Advanced Facies Modelling Manual (COMPLETE) PDF

1 Training Exercise Advanced Facies Modelling Clastic Environments 2 Contents 1 INTRODUCTION ....................................................................................... 5 1.1 Overview ......................................................................................................

1 Training Exercise Advanced Facies Modelling Clastic Environments 2 Contents 1 INTRODUCTION ....................................................................................... 5 1.1 Overview .......................................................................................................... 5 1.2 Course Objective .............................................................................................. 5 1.3 What is a reservoir model? ............................................................................... 5 1.4 The value of a reservoir model ......................................................................... 6 1.5 The elements of the reservoir model ................................................................ 6 1.6 Designing the reservoir model .......................................................................... 8 1.7 Risk analysis and assessing the uncertainty of the reservoir model.................. 8 1.8 Petrophysical property modelling ...................................................................... 9 1.9 Modelling of petrophysical properties in Roxar RMS ...................................... 11 2 FACIES:BELTS .......................................................................................13 2.1 Overview ........................................................................................................ 13 3 FACIES:COMPOSITE .............................................................................15 3.1 Overview ........................................................................................................ 15 4 FACIES:CHANNELS ...............................................................................16 4.1 Overview ........................................................................................................ 16 5 FACIES:INDICATORS.............................................................................18 5.1 Overview ........................................................................................................ 18 6 MERGING OF FACIES PARAMETERS ..................................................19 6.1 Overview ........................................................................................................ 19 7 FACIES MODELLING – FLUVIAL ENVIRONMENTS .............................20 7.1 Concepts ........................................................................................................ 20 7.2 Session Objectives ......................................................................................... 21 7.3 Case 1: High NTG, few wells, basic sand/shale facies interpretation .............. 21 7.3.1 Exercise 1: Investigating the data .......................................................................... 22 7.3.2 Exercise 2: First pass model using Facies:Belts with proportion mode................. 25 7.3.3 Exercise 3: First pass model. Influence of the Variogram type on the lateral facies continuity (OPTIONAL) ..................................................................................................... 29 7.3.4 Exercise 4: Create and edit a Bodylog .................................................................. 30 7.3.5 Exercise 5: Refined model using Facies:Channels and Bodylog .......................... 38 7.3.6 Exercise 6: QC of results ....................................................................................... 43 7.3.7 Exercise 7: Use Crevasses with Facies:Channels (OPTIONAL) ........................... 46 7.3.8 Exercise 8: Use correlation with Facies:Channels (OPTIONAL) ........................... 49 Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 3 7.4 Case 1: Extensive well data, seismic information available ............................ 56 7.4.1 Exercise 1: Create Vertical proportion curve, to be used as a vertical trend for facies modelling ........................................................................................................................... 58 7.4.2 Exercise 2: Edit a Vertical proportion curve ........................................................... 60 7.4.3 Exercise 3: Create a Variogram model .................................................................. 61 7.4.4 Exercise 4: Model the fluvial environment using Facies:indicators ....................... 66 7.4.5 Exercise 5: Facies:indicators with seismic conditioning ........................................ 70 7.4.6 Exercise 6: QC of the facies modelling results ...................................................... 73 8 FACIES MODELLING – SHOREFACE AND DELTA ENVIRONMENTS 76 8.1 Concepts ........................................................................................................ 76 8.2 Session Objectives ......................................................................................... 77 8.3 Case 1: Simple shoreface environment with stacked belts facies ................... 77 8.3.1 Exercise 1: Basic Facies:Belts job set-up – No Well Conditioning ........................ 78 8.3.2 Exercise 2: Stacked Belts – progradational directions (OPTIONAL) ..................... 85 8.3.3 Exercise 3: Stacking Angle (OPTIONAL) .............................................................. 87 8.3.4 Exercise 4: Stacking Belts – Curved Boundaries (OPTIONAL) ............................. 87 8.3.5 Exercise 5: Stacking Belts – Interfingering ............................................................ 91 8.3.6 Exercise 6: Stacking Belts constrained to Well Data ............................................. 94 8.3.7 Exercise 7: Use predefined polygons as belts boundaries (OPTIONAL) ............ 101 8.3.8 Exercise 8: Manual editing of a 3D facies parameter (OPTIONAL) .................... 103 8.3.9 Exercise 9: Facies log generation based on simple PHIE cut off (OPTIONAL) .. 104 8.3.10 Exercise 10: Create intensity and azimuth trends to position the Spit facies in the Shoreface environment, using Facies:Belts report parameters ..................................... 106 8.3.11 Exercise 11: Create intensity and azimuth trends to position the Spit facies in the Shoreface environment, using Facies:Belts report parameters ..................................... 109 8.3.12 Exercise 12: Merge facies parameters to create a final shoreface facies model 112 8.4 Case 2: Complex Fluvial dominated Delta, with stacked belts, fluvial channels and mouthbar facies (OPTIONAL) ............................................................................. 114 8.4.1 Exercise 1: Investigate the data ........................................................................... 117 8.4.2 Exercise 2: Create a combined facies log to define the facies belts ................... 119 8.4.3 Exercise 3: Model the large scale deltaic facies belts ......................................... 121 8.4.4 Exercise 4: Model the channel and crevasses .................................................... 