automated processing of large data volumes for development of the hugoton-panoma geomodel geoffrey...

21
Automated processing of Automated processing of large data volumes for large data volumes for development of the development of the Hugoton-Panoma geomodel Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin Geoffrey C. Bohling, Martin K. Dubois, John Doveton and K. Dubois, John Doveton and Alan P. Byrnes Alan P. Byrnes Kansas Geological Survey

Post on 20-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Automated processing Automated processing of large data volumes of large data volumes

for development of the for development of the Hugoton-Panoma Hugoton-Panoma

geomodelgeomodel

Geoffrey C. Bohling, Martin K. Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan P. Dubois, John Doveton and Alan P.

ByrnesByrnesKansas Geological Survey

Page 2: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Processing StepsProcessing Steps

• Generation of geologic constraining Generation of geologic constraining variables (logs) from tops setvariables (logs) from tops set

• Prediction of lithofacies from logsPrediction of lithofacies from logs

• Porosity log correctionPorosity log correction

• Computation of water saturation, Computation of water saturation, OGIPOGIP

• Back-calculation of free water levelBack-calculation of free water level

Page 3: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Solution strategySolution strategy

• Excel add-in Kipling2.xla used for Excel add-in Kipling2.xla used for prediction of lithofacies from logs prediction of lithofacies from logs (neural net code in Visual Basic)(neural net code in Visual Basic)

• Remaining steps accomplished with Remaining steps accomplished with special-purpose Excel workbooksspecial-purpose Excel workbooks– Spreadsheet input driving VB codeSpreadsheet input driving VB code– Batch processing of LAS filesBatch processing of LAS files

Page 4: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Geologic constraining Geologic constraining variablesvariables• Depositional environment code (MnM)Depositional environment code (MnM)

– 1 for nonmarine1 for nonmarine– 2 for marine2 for marine– 3 for tidal flat3 for tidal flat

• Relative position curve (RelPos)Relative position curve (RelPos)– 0 at bottom of interval to 1 at top0 at bottom of interval to 1 at top

• Added to well logs for facies predictionAdded to well logs for facies prediction

Page 5: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

The solution: GenMnM.xlsThe solution: GenMnM.xls

• Spreadsheet with table of tops, Spreadsheet with table of tops, depositional environment codes for depositional environment codes for each intervaleach interval

• Attached VB code generates LAS files Attached VB code generates LAS files with depositional enviornment code with depositional enviornment code (MnM) and relative position (RelPos) (MnM) and relative position (RelPos) curvescurves

Page 6: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

GenMnM.xlsGenMnM.xls

Page 7: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Prediction of facies from Prediction of facies from logslogs

• Using neural net in Excel add-in Using neural net in Excel add-in Kipling2.xlaKipling2.xla

• Until recently, Kipling2.xla called Until recently, Kipling2.xla called neural net function in “R” statistical neural net function in “R” statistical language for training; now all in VBlanguage for training; now all in VB

• Includes batch prediction over LAS Includes batch prediction over LAS filesfiles

Page 8: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Neural NetNeural Net

Page 9: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Training the neural netTraining the neural net

Training for prediction of L10, Chase

4000

5000

6000

7000

8000

9000

10000

0 10 20 30 40 50 60 70 80 90 100

Iteration number

Ob

ject

ive

fun

ctio

n

Page 10: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Prediction on spreadsheet dataPrediction on spreadsheet data

Count of Pred. L10Pred. L10L10 1 3 4 5 7 9 10 Grand Total

1 241 16 1 2 22 2822 8 1 1 103 5 5 10 204 1 15 5 17 21 7 506 1 2 1 47 3 12 43 3 618 10 1 119 3 4 2 9

10 1 14 15Grand Total 262 23 1 35 81 3 58 463

Facies Probabilities, Shankle

0 0.5 1

2540

2552.5

2564.5

2576.5

2588

2600

2611.5

2623

2635

2647

2664.5

2676

2687.5

2699.5

2711

2722.5

2734

2745.5

2757

2768

2780

Elo

gD

epth

NM Silt

NM ShlySilt

Mar Shale & Silt

Mdst/Mdst-Wkst

Wkst/Wkst-Pkst

Fxln Sucrosic (Dol)

