Advanced Facies Modelling Presentation PDF

Advanced Property Modelling in Roxar RMS Advanced Facies Modelling Petronas Roxar Academy 2016 Ruslan Nasibullin Senior Geologist Roxar Software Solutions AP www.roxarsoftware.com RMS Facies Modelling Techniques Pixel based methods Transitional facies Fluvial facies (same OWC’s) Carbonates Delta...

Advanced Property Modelling in Roxar RMS Advanced Facies Modelling Petronas Roxar Academy 2016 Ruslan Nasibullin Senior Geologist Roxar Software Solutions AP www.roxarsoftware.com RMS Facies Modelling Techniques Pixel based methods Transitional facies Fluvial facies (same OWC’s) Carbonates Delta Regional heterogeneity Object based methods Channels & Levee Fluvial facies (different OWC’s) Carbonates Turbidities Emerson Confidential • 4/30/2018 • Slide 7 RMS Petrophysical Modelling Techniques Deterministic Methods Quick & Simple Models Interpolation Algorithm Trend Modelling Stochastic Methods Realistic 3D Models Facies Heterogeneity Random samples of probabilistic descriptions Emerson Confidential • 4/30/2018 • Slide 9 RMS Facies Modelling Methods Facies: Belts Facies: Indicators Pixel-based methods - F*: Facies proportions - VPC - F: Facies proportions - FPF Method Characteristic property Well Data 1D Trends - Facies proportions 2D - Facies proportions - Intensity trends, etc Facies: MPS Facies: Facies: Channels Composite Object modelling methods Interpreted facies log (discrete type) - FPF - FPF - FPF - VPC - F: Facies proportions - F: Facies proportions - F: Facies proportions - Size and shape of - Size and shape of objects objects - Facies proportions - Facies proportions - Facies proportions - Seismic data - Seismic data - Facies proportions - Azimuth directions - Facies proportions - Training image - Variogram ranges - Seismic data - Azimuth directions - Seismic data 3D Polygons, Points - Facies proportions - Accuracy of the seismic data - Seismic data Facies: SedSeis Pixel-based/Object - Thickness of the objects - Facies proportions - Size and shape of objects (continuous / discrete) - Seismic data Changing of facies belts boundaries Possibility specify any discrete parameter as hard data Additional original data Local update modelling mode − + + Body correlation between wells Possibility to use shape of objects, which have been created manually Possibility to set Object height limitation distance to channel by surfaces edges + + − Note: * F - function, which can reflect the variations of different variables (e.g. facies proportion, object's orientation, resizing objects with depth, etc.). Vertical proportion curves (VPCs) show vertical variations in the volume fractions for different facies. A single data object stores information on the vertical trends for each facies for (optionally) each zone. Emerson Confidential • 4/30/2018 • Slide 18 Facies Probability Functions (FPFs) give a probabilistic relation between facies types and seismic attribute values. They are estimated using the following seismic data types, in combination with a bias facies log or parameter RMS Facies Pixel Based Methods Facies Belts Input Data • Definition of the facies belts and their ordering • Information on migration directions and other geometric attributes • Interfingering variograms • Interpreted facies log Emerson Confidential • 4/30/2018 • Slide 19 Applications Benefits Shoreface Carbonate reef deposits Deltaic Output parameters can be used as trends Facies Belts: Geometry Depositional direction Aggradation Angle Coordinates of boundary Emerson Confidential • 4/30/2018 • Slide 20 Facies Belts: Facies Boundary Types Your choice of cross sectional geometry should be based on the general, large-scale facies architecture of the reservoir grid model • Parallel • Pinch out • Curved Emerson Confidential • 4/30/2018 • Slide 21 Facies Belts: Interfingering Modelled using a Gaussian field with a specified standard deviation and range • Large Standard deviation • Large variogram range • Short variogram range Emerson Confidential • 4/30/2018 • Slide 22 Facies Belts: Stacking Angle Ac  S Retrograding system Ac  S Aggrading system Ac < S Ac = Accommodation space Emerson Confidential • 4/30/2018 • Slide 23 S = Sediment supply Prograding system Facies Belts: Extra Tips (Algorithm) Trend algorithm equivalent to Kriging without well conditioning (even