Академический Документы
Профессиональный Документы
Культура Документы
Panoma (Council Grove) Geomodel Initial Simulations (Single Well) This paper is a snapshot of an ongoing effort with the
ultimate goal of creating a robust three-dimensional geomodel that is
suitable for accurate reservoir analysis and simulation. The work to
date demonstrates:
Formation-Member Data Three Runs
Simulations Using Geomodels 1. Eight lithofacies have distinct suite of petrophysical
Tops Coverage Lithofacies Model Predicted Lithofacies at Well
(upscaled) Alexander D2
1. Upscaled from well
2. GeoModel, Rate Specified
properties and use of those properties in calculating volumetric
original gas in place (OGIP) reduces error in the calculation.
Upscale Lithofacies 1.6 BCF cum. 3. Geomodel, Pressure Specified
(1) Upscale lithcode 2. Neural network models accurately predict lithofacies in non-
curves wells having facies cored wells using log curves and geologic constraining variables
predicted by NNets.
1.0 Parameters
Core
Upscaled
Predicted ! 640 Acre Section after training on wells with cores.
Core
Lith Lith ! Cell Size: 390’ X 415’
Tops Check
0.75
Lith
(2) Populate cells at these ! Layers: Upscale Well – 6 3. The petrophysical model, synthetic capillary pressure curves,
Geomodel -- 41
(in Petrel) wells with upscaled facies.
! Well Location: Center
petrophysical transform equations, facies prediction and porosity
correction appear to be validated by property-based volumetric
Data busts readily apparent
(rotated view looking north).
(3) Model cells between 0.75
1.0 ! .6’ X 315’ Fracture
! 100 Year Run GR Lith OGIP that matches material balance OGIP.
wells for lithofacies A1 Interval ! Swic 30 %
A1 Interval ! BHPi 260 psia 4. Initial simulations using cellular model appear to be working well.
11000+ wells, formation level tops
(Sequential Gaussian). A Cum. Gas (BCF)
constant average curve was
Model fitted creating a constant NM Silt & Sd Wackestone
10^5
5. Vast detailed geomodel is made possible by extremely large set
ALEXANDER D2 - PANOMA of detailed tops, digital logs, core data and automation (facies
Boundary distribution (from "lith-code" NM Shly Silt Dolomite 300
prediction and OGIP calculator, for example).
curves) equal to the BHP/Z 10^4
A1_LM
Data Management: Mar Shale & Silt Packestone 240
Flowing
average probability of that
Monthly Prod.
3D Lithofacies Model
Pressure
Tops data set
24,879 Total wells in initial PETRA project, Panoma facies. Mudstone Grnst & PA Baf 180
including regions outside of model Shrimplin GU-2HI A1 Interval 10^3
12,097 Wells having at least Council Grove Field ~500 120
top pick by KGS geologist
11,367 Wells in initial structure model.
Well
Perm, Layer 1 FURTHER WORK:
Non-biased 2:1 Trend Biased Heavy Trend Biased 10^2
10,836 After screening for tops busts. Biasing Lithofacies (NE-SW) (NE-SW)
Facies Distribution 60
Further screening reduced the well
Model Biased Additional effort will be in several broad areas; 1) “ground truthing”
count to10,700.
Geometry 0 0.4 0.8 1.2 1.6 2.0
10
lithofacies prediction and extrapolation, 2) increasing coverage, 3)
2007
2015
1983
1991
1999
1975
Digital well log data set Gas Cum (BCF)
1,103 Sufficient Council Grove penetration Model facies using different biasing improving the neural network model and Petrel models, and 4)
536 After removing wells with bad curves, parameters based on geologic expanding reservoir simulation.
469
data gaps or other problems
Final count after further screening 500 wells with proper log suite for understanding. Examples include “non- B1_LM
facies and petrophysical modeling biased” using a spherical range of Frac Analysis on Panoma Well 1. Test the Nnet models by comparing core lithofacies (from
100,000' x 100,000', slightly biased (2:1 Newby 2-28R 12/03/1997 undescribed, available core) with those lithofacies predicted.
ratio to NE-SW), heavily biased to NE- A1-LM 2. Test Petrel's stochastic facies modeling by comparing its
A1-LM
Model Architecture SW, and bias based on mapped
regional facies distribution patterns.
Grnst-PA) Grnst-PA)
A1-LM
Grnst-PA)
B2_LM
Range (md)
0.2 7.6
results to Nnet predictions at wells that were not used in
conditioning the Petrel model.
B1-LM (Grnst-PAbaf) 3. Consider alternative workflow: populate Petrel cells with
Slice and Dice Define Framework variables and use Nnet models to predict facies at cells.
Seven cycles modeled 1. Create a "skeleton grid” 4. Increase well coverage (wells with appropriate log curves).
For efficiency while maintaining fine
heterogeneity, reservoir model was
divided into 7 marine-nonmarine cycles
2. Construct top horizon for
Council Grove Group (top A1_SH) Petrophysical Model 3D view of Facies 8 (Grnst/PAbaff) 3D view of Permeabilty B3_LM
5. Improve NNet models, if possible, by incorporating other
statistically based classification techniques (e.g.: fuzzy logic and
Markov chains).
(A1-C) for lithofacies, porosity and 3. Create isochores for subjacent This view of the occurrences of Facies 8 illustrates the (Facies 8) A view of a portion of the adjacent
zones (2/cycle) and “hang” on Upscale Porosity Raw log data was discontinuous nature of this particular facies in 3D. 6. Explore use of Nnet facies probabilities in Petrel facies
permeability modeling. The 7 perm slide illustrates the permeability distribution with B4_LM
models were then joined for simulation. top horizon quality checked and Discontinuity, to different degrees, is the norm for all facies. geometry biasing.
of Facies 8.
Proportional layering in cycles 4. Proportionally layer the zones
0.2 30 0.2 29
7. Establish a detailed, field-wide free water level.
aliased. During the B5_LM 8. Analyze GIP by comparing property based volumetric with
Layer (N) proportional to interval 5. QC structural framework by processing a cross-
thickness (h). Nm (marine) = (mean h + sectioning Core
Lith
X-Plot
Log Phi
Upscaled
Log Phi plot porosity curve Lithofacies 8 (grnst/pa) Permeabiltiy material balance OGIP and compare both with production history.
sd) and Nnm (nonmarine) = mean h. 9. Simulate at a variety of scales.
A1_LM Lithofacies 8 (grnst/pa)
“Dummy” layers hold places of other
cycles in each model (12 per model)
was generated using
the Neutron and
Density Porosity
A1_LM Upscaled from Well Model CHALLENGES:
curves. No shale
correction was made Core Properties Lithofacies Layering The primary challenge in this project is to develop single model with
in the first models. ! Tied to Lithofacies ! Used 6 Layers to capture different facies sufficient detail to be useful at the field, region, area and well scale.
Similar to facies ! Corrected for Klinkenberg & zones
Related to the overall challenge are individual hurdles:
modeling, porosity ! Filter Data> .1 Perm,>.01 Perm ! Simulation under performed actual Pressure Specified Run
curves were Step down pressure curtailment 1. Manage vast data sets required for detailed characterization.
"upscaled" at the wells Layer Porosity Perm Top Bottom Gross Net (Back Pressure Curve) 2. Balance upscaling against model utility.
Interval Isochores for modeling porosity Initial Pressure: 243 Psia 3. Physical limitations of software and hardware.
1 8.679 5.864239 2900.5 2905.5 5 5
in cells between the Recovery 2.5 BCF 4. Upscale simulation exercises from well to field.
wells (Sequential 2 8.475 0.107800 2906.5 2920.5 14 11.5