated values to observed ones


versus time
• Cumulative production and injection for each phase at Field/PF/Well level
• Production and injection rates for each phase at Field/PF/Well level
• Pres
• Salinity, tracers if any

Comparison of simulated values to observed ones at


each time an observation is available
• RFT, PLTs

3D Visu (FloViz, Petrel, Tempest View): P, Sw, Sg, So 


Depleted area, swept zones, drainage zone
Streamline with Studio SL
Wednesday afternoon
SDR advanced 2013 - History matching - 36 -
Petrel RE- History Match Objective function output

Well top attributes: signed mismatch per vector over the


Mismatch per well and per vector as bubble maps at a user-defined period
given time-step  Mapping of signed mismatch for trend detection

PetrelRE – Objective Function


Tuesday afternoon
SDR advanced 2013 - History matching - 37 -
GENERAL MATCHING STRATEGY - 1

pressure matching:
• At this stage: Ps, P@end of BU, RFT. NOT flowing BHP!
• Is performed controlling wells with reservoir voidage rate (RESV)
to free oneself from the phase distribution
• The model acts as a material balance problem:
• We act on the parameters affecting the diffusive equation to
adjust the spatial distribution of accumulations and the degree of
communication between the different energy sources.

SDR advanced 2013 - History matching - 38 -


GENERAL MATCHING STRATEGY - 2

saturation matching (GOR, Wcut, contacts, PLT):


• Not completely sequential as breakthrough are indicators of the degree of
communication
• Switch controlling the wells with surface rate only when Wcut and GOR are
almost matched
• use of WHISTCTL
• LRAT is an alternative when GOR ~ cst
• Act on the parameters of the transport equation.

SDR advanced 2013 - History matching - 39 -


GENERAL MATCHING STRATEGY - 3

Final adjustments
• launch « complete » runs again: production data,
« complex » options (hysteresis, updating COMPDAT …)
• up-date properties if necessary
• Introduce the VFP tables & gas-lifts
• Finalize the PIs (well indices) adjustments to match WBHP (if
available) or WTHP
• Re-check the PLT

SDR advanced 2013 - History matching - 40 -


GENERAL MATCHING STRATEGY - 4

WBHP and WTHP matching:

• At the first stage, when matching the reservoir pressure, we don’t care about
WBHP and WTHP
• it may be useful, however, to increase the “connection factor” to honour the
observed rates (use WPIMULT)

WBHP and WTHP must be matched before the prediction period,


to ensure the continuity from boundary conditions in Q (HM) to
BC in P (forecast)
• matching  Preservoir+ PI  WBHP + VFP  THP

SDR advanced 2013 - History matching - 41 -

You might also like