Pkst/Pkst-Grnst

Grnst/PA Baff

Moldic/sucrosic dol

Marine SS

Page 11: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Batch prediction on LAS filesBatch prediction on LAS files

Page 12: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Porosity log correctionPorosity log correction

• Variations from reference mineralogy Variations from reference mineralogy (limestone)(limestone)– Regression of log vs. core porositiesRegression of log vs. core porosities

• Washouts in shales Washouts in shales – Remove spikes and associated shouldersRemove spikes and associated shoulders

• Excel worksheet with VB code attachedExcel worksheet with VB code attached– Processes LAS files in specified folderProcesses LAS files in specified folder

Page 13: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Log-core porosity relationshipsLog-core porosity relationships

0

5

10

15

20

25

0 5 10 15 20 25Density porosity (ls equiv)

Co

re p

oro

sity

Facies 1

Facies 2

Facies 3

Facies 4

Page 14: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Washout correctionWashout correction

0.20 0.15 0.10 0.05

Porosity

2840

2835

2830

2825

2820D

ep

th

Wa

sho

ut

Th

resh

old

= 0

.20

mean = 0.07

drops below mean

turns back up

Before replacing peakAfter replacing peakAfter replacing shoulders

Page 15: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Porosity correction Porosity correction worksheetworksheet

Page 16: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Batch OGIP calculationBatch OGIP calculation

• Computes Sw & OGIP vs. depth at wells Computes Sw & OGIP vs. depth at wells from facies, porosity in LAS filesfrom facies, porosity in LAS files

• Uses porosity-Sw transforms by facies; Uses porosity-Sw transforms by facies; need HFWL and initial pressureneed HFWL and initial pressure

• Constant FWL (from spreadsheet) or by Constant FWL (from spreadsheet) or by well (from headers)well (from headers)

• Ditto for initial pressureDitto for initial pressure• Generates LAS files with Sw, OGIP Generates LAS files with Sw, OGIP

curvescurves

Page 17: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Batch OGIP spreadsheetBatch OGIP spreadsheet

Page 18: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Backcalculating FWL from Backcalculating FWL from OGIPOGIP

• Computes FWL at each well that Computes FWL at each well that produces best match between produces best match between volumetric OGIP and mass balance volumetric OGIP and mass balance OGIPOGIP

• Mass balance OGIP (target) for each Mass balance OGIP (target) for each well read from LAS headerswell read from LAS headers

• Other than fitting FWL, works like Other than fitting FWL, works like Batch OGIPBatch OGIP

Page 19: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Bracketing search for FWLBracketing search for FWL

Page 20: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Concluding remarksConcluding remarks

• Flexible, easy batch processing Flexible, easy batch processing crucial to a project of this scalecrucial to a project of this scale

• Excel with VB provides reasonably Excel with VB provides reasonably accessible & inexpensive accessible & inexpensive environment for developing environment for developing customized processing codecustomized processing code

Page 21: Automated processing of large data volumes for development of the Hugoton-Panoma geomodel Geoffrey C. Bohling, Martin K. Dubois, John Doveton and Alan

Hugoton Asset Hugoton Asset Management Project Management Project industry partnersindustry partners• Anadarko Petroleum CorporationAnadarko Petroleum Corporation• BP America Production CompanyBP America Production Company• Cimarex Energy CompanyCimarex Energy Company• ConocoPhillips CompanyConocoPhillips Company• E.O.G. Resources Inc.E.O.G. Resources Inc.• Medicine Bow Energy CorporationMedicine Bow Energy Corporation• Osborn Heirs CompanyOsborn Heirs Company• OXY USA, Inc.OXY USA, Inc.• Pioneer Natural Resources USA, Inc.Pioneer Natural Resources USA, Inc.