Visualization: Rapid Visual Access to Multidisciplinary Datasets

Paul White, Graham Brew, and Skip Pack—Dynamic Graphics Inc.

Summary

Multidisciplinary hydrocarbon development and production teams need multidisciplinary tools. To fully leverage team synergies, the geologist, geophysicist, reservoir and production engineers must all have access to the data from each other′s disciplines. Moreover, all these data should be easily available from the desktop, and should be visualized and interrogated in a unified 4D georeferenced space. This is increasingly important as data volumes increase, and time-lapse monitoring becomes the norm.

We present a new solution for this operational requirement. The system requires only limited data migration and reformatting. The fully georeferenced 4D viewer is capable of simultaneously displaying time-lapse seismic data, geologic models, well data, production data, reservoir properties, 4D reservoir simulations, drilling hazards, and many others.

The simultaneous visualization of reservoir models with geophysical and production data is a key component—yielding significant new insights that might otherwise be missed. Time-lapse seismic response can be interactively compared to the reservoir simulation prediction. Well log data can be compared for consistency with reservoir property models. Time-lapse seismic data can be animated to reveal pressure changes which can be instantly compared to the measured pressure change from production well data.

Usage is also widespread within collaborative team meetings and reviews where the ability to view all data in one viewer, rather than several disparate software packages, makes meetings easier, smoother, and more productive. When all relevant data can be display graphically in an easily accessible manner, interrogation becomes simpler, and associations become more apparent, leading to better decisions.

1. The Need for Visualization

Multi-disciplinary teams are at the heart of most modern hydrocarbon exploration and development.
The decisions made by these teams are based on the data relevant to each of the disciplines. These decisions are critical to the efficient development of valuable resources.
The data used in this decision-making can vary enormously in scale, detail, content, history, and utility.
The multi-disciplinary teams work best when they can incorporate all relevant information into the decision-making processes—not just the data for one or two disciplines.
Well review meetings, and other key gatherings of the decision-making staff, often require that all relevant data be available quickly and easily, and preferably geospatially referenced to all other data.
Each disciplinary expert uses their own software to analyze and visualize their data.
No single member of the team can be an expert in all these different data types
Well reviews with lots of software etc.

Conclusion:

The multi-disciplinary team could potentially work much more efficiently, and effectively, if they had rapid visual access to all data relevant to the decision-making process. In this poster we present a software application that could potentially address some of these needs. We also examine short-comings and the needs for further development.

2. Visualization Requirements

Any application that aims to satisfy the established visualization requirements should have the following characteristics:

Flexible Imports: Can handle any of the teams multi-disciplinary data
Minimal front-end loading: Data entry should be as easy and fast as possible, handling all necessary formats quickly and with minimal duplication
Intuitive⁄Easy to use: The team should not be burdened by another complex, difficult to learn software package. Rather, the application should build on existing skills, paradigms, and intuition
Multi-platform: The application should work seamlessly and effortlessly across all available platforms⁄operating systems (Windows^®, Linux^® etc.)
Fast: Start-up, data-loading, rendering and usage should all be suitably fast to encourage team usage and data exploration
Functional: The application should display⁄manipulate data in a fully flexible fashion⁄ thus aiding, rather than impeding, decision making

3. Visualization Spatial Data

As we have seen, the demands of the visualization environment are data-driven. At a minimum, the environment should be capable of visualizing the following spatial data (non-spatial data, and data with a ″soft″ spatial components are treated in a later panel):

Seismic Data: Time & depth
Seismic attributes: Calculated from 3D and 4D seismic
Seismic Interpretations: Horizon picks, fault picks.
Well Locations: Well paths, surveys etc.
Mechanical Data: Casing points and borehole completions
Annotation Data: Lease boundaries, facilities, coastlines
Well Logs: Logging data, televiewer images, core photos
Cellular Property models
Reservoir simulation output: history match and predictions
Structural Models: faults, horizons
Images: Aerial⁄satellite images, draped on surfaces

4. Visualization Examples

Complex Structure⁄Time Lapse Seismic

In the example below, a complex field that has been on production for some years is visualized. Imperative to the development of this field was the use of time-lapse seismic data. This highlighted the effectiveness of the water injection, and the regions of the reservoir with undeveloped reserves. The temporal aspects of this visualization are discussed in a later panel.

In this broad field-scale image, the following elements are shown:

Initial structural model (incomplete)
Seismic attribute map
Seismic planes
Seismic volume (filtered to show only high values⁄blue clouds)
Well locations (producers, injectors, and exploration holes)
Select well log data
Cellular property models
Casing Points

Manipulation of this kind of 4D visualization allows the multi-disciplinary team to see connections and associations previously only available by generating and combining static images from numerous software packages.

Reservoir Simulation Data⁄Well Data QC

Shown below is an illustration of how the visualization environment can render reservoir simulations models with a large number of properties. Filters can be performed on the individual properties, and different color tables applied to each. Furthermore, the well data shown along side the cellular model can highlight where production and injection is taking place, and illustrate where incompatibilities exist between the finely sampled well data and the upscaled reservoir simulation grid.

Flexibility of Data Scales (UK coastline and bathymetry)

In this example, the extensibility and flexibility of the visualization environment is illustrated. Not only can the CoViz 4D environment render spatial objects as small as a single logging sample, or casing point, it can also handle features as large as the coastline of the UK. All these scales can be achieved within the same sessions and viewing window, allowing for the development team to consider multiple scales of data, from regional basin-size work, through facilities maps, down to the size of individual wells. This allows for the better appreciation of the ″big-picture″ when making reservoir-scale decisions.