126 8.4.5 Exercise 5: Create a trend to position the mouthbar objects in the appropriate facies belts, close to channel objects ................................................................................................. 132 8.4.6 Exercise 6: Model the mouthbar objects .............................................................. 135 8.4.7 Exercise 7: Merge the facies to get the final facies model .................................. 139 8.4.8 Exercise 8: Result QC .......................................................................................... 141 9 FACIES MODELLING – TURBIDITES ..................................................143 9.1 Turbidities Key Characteristics ..................................................................... 143 9.2 Session Objectives ....................................................................................... 144 9.3 Data and Conceptual model ......................................................................... 145 Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 4 9.3.1 Exercise 1: Well data and data analysis .............................................................. 146 9.3.2 Exercise 2: Facies:Composite panel and basic job set-up .................................. 150 9.3.3 Exercise 3: Using trends in Facies:Composite .................................................... 156 9.3.4 Exercise 4: QC trends using ‘Average map’ ........................................................ 161 9.3.5 Exercise 5: User defined shapes ......................................................................... 162 9.3.6 Exercise 6: Turbidite modelling using ‘backbone’ objects ................................... 165 10 SEDSEIS .............................................................................................169 10.1 Background............................................................................................ 169 10.2 Project ................................................................................................... 169 10.2.1 Exercise 1: Pixel mode ........................................................................................ 170 10.2.2 Exercise 2: Parametric mode ............................................................................... 171 10.2.3 Exercise 3: Intra body trends ............................................................................... 172 Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 5 1 INTRODUCTION 1.1 Overview Roxar RMS is a tool designed to support all aspects of reservoir management. It allows the geoscientists to combine their knowledge with all available data from the reservoir, such as well data and seismic data, into a consistent 3D reservoir model. Roxar RMS further has the functionality to extract the decision support information, such as volumes, well plans and production profiles, required for prudent reservoir management. With the best workflow management in the industry, a modern user interface, unique analysis and visualisation tools, Roxar RMS provides an environment where the reservoir disciplines together can maximise the output from their reservoir. 1.2 Course Objective The objective of the course is to give the user an understanding of the principles of building reservoir models of complex geological environments. This will be done through a series of practical exercises resulting in a number of different geological models. The emphasis will not only be on learning how to use the software toolbox, but also the workflows and methods required to build a successful 3D property model. 1.3 What is a reservoir model? For the purpose of this course the following definition will be used for the term reservoir model: ‘A reservoir model is a consistent representation of all data and knowledge about a reservoir relevant to the management of the reservoir.’ Normally the reservoir model consists of the following elements:  A description of the structural framework  A description of the petrophysical properties of the reservoir  A description of the initial distribution of fluids and pressures  A description of the dynamic fluid behaviour and properties Reservoir modelling is the process of building and maintaining the reservoir model. For the reservoir model to be as effective as possible in supporting the management of a reservoir it needs to include all available data from the reservoir and the combined knowledge of the professionals interacting with the model. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 6 It is therefore critical that the model is straightforward to build, can be maintained effectively and it is easy to extract the decision support information necessary for prudent management of the reservoir. As the human and financial risk involved with hydrocarbon production is very high it is also critical that the uncertainty of the reservoir model can be realistically assessed. 1.4 The value of a reservoir model Using a reservoir model, as defined above, adds value to the management of an asset by maximising the use of available data and introducing better decision-making processes. By integrating technical understanding with available data the reservoir model provides an environment in which a multi-disciplinary team can improve their understanding of the reservoir. This will allow them to see new opportunities and avoid potential dangers when managing the reservoir. By combining this knowledge and data the reservoir model will give the best possible answer to questions regarding volumetric estimation, drainable volumes, production profiles, optimal well locations etc. Correctly interpreted this will allow the reservoir team to make the optimal decision at any time in the reservoir life cycle. 1.5 The elements of the reservoir model As mentioned in the definition, the reservoir model normally consists of the following components:  A description of the structural framework  A description of the petrophysical properties of the reservoir  A description of the initial distribution of fluids and pressures  A description of the dynamic fluid behaviour and properties The emphasis and even the presence of these components will strongly depend on the purpose of the reservoir model. The Structural Framework The structural framework consists of the surfaces bounding the reservoir intervals, the faults and the relationship between them. The structural framework is normally based on an interpretation of seismic data, well observations of surfaces and faults and a model of the relationship between them. Typical challenges in establishing a structural framework are depth conversion of the seismic data, interpreting the relationships between faults and combining all of the information into a consistent model. The Petrophysical Properties Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 7 The petrophysical model describes the porosity and permeability distribution in the reservoir. There are normally two approaches to arriving at a petrophysical model. The simplest approach is by distributing the properties based on statistical inference, either by interpolation or data driven geostatistical techniques. A more sophisticated technique is