if well conditioning is toggled on!) It can be used for a quick result to check that the settings are adequate Only the Simulation algorithm will produce interfingering (using the variogram) Emerson Confidential • 4/30/2018 • Slide 25 Facies Belts: Extra Tips (Algorithm) Trend ▪ ▪ Well data not honoured Trend is used only to run a quick check on the geometry settings Emerson Confidential • 4/30/2018 • Slide 26 Kriging ▪ ▪ Kriging option is equivalent to Trend with honouring Well data if specified No interfingering Simulation ▪ Honouring Well data if well conditioning toggled ON and interfingering Facies Belts: Extra Tips (Simulation sSettings) Save memory option in order to run the simulation faster in big fields User defined values: 0.01 = 1% of blocked well cells might not be honoured, but fast simulation 0.1 = 10% of blocked well cells might not be honoured, but very fast simulation Emerson Confidential • 4/30/2018 • Slide 27 Facies Belts: Extra Tips (Proportions Mode) Good for modeling lensoid shapes The ordering of facies is important (if more than 2 facies), the succession will always be honoured This option is also very useful when trying to model big shale barriers Emerson Confidential • 4/30/2018 • Slide 28 Facies Belts: Extra Tips (Proportions Mode with Trends) 1D Trend Very useful to reproduce exactly 1D trends like VPC with very low or high values The facies ordering will be honoured as well 3D Trend 3D facies proportions trends have to be created first, and this is not trivial and may be time consuming Emerson Confidential • 4/30/2018 • Slide 29 Facies Belts: Extra Tips (Trends & Threshold) Trends and threshold option used to produce very basic seismic conditioning Emerson Confidential • 4/30/2018 • Slide 30 Facies Belts: Extra Tips (Pinchout Polygons) Emerson Confidential • 4/30/2018 • Slide 31 Facies Belts: Extra Tips (Interfingering) It is recommended to use the General Exponential variogram (very stable) for the interfingering, and to play with the value of the exponent for a more cosmetic result, and more continuous facies (important if connectivity issues) The default variogram (Spherical) for noisier boundaries If an even more smooth effect is needed, use General exponential with a high exponent ~1.9) instead of a Gaussian, which can be less stable Emerson Confidential • 4/30/2018 • Slide 32 Facies Belts: Extra Tips (Interfingering) Example of proportion mode with fixed Ranges 3000x3000, but different variograms Spherical noisy General Exponential exponent=1 (~ equivalent to Exponential variogram, very noisy, like the Spherical variogram) Emerson Confidential • 4/30/2018 • Slide 33 General Exponential exponent=1.5 Smoother boundaries General Exponential exponent=1.9 Very smooth (close to Gaussian variogram) Facies Belts: Extra Tips (Interfingering) To determine XY ranges use geological understanding together with well information and trial and error if you have few wells Depth range : use the BW statistics – Start with the average facies thickness – Run the Facies Belts job – Once this is done, create a new log from the created facies parameter on the BW – Check if the statistics are compatible with the original ones. If not, modify the depth range and run the job again Emerson Confidential • 4/30/2018 • Slide 34 RMS Facies Pixel Based Methods Facies Indicators Input Data Applications Benefits • Definition of the facies • 1D, 2D or 3D trends • Vertical proportion trends The generation of very flexible facies patterns, for any number of facies Flexible trend incorporation • Seismic constraints (cosimulation and seismic as probability trends) • Facies probability functions • Depth converted seismic attribute (optional) Emerson Confidential • 4/30/2018 • Slide 35 Facies Indicators: Handling of Mature Fields Performance is independent of number of wells and use of seismic data Well data Facies realisation Important complement to the object modeling techniques in RMS which are more suitable for fields with less data Emerson Confidential • 4/30/2018 • Slide 39 Facies Indicators: Flexible Trend Incorporation A wide variety of 1D, 2D and 3D volume fraction trends can be used as input Can model any number of facies Cross-section through Indicator realisation Vertical proportion curves Emerson Confidential • 4/30/2018 • Slide 40 Facies Indicators: Vertical Proportion Curves Input Vertical Proportions Upwards - fining, Coal at top Emerson Confidential • 4/30/2018 • Slide 41 Upper layer Shale layer Lower layer Facies Indicators: Seismic Constraints Two main methods for incorporation of seismic data • • Cosimulation Seismic as trend Vshale map from seismic attributes Emerson Confidential • 4/30/2018 • Slide 42 Slice through facies model Net sand map RMS Facies Object Based Methods Facies Channels Input Data Applications Benefits • Definition of the background and object facies Variety of channel depositional settings (fluvial • delta plain • deep marine) Modelling highly complex channel systems (single- & multi-channel-belts) • Volume fraction of the different facies • Information on the geometry dimension an orientation of the object facies Emerson Confidential • 4/30/2018 • Slide 45 Facies Channels Models fluvial or tidal systems Models crevasses attached to the channels Models barriers within the channels Emerson Confidential • 4/30/2018 • Slide 46 Facies Channels: Different Fluvial Systems Two Main Operating Modes “Sandbody” mode “Multi-channel” mode For single cut-and-fill channel sand bodies and direct modeling of channelbelt geometry For modeling multiple channels within channel belts Requires minimal user input Requires more complex user input Emerson Confidential • 4/30/2018 • Slide 47 Facies Channels: Extra Tips Common mistake : Algorithm is not suited to model internal channel bodies (point bar, sand bar…). It should be used to model the main channel belts Same with crevasses : designed to model overbank complexes, not splays Object modeling algorithm : • • • • First channel is drawn randomly in the simulation box Calculate its probability according to all constraints (wells, seismic, geometry…) If probability is high, the channel is kept, Next iteration tries with another object Iterative technique avoids bias at the wells, and takes into account that the wells are more likely to reach wide channels than narrow ones Emerson Confidential • 4/30/2018 • Slide 57 Facies Channels: Extra Tips (Volume Fraction) Emerson Confidential • 4/30/2018 • Slide 58 Facies Channels: Extra Tips (Volume Fraction) Emerson Confidential • 4/30/2018 • Slide 59 Facies Channels: Extra Tips (Geometry Settings) Typical channel width distribution Azimuth direction : always allows some variability Sinuosity : remember you are modelling the channel belts, not the meandering intra channel bodies Maximum recommended sinuosity value is 1.8 Emerson Confidential • 4/30/2018 • Slide 60 RMS Facies Object Based Methods Facies Composite Input Data Applications Benefits • Definition of the background and object facies High perm thief grid models • Shale baffles • Deltas lobes • Calcite layers • Volume fraction of the facies / Target number of bodies Conditioning to a large number of wells • Multiple well-to-well correlation • Modelling with multiple trends • Information on the geometry, dimension and orientation of the facies Emerson Confidential • 4/30/2018 • Slide 62 Facies Composite Advanced object based Multi-well conditioning Object Interactions Implicit Erosion Rules 3D Seismic Conditioning Powerful Trend Control Local Volume Fractions Emerson Confidential • 4/30/2018 • Slide 63 Facies Composite: Objects Rectangle Axial Ellipsoid Angular Cone Backbone Note : a variant of the axial shape, where the object is deformed locally to follow a direction parameter Emerson Confidential • 4/30/2018 • Slide 64 Facies Composite: Example of User Defined Shapes General channel (Axial) Lobe (Axial) Turbidite channel (Axial) • Fluvial and deltaic channels • Delta mouthbars • Turbidite lobes • Typical “Detailed Shape” settings − Centre-line amplitude: 0.1 - 1.0 − Width rel. amplitude: 0.05 - 0.25 − Relative wavelength: 0.1 - 1.0 • Typical “Detailed Shape” settings • Typical “Detailed Shape” settings − Centre-line amplitude: 0.05 - 0.25 − Centre-line amplitude: 0.1 - 0.5 − Width rel. amplitude: 0.05 - 0.25 − Width rel. amplitude: 0.05 - 0.25 − Relative wavelength: 0.5 - 1.0 − Relative wavelength: 0.1 - 1.0 Emerson Confidential • 4/30/2018 • Slide 65 • Fluvial and deltaic channels Facies Composite: Example of User Defined Shapes Arcuate bar (Axial) “Flexible” ellipsoid (Axial) • Wave reworked mouthbars • Wide spread shale, calcite or coal barriers • Flexible and realistic geometries for multi-well conditioning of widespread barriers • Typical “Detailed Shape” settings − Centre-line amplitude: 0.05 - 0.25 − Width rel. amplitude: 0.05 - 0.25 − Relative wavelength: 0.5 - 1.0 • Typical “Detailed Shape” settings − Relative amplitude: 0.1 – 0.5 − Relative wavelength: 0.5 - 1.0 Emerson Confidential • 4/30/2018 • Slide 66 Facies Composite: Trends 1D, 2D or 3D trends for: – object sizes – orientation – volume fractions – distribution Control of facies distribution by 2D or 3D seismic attributes Large scale facies realizations can be input as intensity functions to control smaller scale facies distribution Emerson Confidential • 4/30/2018 • Slide 67 Facies Composite: Reference Point of Objects To position an object, the algorithm consider the reference point to start with With reference point is 0,0,0 (default) in the object, it means that it is exactly at the centre of the object shape Reference point specification For turbidites, put the reference point at the proximal end of the shape, to force bodies positioning even more Emerson Confidential • 4/30/2018 • Slide 68 Facies Composite: Extra Tips (Volume fraction) Volume fraction: – Global → no systematic trends in the volume fraction, but relative intensity trend can be used – Local → used to introduce trends in the volume fraction. The settings are specified as attribute distribution Emerson Confidential • 4/30/2018 • Slide 69 Facies Composite: Extra Tips (Geometry settings) Standard deviation of object size : can use 1/5th of the average, up to the average value itself Height : it is the maximum thickness of the object, related to the reference point There is a lot of uncertainty on the bodies height : – Can easily add 20-25% more to the average observed at BW – Dip : leave the default value ! If thin objects, a higher dip will produce lateral discontinuity Emerson Confidential • 4/30/2018 • Slide 70 Facies Composite: Extra Tips (Simulation settings) High T0 → much more variability in volume fraction at the beginning of the simulation Low T0 → less variability If volume fraction is difficult to reach, then reduce T1 (but not recommended) Emerson Confidential • 4/30/2018 • Slide 71 Facies Realisation Merging Used to create complex environments Hierarchical modeling (by defining ’erosion’ rules) Emerson Confidential • 4/30/2018 • Slide 72 RMS Facies Object Based Methods Facies SedSeis Input Data Applications Benefits • Digitized Polygon or Point Data information • Seismic data Two different methods can be run as one job: Pixel Based / Object Based • Facies log High-permeable flowregions • Facies bodies identified on good resolution seismic Emerson Confidential • 4/30/2018 • Slide 73 RMS Facies Pixel Based Methods Facies Multipoint Statistics (MPS) Input Data • Training Images (TI) • Discrete parameter representing different areas (region parameter) • Volume fractions for each facies type / region / trend • Seismic data (attributes) Emerson Confidential • 4/30/2018 • Slide 74 Applications Benefits Tool for creating realistic facies architectures in 3D reservoir models Flexibility & Speed • Multiple sources of hard data such as well data, seismic, trends, regions RMS Facies Object Based Methods Facies Elementary Input Data • A 3D grid layout (volume and resolution) • Well data (facies body distribution) • Trend lines / surfaces Emerson Confidential • 4/30/2018 • Slide 75 Applications Benefits Stochastic reservoir modelling during the exploration stage • Include deterministic elements when moving from appraisal to production Combines both deterministic and stochastic approaches with object modelling. Integrated with the internal programming language IPL. Exercises#1 Facies Modelling – Fluvial Environments • Case 1 : High NTG, few wells, basic sand / shale facies interpretation Project : Ruby.pro, Grid Model → Fluvial • Use of Facies Belts as a first pass model with Proportion mode • Use of Facies Channels with a conceptual model • Case 2 : Extensive well data, seismic attribute available Project : Sapphire.pro, Grid Model → Mod_Grid Zone. Zone 2 • Use of Facies Indicators with VPC trend, seismic conditioning Emerson Confidential • 4/30/2018 • Slide 76 Case 1 : High NTG, few wells, basic sand/shale facies interp. Project : Ruby.pro, Grid Model → Fluvial Use of Facies