1 Predicting Porosity and Hydrocarbon Saturation of Rock Formations During Drilling Using Genetic Algorithms & Fuzzy Logic Alexandra D. Pinto de ANDRADE, Stephen J. CUDDY & Paul W.J. GLOVER University of Aberdeen & PetroInnovations
1
Predicting Porosity and Hydrocarbon Saturation of Rock Formations During Drilling Using Genetic Algorithms &
Fuzzy LogicAlexandra D. Pinto de ANDRADE, Stephen J.
CUDDY & Paul W.J. GLOVERUniversity of Aberdeen & PetroInnovations
2
Structure
The Role of Improved Data Analysis in Minimising the Impact of Hydrocarbon ExtractionThe Problem of Porosity and Hydrocarbon Saturation PredictionCuttings Gas Logs & Genetic AlgorithmsA Field ExampleSummary
3
Minimising Environmental Impact
Can take many formsRemediation of current pollutionThe limitation of spills at refineriesImproved methods of transporting oilImproved rig decommissioningThe reduction and prevention of pollutant gas emissionExtending current field life using improved analysis techniquesImprovements to the design of drilling and production rigsImproved seismic methods to reduce their impact upon the sea-life
4
The Problem
Improved analysis of reservoirs depends on accurate knowledge of the porosity and hydrocarbon saturation at depthExpensive, time-consuming well log techniques are usedThese techniques can be environmentally damaging
5
Cuttings Gas Logs - What Are They?
The log of the different hydrocarbon gases evolved from drilling cuttings
Are done in every well during drillingAre a statutory obligation on the grounds of safetyAre therefore “free” and immediately available
They have an extremely poor vertical resolutionThey have not been successfully linked to useful reservoir properties such as porosity and saturation
6
Cuttings Gas Logs - Types
Commonly the lighter alkanes are analysedC1, C2, C3, iC4, nC4, C5, C5+
Sometimes expressed as gas ratiosHydrocarbon Wetness (Wh)
Hydrocarbon Balance (Bh)
Hydrocarbon Character (Ch)
100 x)C + C + C + C + C(
)C + C + C + C( = Wh54321
5432
)C + C + C()C + C( = Bh
543
21
C)C + C( = Ch
3
54
7
Genetic Algorithms (GAs) - What Are They?
Computer-basedTake a general form of an equationEvolve the equation constants and operators until a best fit to some calibration data is foundThe evolution may include random changes, cloning, sexual reproduction etc.The evolved equation uncovers the mathematical relationships hidden in the calibration dataThe equation can be used to predict any desired parameter
8
Genetic Algorithms - The Equation
General Form
Y(A, B, C, D…) = aAb1 cBd
2 eCf3 gDh
4 …
where:
i = Either + , - , ÷ or ×a, b, c, d, e, f ... are constant parametersA, B, C, D … are variables in the calibrating data setY(A, B, C, D…) is the parameter that is required
9
Genetic Algorithms - Uses
Can discover the mathematical relationship linking complex patternsCan be used to predict the porosity and saturations in the sub-surface from well-log dataBut also from any data that contains information about porosity and saturation no matter how complex or slight the relationshipIf CGLs contain information, GAs will find itThen porosity and saturation can be predicted
10
GA - Example
Input (Calibration Data)CALIGRDT
Output (Predicted Data)DTS_GA_4DTS_FL_4
Test (Comparison Data)DTSDTS_1
11
The Use of Cuttings Gas Analysis and Genetic Algorithms to Predict
the Porosity and Hydrocarbon Saturation
A Field Example
12
Results IGas Data
Individual Cuttings Gas: Track 1Total Gas: Track 2Cuttings Gas Ratios: Track 3
Hydrocarbon Character (HC)Hydrocarbon Balance (HB)Hydrocarbon Wetness (WET)
Comparison of HydrocarbonCharacter Ratio with Gamma Ray
GR and HC Ratio: Track 4GR alone (filled): Track 5HC alone (filled): Track 6
13
Results IIGas Data
Individual Cuttings Gas: Track 1Total Gas: Track 2Cuttings Gas Ratios: Track 3
Hydrocarbon Character (HC)Hydrocarbon Balance (HB)Hydrocarbon Wetness (WET)
Comparison of HydrocarbonCharacter Ratio with Gamma Ray
GR and HC Ratio: Track 4GR alone (filled): Track 5HC alone (filled): Track 6
14
Results IIITrack 1: Comparison of HB with Density Log
Hydrocarbon Balance (HB)Density Log (RHOB)
Track 2: Comparison of HydrocarbonWetness with Neutron Porosity Log
Hydrocarbon Wetness (Wet)Neutron Porosity Log (NPHI)
Tracks 3 & 4: Comparison ofNPHI/RHOB Combination with Hydrocarbon Balance/WetnessCombination
Green = ShalyYellow = Sandy
15
Results IVTrack 1: Comparison of HB with Density Log
Hydrocarbon Balance (HB)Density Log (RHOB)
Track 2: Comparison of HydrocarbonWetness with Neutron Porosity Log
Hydrocarbon Wetness (Wet)Neutron Porosity Log (NPHI)
Tracks 3 & 4: Comparison ofNPHI/RHOB Combination with Hydrocarbon Balance/WetnessCombination
Green = ShalyYellow = Sandy
16
Implications
Clearly, there is some relationship between:The Gamma Ray Log and the Hydrocarbon Character RatioThe Density Log and the Hydrocarbon Balance RatioThe Neutron Porosity Log and the Wetness Ratio
Genetic Algorithms can find this relationship, and use it to predict porosity
17
Results VTrack 5: Comparison of Porosity
From Gas Ratios using GAsFrom Conventional Logs
The conventional logs are:ExpensiveTake many extra days to do
The gas log/GA method is:FreeAvailable during drilling
Track 6: Final Rock Analysis fromthe Gas Ratio Data
ShaleSandOilWater
18
Results VITrack 5: Comparison of Porosity
From Gas Ratios using GAsFrom Conventional Logs
The conventional logs are:ExpensiveTake many extra days to do
The gas log/GA method is:FreeAvailable during drilling
Track 6: Final Rock Analysis fromthe Gas Ratio Data
ShaleSandOilWater
19
Summary I
Human development requires the energy and raw materials provided by oilThe analysis of these reserves uses techniques that are expensive and can be damaging to the environmentImproved analysis using gas cuttings measurements and GAs provide good porosity, permeability and lithofacies dataEnvironmental damage is reduced by reducing the use of invasive exploration techniques and obviating the need for new wells and reservoirsHowever, oil production is optimised with less environmental impact