to combine a spatial model of the depositional environment and diagenetic effects with the well observations. This latter approach is based on geostatistical techniques and often involves building a facies model to guide the distribution of petrophysical properties. Typical challenges in establishing the petrophysical model can be to understand how the depositional and diagenetic processes impact on the rock and fluid properties and how to combine the well data and the conceptual geological model. The Initial Fluid and Pressure Distribution The initial fluid distribution is a full description of the presence of oil, water and gas in the reservoir. Associated with the fluid distribution is the initial pressure distribution of the reservoir. The fluid distribution is typically based on fluid contacts observed in the wells and a model for the fluid presence as a function of the height above these contacts. The pressure is based on measurements in the wells. In the exploration stage a challenge in establishing the fluid distribution can be that the contact has not yet been observed. For the pressure description a challenge can be identifying whether the reservoir has one pressure regime or is divided into several isolated compartments. The Dynamic Fluid Behaviour and Properties The dynamic fluid behaviour is a description of the fluid movement, through the reservoir and into the wells and to the wellhead, both as a function of historic production and prediction for future production. To describe dynamic fluid behaviour in the reservoir an understanding of the dynamic properties is necessary. This includes the pressure-volume-temperature (PVT) relationships for the fluid composition in the reservoir and the relative permeability (rel-perm) for each fluid for the rock types of the reservoir. These can be measured in laboratories or taken from similar reservoirs. The fluid behaviour from the reservoir to the wellhead is described by lift-curves, giving the pressure drop as a function of dynamic factors. A significant challenge in describing the fluid properties is obtaining correct descriptions of permeability and relative permeability. For fields on production the biggest challenge in describing the dynamic fluid behaviour is to match the historic production from the individual wells and the whole field. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 8 1.6 Designing the reservoir model As mentioned above the purpose of building a reservoir model is to help manage the reservoir. It is therefore critical that the model is designed to support the objectives at hand. If the challenge is to develop an appraisal well the requirements of the model will be significantly different than for an improved oil recovery project on a mature field. The first step of the reservoir modelling process should be to investigate what is the purpose of the exercise, and what are the critical factors that must be handled. As the objective in the first case is to validate the structural constraints of the reservoir, the depositional environment and the fluid distribution, most work should probably be spent on investigating different scenarios for depth conversion and property models. In the fairly constrained environment of a field on decline the main challenges may be the detailed geological model and the fluid behaviour. The emphasis should then be placed on understanding the fine-scale geology and the impact of different recovery mechanisms. The design of the reservoir model therefore needs to include all parties that will influence or will be affected by the model. In other words, geoscientists and engineers have to team up and collectively answer the following questions:  Which decisions do we need to make on the basis of the model?  What is the information that we need from the model to make these decisions?  What is critical to secure the validity of this information?  How do these critical issues translate to the different elements of the reservoir model? Once these questions have been answered each of the disciplines need to take a critical look at how to incorporate their areas of responsibility into the model, keeping in mind that over-modelling may be just as counterproductive or damaging as being too simplistic. 1.7 Risk analysis and assessing the uncertainty of the reservoir model As already mentioned, there is significant human and financial risk associated with reservoir management. One of the main reasons for this risk is the significant uncertainty associated with the reservoir model. The uncertainty of the reservoir model ranges all the way from measurement uncertainty of the core data to uncertainty in whether the chosen well configuration will be optimal to drain the reservoir. The uncertainty of the reservoir model therefore needs to be assessed to properly analyse the risk. The uncertainties of a reservoir are many and have such varying impact that it is necessary to know what to look for. The critical risk factors should be kept in mind when designing the model such that the appropriate uncertainties can be understood. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 9 Uncertainty will normally occur at two different levels; is the scenario we are looking at correct and if so, how certain are we of the details? The importance of each level will vary for each property we are dealing with and need to be investigated. Scenario Uncertainty The scenario level uncertainty may be addressed by looking at different values of critical variables.  The number of wells in the development plan  Is this a prograding or a retrograding environment?  Is the central fault open or closed? To fully understand uncertainty at this level we need to model each of the scenarios and look at the implications on the decision we are trying to make. If the number of variables becomes exhaustive techniques like Experimental Design and Analysis of Variance will assist in improving the analysis. Parameter Uncertainty The best way to address the impact of parameter uncertainty is to create a geostatistical model for the parameter and simulate multiple realizations. This can easily be done for important parameters like:  Reservoir surfaces  Facies models  Petrophysical models. Total Uncertainty For a complete picture these two levels of uncertainty need to be combined into a total uncertainty model. This will make it possible to span the entire uncertainty of the reservoir model associated with any response parameter. It must be stressed that the uncertainty analysis should be closely linked to the design of the reservoir model, and the objectives behind the design. It is easy to get lost trying to describe and understand uncertainties that are not relevant to the purpose of the model. Remember: over-modelling can be just as counterproductive as being too simplistic. 1.8 Petrophysical property modelling The key focus when performing petrophysical modelling, as with any other modelling, is building a model that is fit for purpose. The petrophysical model needed for doing quick volumetric analysis should be vastly different from the model upon which major field development will be based. By building a model that is fit for purpose, modelling resources will be optimized and better answers will be the result. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 10 Designing the Petrophysical Property Model As mentioned previously the following questions need to be answered when designing the model:  Which decisions do we need to make on the basis of the model?  What is the information that we need from the model to make these decisions?  What is critical to secure the validity of this information?  How do these critical issues translate to the different elements of the reservoir model? If these questions have been answered properly you should have a fairly good understanding of how to focus the modelling effort. For instance, if the objective of the study is to get updated volumetrics and well control is reasonably good then a relatively quick 2D interpolation of porosity will suffice. If, however, the challenge is to explain unexpected water flooding in a well, the solution could be to build a detailed model of the facies to look for thief zones etc. As all fields are different and the objectives of modelling vary, it is difficult to generalize principles for model design. Some general directions may be:  If the objective of the study is to provide input to flow simulation it is important to explicitly model and capture the heterogeneities that influence fluid flow.  In a mature field with abundant well, seismic and production data the focus should be on integrating and honouring all of the data.  In the exploration phase, with limited well data, more emphasize should be placed on the conceptual geological model.  In deepwater environments, where wells are expensive, it is important to model the sediment distribution in time and space when well planning. Once the model has been designed it is necessary to select which modelling methodology should be used. Does the model call for simple interpolation, advanced object modelling conditioned to seismic data or trend based petrophysical simulation? Experience and knowledge of the available techniques are the best basis for making this decision. The objective of this course is to provide some of the knowledge and experience required to make the right choice. Reassessing the Data Interpretation Once the model has been designed and the modelling techniques have been selected it is necessary to revisit the interpretation of the data:  Do we have the logs we need for the wells (for instance facies)?  Is the resolution of the interpreted logs correct or is it too coarse or too detailed?  Is the zonation of the reservoir inappropriate to the modelling technique? An advanced facies model can often handle significantly thicker intervals than a traditional 2D layer cake approach  Does the grid resolution capture the necessary detail, without being too detailed?  Do we have the appropriate seismic attributes as conditioning data for the modelling? Is, for instance, acoustic impedance available and correlated to porosity?  Do we understand the relationship between different parameters, e.g. poro/perm etc. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 11 Building the Model If the model has been thoughtfully designed and the relevant data are available, the modelling itself should be fairly straightforward. Only a few things need to be remembered:  Do a thorough data analysis to quantify model parameters. Keep in mind that the stochastic modelling is performed in a regularised simulation domain. Make sure that data analysis is performed in this domain.  Start with a basic model that only includes the main features of the conceptual model without any data constraints (unless the model is based on interpolation)  QC the result and remodel until the result is satisfactory  Refine the model step by step until it gives a realistic representation of the model  Add geological detail  Add conditioning data like seismic and wells QC the result after each constraint is added and making sure that the constraint is honoured before further constraints are added. Taking a systematic approach to the modelling and confirming the results at every stage is the most efficient way of building a quality model that is fit for purpose. If something is rushed and too many things are introduced at the same time it is difficult to troubleshoot if the result is not correct. At any stage keep the focus on the end result and make sure that the modelling objectives are met. 1.9 Modelling of petrophysical properties in Roxar RMS Roxar RMS has a wide variety of modelling techniques, ranging from interpolation techniques to advanced object modelling, available to the modeller. Properly used these techniques may be employed, individually or in combination, to build property models fit for any reservoir management purpose in any geological environment. As mentioned previously there are two approaches to modelling the distribution of petrophysical properties in the reservoir, statistical inference and description of depositional and diagenetic processes. Roxar RMS has a variety of options available for each approach such that the model may be adapted to the reservoir and the modelling objectives. The main objective of using statistical inference to describe petrophysical properties is to reproduce a statistical measure of the data. This measure may be the distribution of the data or the spatial continuity of the data. The following techniques are available in Roxar RMS:  2D interpolation  3D interpolation  2D and 3D Kriging  Sequential Gaussian Simulation (SGS) Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 12  Sequential Indicator Simulation (SIS)  Cloud transform When trying to take the depositional environment and diagenetic processes into account when describing the petrophysical properties the important features of the conceptual geological model need to be considered. This will often require a staged approach where the large-scale variation is captured in a facies model. It may also be necessary to combine several different techniques to achieve the required refinement of the model. In addition to the techniques above, the following techniques are available in Roxar RMS.  Facies-associated modelling (belts)  Object modelling  Depositional and diagenetic petrophysical trends Both approaches will honour the well data and can be conditioned to seismic attributes. If the petrophysical properties are used as input for dynamic fluid analysis the resolution needs to be reduced from geological modelling scale to flow simulation scale. Roxar RMS has state of the art techniques to preserve the fluid flow properties of the petrophysical properties in this upscaling. When the petrophysical properties have been described it is possible to extract more precise information in the form of net pore volumes of the reservoir, drainable volumes from well locations, well targets based on for instance net sand analysis etc. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 13 2 FACIES:BELTS 2.1 Overview ’Facies:Belts’ is a grid based stochastic facies modelling tool originally designed to model transitional geological environments in progradational and retrogradational depositional systems. The model uses progradation directions and stacking angles as input to build a framework for each facies belt. The boundaries of the facies belts are simulated stochastically. A variety of equiprobable output models are possible. Each model will honour the well data, but will display differences in the interfingering between the different facies belts. ’Facies:Belts’ can be used to simulate a number of different geological environments, including shoreface reservoirs, deltaics and carbonate reef deposits. The flexibility of the algorithm allows the facies boundaries to be straight or curved, parallel or divergent, enabling the user to model a variety of environments. ’Facies:Belts’ is often used in addition to an object based method to define the large-scale facies framework of the reservoir zone, which is then used as a background to further object based facies modelling. The objects are used to describe smaller-scale heterogeneities conditioned on the facies-belt distribution. It also allows easy modelling of facies environments where the facies’ volume proportions vary vertically, laterally, or both. A ‘Trend and Threshold’ option can be used to produce first-pass basic seismic conditioning. The statistical model used in ’Facies:Belts’ is the truncated Gaussian simulation where various trend and threshold functions can be incorporated (the algorithm is described in the RMS user guide manual). Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 14 Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 15 3 FACIES:COMPOSITE 3.1 Overview Facies:Composite is an object-based facies modelling algorithm that simulates facies bodies as geometrical objects. The modelled objects have a defined shape and the algorithm can use well data, trend maps or seismic attributes to position the objects within the reservoir. The sizes and orientation of the bodies are drawn from distributions specified by the user. The algorithm uses an iterative simulation technique to ensure correct conditioning to well data and correct use of complex trend and seismic information. ’Facies:Composite’ is a very flexible tool which is ideally suited for modelling reservoir architecture where the important heterogeneities are defined by distinct facies bodies. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 16 4 FACIES:CHANNELS 4.1 Overview The RMS Stochastic Facies:Channels module is an advanced object based facies algorithm that can model complex channel depositional systems. Facies:Channels is a modeling tool that allows lateral and vertical trends in the distribution and geometry of modelled channels. While objects in Facies:Composite have a predefined length (causing the objects not to cross the reservoir in most cases), channels generated in Facies:Channels will always cross the entire reservoir. The basic concept is that the facies within a zone can be subdivided into a background facies and channel objects. Optionally, crevasse splays and intra-channel barriers can be modelled. These are linked to the channel facies. The crevasses will be attached to the channel margin, whereas the barriers will be positioned inside the channel body. The modelling is done on a 3D-geological grid with fault splits. The orange is the background facies and the green is an object facies. Two main operating modes are present, the ’sand-body’ mode and the ’multi-channel-beltmode’. The ’sand-body’ mode is used for single cut-and-fill channel bodies, and for direct modelling of channel belt geometries. ’Multi-channel-belt’ mode is used for modelling multiple channels within channel belts. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 17 The modelling can be conditioned to facies logs, object logs and 3D seismic attributes. The algorithm is flexible for incorporating trends such as for position and size of the object facies. The resulting facies parameter is typically used as input to stochastic petrophysical modelling. It is also possible to do hierarchical modelling by merging a Facies:Channels parameter with other discrete parameters, e.g. a facies parameter from Facies:Belts or a facies parameter from Facies:Composite. Facies:Channels includes a previewer which gives instant feedback on the user input. The previewer performs a simplified 2D simulation based on the current parameter settings. This allows the user to tune the parameter input prior to performing actual simulations. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 18 5 FACIES:INDICATORS 5.1 Overview Facies:Indicators is a flexible pixel-based modelling technique that samples the local conditional probability distribution for each grid cell. The method can incorporate any type of trend, and conditioning to a 3D seismic attribute. The algorithm is ideally suited when conditioning to a large number of wells. It is often used for reproducing irregularly shaped facies bodies. Facies:Indicators major benefits include:  Flexibility. The Facies:Indicators method allows the generation of very flexible facies patterns, for any number of facies. It also allows conditioning of the results to well and seismic data.  Speed. The method provides fast results, irrespective of the number of wells. It is therefore especially suited to modelling mature fields with a large number of wells. A wide variety of 1D, 2D and 3D volume fraction trends can be used as input, and any number of facies can be modelled. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 19 6 MERGING OF FACIES PARAMETERS 6.1 Overview Merging of facies parameters is used to combine two facies parameters into one. The parameters are merged together honouring erosion rules. This functionality is particularly useful for combining the large scale facies framework simulated using ’Facies:Belts’ with the finer heterogeneity simulated using ’Facies:Composite’ or ‘Facies:Channels’. Only facies parameters from the same zone can be merged. The parameters can be either standard facies parameters or BodyFacies parameters; for BodyFacies parameters, the parametric body information is retained in the merged parameter. If one or both of the input parameters are BodyFacies parameter, the output parameter will be BodyFacies format. The parameters are merged honouring an erosion scheme defined in the list boxes. A merged realization can be merged again with another realization. The merge function is also available for separate zones where different erosion rules can be given for each zone. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 20 7 FACIES MODELLING ENVIRONMENTS – FLUVIAL 7.1 Concepts When modelling fluvial environments, it is important to know which fluvial system you are trying to model. It can either be:  Coarse alluvial systems – fans and braidplains  Sandy fluvial systems – meandering, anastomosed and braided rivers The typical facies represented in a fluvial system are:  Channels: stacked or isolated; active or abandoned  Overbank: crevasse splays and sandflats  Flood plain: background sedimentation The information should come from analogue data, field studies and regional knowledge. Cores and logs provide detailed studies and calibration. Resulting in a core constrained, log derived facies code for every measurement level in the well. Analogue data have provided global statistics for channel width to thickness relationships depending on type of fluvial system (references: fielding et al.). Rarely does the measured sand body thickness (t) from logs equal the channel depth (h): t ~ 0.55h Relationship of channel depth to width has been defined as:  Meandering channels ~ 0.95h^2  Average of all data ~12h^1.85  Braided channels ~513h^1.35 These averages should be defined locally as data becomes available. Different Facies modelling techniques can be applied, depending on the available input data, number of wells in the project, NTG, presence of trends etc… The most appropriate RMS technique to model fluvial environments is Facies:Channels. But in some cases, other techniques may produce a better result and should not be completely discounted. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 21 This section will provide a few examples of different Facies Modelling techniques being used for fluvial environments. 7.2 Session Objectives Different cases will be investigated as Fluvial environments, and different facies modelling techniques will be used according to the various data set up. For a given depositional environment, different modeling techniques can be applied. The choice must be made according to the amount and type of data available, existence of a conceptual model or not, the expected facies geometry/connectivity… etc. The first example has little data and will require a basic conceptual model. Facies:Belts using proportion mode, and Facies:Channels algorithms will be used. In the second example, there is a lot of data, and Facies:Indicator algorithm (more data driven) will be used. 7.3 Case 1: High NTG, few wells, basic sand/shale facies interpretation RMS Project – Ruby.pro Grid Model: Fluvial In this case, we only have 4 wells available. GR, PHIE and Facies_Fluvial logs are available. The facies interpretation was kept very simple, based on a PHIE cut off, with 2 facies: shale and sand (‘sand’ >7.5% PHIE). Note that the ‘sand’ facies combines poor and good sands. The environment has been interpreted from regional studies as a coarse braided fluvial system. The channels are oriented South East-North West, (N300). Well statistics show a relatively high NTG. Almost 46% of the facies volume fraction is sand. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 22 7.3.1 Exercise 1: Investigating the data Open the project: ► File ► Open Project ► Ruby.pro Observe the BW Statistics of the Fluvial reservoir: ► Grid models ► Fluvial ► MB1 on BW Facies_Fluvial in the Data tree ► From the Task pane select Statistics in the BW Fluvial_Facies task group Note : Blocked wells statistics give very useful information about the volume fractions of the different facies observed at the wells, their average thickness (and also min, max and Standard deviation). These values can be used later on when setting up the facies modelling job. Check the blocked volume fraction (i.e percent), average, min and max thickness, and Std deviation, and compare with the original well data (before well blocking). Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 23 Well Log Editor/calculator: ► Grid models ► Fluvial ► MB1 on BW Facies_Fluvial in the Data tree ► From the Task pane select Well log editor/calculator in the BW Fluvial_Facies task group ► Display the Facies_Fluvial, PHIE_Fluvial, GR_Fluvial logs as curves and as background display ► Investigate the different well data Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 24 Note : When a Set up is applied, it is stored in the project after saving Use the Multi-well viewer: ► Grid models ► Fluvial ► MB1 on BW Facies_Fluvial in the Data tree ► From the Task pane select Multi-well viewer in the BW Fluvial_Facies task group ► Display all 4 wells, and GR and facies logs Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 25 7.3.2 Exercise 2: First pass model using Facies:Belts with proportion mode For this first pass approach, a facies model must be run quickly, and there is no strong conceptual model ready at the time of the study. As object modelling techniques require a lot of geometry information, based on a conceptual model, they could not be applied at this stage of the project. The alternative used here is based on a pixel modelling technique, Facies:Belts using Proportion Mode to represent lenses of facies. Open the Facies:Belts modelling job: ► Grid models ► Fluvial ► MB1 on Grid in the Data tree ► From the Task pane select Belts from the Facies modelling task group First, rename the Job to ‘Proportion_Mode’: ► Job dialogue box ► MB1 ► Rename… Then in the General tab: ► Define the output parameter name: belts_proportion ► Select the BW and Condition on well data Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 26 ► Use the Facies_Fluvial log ► Select sand, then shale facies to model using the arrow (make sure the sand facies is on the top of the list) In the Geometry tab: ► Select the Proportions geometrical class ► Select the proportion trend as ‘Lenses’ ► Specify the Volume fraction of sand: 0.46 (as noted from the Block wells statistics) In the Interfingering tab: ► Define the azimuth as 300 (N300 was provided from regional interpretation) ► Ranges: Parallel to Azimuth = 1000m Perpendicular to Azimuth = 300m In depth = Mean: 7m, Std Dev: 4m (Normal dist.) Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 27 Note : The variogram ranges will define the lateral continuity of the sand facies. In this case, we want to represent elongated sand patches, so they look closer to channel objects. Without enough regional information in this case, the dimensions of the elongated sand patches can only be set similar to standard analogues. Hence, the main range (parallel to azimuth) has a value of 1000m, the perpendicular range a value of 300m. For the depth range, the BW statistics can be used to set the Mean and Standard Deviation. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 28 ► Leave the other Tabs as default ► Save and Run the job ► Display the results in a 3D view It may be necessary to change the visual settings of the facies parameter to display the appropriate colour coding: ► Grid models ► Fluvial ► belts_proportions facies parameter ► MB3 ► Visual Settings Observe the preferential orientation imposed by the variogram, and the ‘pixel’ look of the facies parameter (Facies:Belts is a pixel based algorithm…). If this can be acceptable as a quick first pass facies model, you can immediately see the limitations on the facies continuity. This can be improved by changing the variogram settings (see next exercise). But object modelling using Facies:Channels will give much better results for facies continuity (described later with several exercises). Save the project as Ruby_<your_name>.pro: ► File menu ► Save project as… Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 29 7.3.3 Exercise 3: First pass model. Influence of the Variogram type on the lateral facies continuity (OPTIONAL) In order to get more continuity for the sand facies the Variogram Type must be changed. First, open the Facies:Belts job ‘Proportion Mode’: ► Grid models ► Fluvial ► MB1 on Grid in the Data tree ► From the Task pane