Belts as a first pass model with Proportion mode Use of Facies Channels with a conceptual model Background & Available Data: • 4 wells with GR, PHIE, Facies_Fluvial logs • Facies interpretation kept simple – based on PHIE cutoff • Sand [>7.5% PHIE] ~ combines poor & good sands • Shale • Environment: coarse braided fluvial system (from regional studies) • Channel orientation: SE-NW [N300] • Well statistics: relatively high NTG [46% of facies is sand] Emerson Confidential • 4/30/2018 • Slide 77 Case 2 : Extensive well data, seismic attribute available Project : Sapphire.pro, Grid Model → Mod_Grid Zone. Zone 2 Use of Facies Indicators with VPC trend, seismic conditioning Background & Available Data: • Structure: simple rotated fault block, depth around 2000 m • 2 main reservoir zones bounded by 2 thick field-wide marine shales • 57 wells with GR, RHOB, Por, Perm, Facies logs • Upper zone: good, relatively homogeneous sands deposited in • • • • • shoreface environments Lower zone: heterogeneous fluvial environment, with available Vsh map Paleogeographic setting: widespread alluvial floodplain with a hinterland to NNE Four main facies: shales, clean sands, shaley sands, coals Field-wide FS separates the 2 reservoir units Common OWC at 2290 m Emerson Confidential • 4/30/2018 • Slide 78 Emerson Confidential • 4/30/2018 • Slide 79 Exercises#2 Facies Modelling – Shoreface & Delta • Case 1 : Simple shoreface environment with stacked belts Project : Ruby.pro, Grid Model → Shoreface • Use of Facies Belts to model the transitional units • Use of Facies Composite to represent the spit deposits, using trends • Merge facies realisations • Case 2 : Complex fluvial dominated delta with stacked belts, fluvial channels and mouthbar facies • • • • Project : Sapphire.pro, Grid Model → Delta Use of Facies Belts for the large scale deltaic belts Model the fluvial channels using Facies Channels Model the mouthbars using Facies Composite Merge facies realisations Emerson Confidential • 4/30/2018 • Slide 80 Exercises#3 Facies Modelling – Turbidite Environment • Case : Turbidite Reservoir Project : Ruby.pro, Grid Model → Turbidite • Basic Facies Composite job set up • Use of trends in Facies Composite • Model the turbidite objects using the backbone shape Emerson Confidential • 4/30/2018 • Slide 81 Exercises#4 Turbidite modelling with SedSeis • Case : Turbidite Reservoir Project : SedSeis.pro, Grid Model → Sedseis • Pixel mode • Parametric mode • Turbidite modelling with intra body trends Emerson Confidential • 4/30/2018 • Slide 82 Case Study : Ruby Field • There are 3 main reservoir zones in the Ruby Field − An upper zone deposited in a coarse braided fluvial system − An intermediate zone deposited in a shallow marine environment − A lower heterogeneous zone deposited in a deep-water turbidite environment • Two reservoir lowest units are separated by a thick mud-rich slope sequence • There are different OWCs in the two reservoirs − 2010 m in the upper shoreface unit − 2270 m in the lower turbidite unit • Two main faults have been mapped within the field Emerson Confidential • 4/30/2018 • Slide 83 Case Study : Sapphire Field • • The structure of the Sapphire field is a simple rotated fault block that lies at a depth of around 2000 m There are two main reservoir zones bounded by two thick field-wide marine shales: − The upper zone comprises good, relatively homogeneous sands deposited in shoreface environments − A lower heterogeneous zone has been deposited in a fluvial environment. • • • • The paleogeographic setting for the unit is a widespread alluvial floodplain with a hinterland to the NNE Four main facies have been identified in this zone: Shales, Clean sands, Shaley sands and Coals A field-wide flooding surface separates the two reservoir units There is a common OWC for both reservoirs at a depth of 2290 m Emerson Confidential • 4/30/2018 • Slide 84 Case Study : Turbidite (Gulf of Mexico) • Synthetic example • Fault bounded structure • 6 exploration and appraisal wells • Reservoir interval 30-150 m thick • Focused entry point for turbidites • Basin floor topography reflected in isochore OWC Emerson Confidential • 4/30/2018 • Slide 85 Thickness map Entry point for turbidites Better Decisions. Better Returns. Ruslan Nasibullin Senior Geologist [email protected] www.roxarsoftware.com