5. Visualizing Property Information

Property information is clearly key to fully understanding reservoir conditions and responses. Much of this data comes from well logging curves. Any visualization environment must be able to load, filter, manipulate and display property information from a wide variety of sources, and in as many different ways as possible, to aid the decision-making process.

Property information can be contained within structure models and cellular models⁄reservoir simulations input⁄output. The visualization environment should allow for rapid methods for the visual and analytical comparison and QC of well data alongside these models. Furthermore, this visual comparison can illustrate any adverse effects of the upscaling process that created the cellular grids⁄reservoir simulations input.

In the visualization environment discussed herein, there are a wide variety of methods for visualizing the well log data. These include ″lathe″ plots, API-style curves, changing colors, labels, and symbols. These can all be displayed simultaneously with the other elements of the visualization environment⁄in this example the structural (horizon and fault) model.

6. Multiple Model Spaces

The visualization environment is clearly capable of displaying large volumes of varied spatial information. When numerous data sets are displayed concurrently, the display may become confusing, and difficult to interpret. Furthermore, data sets can commonly have different vertical units. Consider a time-migrated seismic cube scaled in seconds, or a magnetic anomaly map with a vertical scale in mantels.

To accommodate these requirements, the visualization environment allows multiple ″spaces.″ These are duplications of the z (vertical) space stacked above one another and spatially referenced in x and y. This can make individual data components much easier to visualize, and it allows for different vertical units in the different spaces. Furthermore, this method allows for the interactive tracking of features between the model spaces (i.e., track a structural horizon in one space on a seismic amplitude cube in another space). The examples here illustrate just some of the many advantages of this approach.

7. Temporal Information

The efficient development of subsurface resources is very much a 4D, rather than a 3D, problem. Hence any visualization system should be fast, efficient, and flexible in displaying a wide variety of temporal data. The time-variant data utilized by the asset team could include:

Reservoir simulations: both history matches and predictions for a potentially huge variety of reservoir properties
Time Lapse Seismic Data: many gigabytes of data, collected possibility as frequently as every several months. Potentially dozens of coincident seismic amplitude cubes
Time Lapse Seismic Attributes: Calculated from the seismic data. Various combinations of attributes, particularly the examination of differences between adjacent time-steps, can lead to thousands of seismic attribute maps displayed on the various reservoir horizons
Production Data: Production, injection, fluid compositions, pressures
Well Data: Date of drilling, plus potential changes in the status of the wellbore (e.g., producer to injector) that span the entire life of the field
Mechanical Data: which perforations were opened and when, time for pump on⁄off, workovers, flow tests, choke settings etc.
Well Logs: Multiple well logs may be collected from a single wellbore over the course of reservoir development and maturity

The example shown in this panel (right) illustrates just some of these temporal visualization possibilities. This visualization was used in an actively producing field to make decisions regarding work-overs and new drilling opportunities.

8. Non-Spatial Information

The subsurface development team uses huge quantities of spatial information, as demonstrated in the previous panels. However, equally important can be non-spatial information, or information with only ″soft″ spatial associations. For example:

Production and surveillance data
Core photos⁄Drilling tests
Well review documents
Mechanical schematics
Risk Assessment⁄financial analysis

Furthermore, these data could be in a variety of non-spatial formats:

PowerPoint slides
Excel spreadsheets
Text documents
Images⁄PDF files
Websites⁄intranet sites
Various database formats

The quantity and varied nature of these data make conversion to true spatial entities almost impossible. Hence, the visualization environment should allow for connections to be made to these data sources from directly within the 3D Viewer. Hence the user can access these information as readily as any ″true″ spatial object.

Production Data

A key deliverable in the visualization environment is the efficient linkage between the 4D visualization and production⁄surveillance data allowing correlation between the abstracted simulation models, and the ground-truth in terms of hydrocarbon production. Production declines, pressure spikes, injected volumes, gas-oil ratios and so forth can all be investigated alongside the structural⁄simulations models to guide decisions for optimal reservoir development. This allows everyone in the multi-disciplinary team—from geophysicists to production engineers—to access their data within the same software environment. Required are:

Speed: The data should be available near-instantaneously to facilitate decision-making in a team- collaborative setting
Currency: The data should be as up-to-date as possible (e.g. daily⁄hourly)
Flexibility: The environment should allow for easy mixing of property types, generation of new plots, and rapid comparison between wells⁄well groups

February 1999

The following items are shown in this temporal plot: seismic amplitude cube from a single time point, seismic attribute (difference) map calculated from two different seismic amplitude cubes displayed on the top-sand surface, well data, perforation data, and a reservoir simulation prediction colored by Water Saturation, filtered to >30% Sw.

April 2000

Note the large blue seismic attribute anomaly. The seismic response indicates strong pressure support⁄water injection in this (upper) portion of the sand. Yet the reservoir prediction model shows only small levels of water in the sand at this time point (blue well is water injector).

September 2001

The mismatch between reservoir prediction and seismic observations persists. Using further analysis of a lower sand (not shown) the team concludes that the perforations to the lower sand in the injector well are blocked, leading to elevated levels of injection into the upper sand.

February 2003

New seismic data have been acquired since the last time step. Hence this temporal image shows a new seismic cube and a new seismic attribute difference map. The mismatch between seismic attribute and the reservoir prediction is lessened, but still requires a reevaluation of the reservoir prediction model.

July 2004

The final time step. Production data plots are shown from this time step. The water injection pressures showed a spike in the pressures of the well injecting into the sand, thus supporting the hypothesis that blocked perforations were impeding water injection into the lower sand.

Acknowledgements

The authors wish to thank all the staff at Dynamic Graphics, Inc. for their dedication and commitment to the software development process. Many thanks also to DGI clients who provided many of the data sets used in the examples on this poster.