select Belts from the Facies modelling task group ► Rename the output parameter ‘belts_proportions_GE’ ► Save As (to duplicate the job) and name the copy: ‘Proportions_GE_Vario’ ► OK Note : When you want to re-use most of the settings from a modeling job, but create a new output parameter, the easiest way is to change the name of the output parameter in the first job, then duplicate the job, discarding any prompted changes. In the Interfingering tab: ► Choose a General exponential variogram, with Exponent 1.8 ► Leave the other values Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 30 ► Execute and display the created Facies parameter with the appropriate colour table ► Compare with the previous results using a Spherical variogram Observe that the sand patches look much more continuous when using the General Exponential variogram. Note : The General Exponential variogram can be used in Facies:Belts with exponents between 1.5 (then ~ similar to the Spherical variogram) and 1.99 (then ~ equivalent to the Gaussian variogram). The higher the exponent, the more continuous the sand patches will be. 7.3.4 Exercise 4: Create and edit a Bodylog The objective is now to model channel objects, using Facies:Channels, an object based modelling technique. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 31 Because the volume fraction of sand is relatively high from well observations (46% sand) it is essential to identify the different channel bodies at the wells to make sure stacked channels can be accurately represented in the simulation. When the volume fraction of sand is high, it is very likely that sand objects are stacked vertically. In order to represent channel objects in a realistic way with object modelling technique it is very important to have a bodylog. If a bodylog is not used, the simulated channel objects will have unrealistically big thickness Bodylogs can be created directly on the original well data, or on the BW data. In this example, we will create the bodylog directly on the blocked wells. Open the Well log editor/calculator: ► Grid models ► Fluvial ► MB1 on BW Facies_Fluvial in the Data tree ► from the Task pane select Well Log editor/calculator in the BW Fluvial_Facies task group ► Press the ‘Create’ button Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 32 The ‘Create log’ panel will then pop up. Choose the following options: ► Create BODY from FACIES log Direct ► Select the ZONELOG, and Facies_Fluvial logs ► Select the ‘sand’ facies to be the object facies ► Type new log name ‘Bodylog_Fluvial’ ► Select the option ‘Associate with facies log’ ► Press OK Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 33 The ‘direct’ option to create a bodylog is the simplest. When using thickness criteria, much more information is required, implying that a strong conceptual model is available. With the ‘direct’ option, every sand interval is assigned a different body number. Then, the user needs to investigate the created bodylog, for each well, to identify possible stacked objects, and edit the bodylog where necessary. The editing process is quick when there are just a few wells in the project, but can become more tedious with lots of well data. However, in sand-rich systems, this is time well spent. Now check that the created bodylog is associated with the facies log: ► Zones ► Fluvial ► BW Facies_Fluvial ► MB3 ► Information The Facies_Fluvial log should have the Bodylog_Fluvial as its associated Object log. Now, still using the Well log editor/calculator, investigate the created bodylog, for each well, and look for possible stacked channel bodies. It may be necessary to change the visual settings of the created bodylog. ► In the Well log editor/calculator ► Press the Visual settings button Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 34 ► Then, still using the Well log editor/calculator, display the GR_Fluvial log as a curve, PHIE_Fluvial in the background left display, and the Bodylog_Fluvial in the background right display ► Select Mode : ‘Edit value’ ► Investigate each well in turn The GR log will help to identify channel objects. A typical channel signature from the GR log would be: low GR values at the base, progressively increasing towards the top. The bodylog on appraisal_1 and appraisal_2 needs to be edited (the 2 other wells could be left as they are). Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 35 To edit the value of the bodylog: ► First, make sure this log is displayed on the right background ► Use the mouse to define the BW cells that you interpret as a different channel body ► Type the body number you want to assign to this defined interval (Tip: check the number of created channel bodies, and start to add extra bodies from this number) ► Press the ‘Set to value’ button Note : The first observed body in a well must have the smaller number; otherwise the facies simulation job will show an error message. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 36 ► Repeat the process until satisfied with the facies and body interpretation Hint : in this simple example, it is necessary to edit the bodylog on wells appraisal_1 and appraisal_2. Discovery and trub_appr_1 wells bodylogs can be accepted as they are. The created and edited bodylog can now be used with an object based modelling technique. Save the project: ► File menu ► Save project… Hint (Optional) : To create bodylog on the original well data: ► Wells ► from the Task pane select Well utilities ► Log operations Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 37 Hint (Optional) : To edit bodylog on the original well data: ► Display appraisal_1 well in the correlation view ► Display created bodylog in the log track ► Press the ‘Edit discrete log’ button ► Press the ‘Create interval’ button ► MB1 at the depth were porosity indicates a separate channel ► MB3 to select the new code for new channel Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 38 7.3.5 Exercise 5: Refined model using Facies:Channels and Bodylog When a conceptual model is available, providing minimum information on channel dimensions, it is recommended that an object based facies modelling technique is used. This is particularly useful later on when performing petrophysical modelling, as intrabody trends are likely to be present. In this example it is assumed that more information is available for the channel objects, based on analogue studies and regional knowledge. Open the Facies:Channels modeling job: ► Grid models ► Fluvial ► MB1 on Grid in the Data tree ► From the Task pane select Channels from the Facies modelling task group In the General tab: ► Type in the output parameter name ‘channel’ ► Select Facies with body option ► Condition on BW data ► Select the sandbody mode ► Select from the available facies: sand to represent the Channel objects, and shale to represent the background Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 39 In the Volume fraction tab: ► Leave the Channel system volume fraction as ‘Global’ ► Type in the Volume fraction value: 0.46 (46% of sand, observed at the wells – see BW statistics) ► Increase the Tolerance: set it to 0.02 Note : The volume fraction value is typically derived from well observations. NB In the early exploration stage, what is observed at the wells may not be necessarily what is happening in the entire reservoir. A recommended approach is to use different scenarios to investigate possible sand volume fractions. Object modelling techniques require a minimum of flexibility to reach the Volume fraction of sand objects in the reservoir. This is governed by the ‘Tolerance’. In this case, a tolerance of 0.02 (quite a high value) means that the final simulated volume fraction of objects will be 46% +/- 2%. While a very small tolerance will not be a problem with just a few wells, it will be necessary to allow more flexibility when there is a lot of well data, so the algorithm can honour all the constraints (channels geometry, trends, well conditioning, well correlations etc…) without any problems. In the Geometry tab: ► Define the channel dimensions as follows: Thickness: Mean=7m, Std Dev=4m, Min=1m, Max=20m Width: Mean=300m, Std Dev=150m, Min=150m, Max=600m Amplitude: Mean=800m, Std Dev=300m Sinuosity: 1.15 Azimuth: Mean=300 deg N, Std Dev=10 deg Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 40 Channel dimensions should be defined using a conceptual model applicable to the reservoir. The information can come from the well observations (when there is a lot of well data) analogue studies, nearby reservoirs with plenty of information. Most oil companies have access to internal analogue/fields databases with such information. Analogue data has provided global statistics for channel width to thickness relationships depending on type of fluvial system. ► Press the Previewer button on the lower right corner of the job panel ► Inspect to investigate the possible channel outcomes The previewer will show possible outcome 2D channel geometries based on the chosen constraints, without having to run the job itself. This is particularly useful, helping to define Amplitude and sinuosity values, and also when a lot of well data will make the simulation run slowly. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 41 ► Test different Amplitude and sinuosity values and visualise results in the previewer ► Leave the other Tabs (Form/Repulsion, Crevasses/Barriers, Seismic) with the default values ► Press Save and Run the job When the job is executed, a status panel will pop up and the Channel system volume fraction will be updated while the job is running, and as iterations are performed. When the simulation is completed, scroll up through the history log to find the final volume fraction of the Channel objects. Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 42 Check the final volume fraction is compatible with the value specified in the job, within tolerance. ► Display the channel parameter in a 3D view, and play through the layers to visualise the results Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 43 Save the project: ► File menu ► Save project… 7.3.6 Exercise 6: QC of results Sand map creation Make a sand average map from the channel facies parameter ► MB1 on channel facies parameter in the Data tree ► from the Task pane select Extract surface from the Parameter utilities task group ► Select Single code: sand ► Select Extract Surface ► Clipboard ► Select Extract grid Parameter ► Rename the output of surface and parameter: channel_sand_map ► Run Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 44 ► Visualise the results in a 3D view ► Make sand maps from the other facies parameters (belts_proportions, belts_proportions_GE) and compare Data Analysis Make a chart object from the channel facies parameter ► MB1 on channel facies parameter in the Data tree ► from the Task pane select Histogram from the Charts task group ► MB1 on ‘Source data and filter’ in the chart view toolbar to specify source object and filters ► MB1 on channel parameter in data tree, then drag and drop it into ‘colouring data’ drop site in the chart view to colour the histogram. Any desired object can also be used ► MB1 on ‘Edit properties’ in the chart view toolbar ► from the Histogram properties dialog box select ‘Show bin height label’ to reveal volume fractions in the histogram view Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 45 The volume fractions of the different facies codes can be checked from the information of the facies parameter icon or within the chart window. ► MB3 on channel facies parameter in the Data tree ► Information ► Press the Volume fractions button and check the sand/gross value Save the project: ► File menu ► Save project… Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 46 7.3.7 Exercise 7: Use Crevasses with Facies:Channels (OPTIONAL) This example shows how crevasses can be modelled together with channel objects. However, as the well data does not show any crevasse facies, we will not condition on the BW. Open the Facies:Channels job again: ► Grid models ► Fluvial ► MB1 on Grid in the Data tree ► From the Task pane select Channels from the Facies modelling task group ► Change the output parameter name to ‘channels_w_crevasses’ Duplicate the job. ► In Job window ► Save as ► Save copy of job as ‘with_crevasses’ ► OK In the General tab: ► Deselect the BW (as we will NOT condition on well data) ► Toggle on the Crevasses option Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 47 In the Volume fraction tab: ► Select the option ‘From geometric parameters’ Note : The option ‘From geometric parameters’ will add the crevasses on top of the channels volume fraction, based on the geometry constraints defined in the In the Crevasse/Barrier tab choose the following settings: Channel margin coverage: 0.5 Relative lobe width: 1 Absolute lobe length: 800 Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 48 Note : The width of a crevasse splay is specified relative to the drawn average width of the channel. Therefore wide channels will have wide crevasse splays and vice versa. When modelling ‘Isolated crevasses’, Relative lobe width is approximately a crevasse belt’s maximum horizontal extension from the channel margin. When modelling ‘Continuous crevasse belt’, Relative lobe width is approximately a crevasse belt’s maximum horizontal extension from the channel apron. ► Open the previewer (previewer icon in the lower right corner of geometry tab) and inspect the crevasses geometry ► Run the job ► Change some of the crevasse geometry settings and visualise the results in the previewer (but do not execute) ► Visualize the result in a 3D view Roxar Software Training Exercise: RMS Advanced Facies Modelling: Clastic Environments 49 It may be necessary to change the visual settings of the channel_w_crevasses facies parameter: ► Choose 3 facies colour table ► Calculate min and max values 7.3.8 Exercise 8: Use correlation with Facies:Channels (OPTIONAL) In this example, we want to force the Facies:Channel job to position a channel object through a specific interval between appraisal_1 and appraisal_2 wells. (This is a synthetic example, but in a real life project there may be strong indications that it is necessary to correlate sand objects between wells). Note : Based on facies interpretation (logs pattern), pressure data, or core data analysis (heavy mineral analysis, bio