4D Subsurface Modelling: Predicting the Future/media/shared/documents/Events/2019… · 4D Subsurface Modelling: Predicting the Future – Burlington House, 20-21 Feb 2019 20th-21st

0

4D Subsurface Modelling: Predicting the Future – Burlington House, 20-21 Feb 2019

20th-21st February 2019 Page 1

Contents

Conference Programme Pages 2-4

Oral Abstracts Pages 5-37

Poster Abstracts Pages 38-42

Geological Society Fire Safety Information

Page 43

Geological Society Floorplan Page 44



4D Subsurface Modelling: Predicting The Future

20th – 21st February 2019

Programme

Wednesday 20th February 2019

08.30 Registration & tea, coffee & refreshments

09.00 Welcome – John Booth (Geological Society), Glen Burridge (Conference Co-Chair)

Session I: Keynotes Chairs: John Booth & Glen Burridge

09.15

Oil & Gas Keynote - Modelling & Monitoring Reservoirs over their Lifetimes M. Oristaglio (SEAM Project/ Yale Univ.)

09.45 Civil Eng Keynote - 4D Ground Modelling in Civil Engineering: Recent Developments R. Talby & E. Russell (Mott MacDonald)

10.15 Mining Keynote - 4D Data Management & Modelling in the Assessment of Deep Underground Mining Hazard W.J. McGaughey (Mira Geosciences)

10.45 Break: Tea, coffee, refreshments and posters

11.10 Geothermal Keynote - The Geothermal Space: What's in it for me? C. Baxter (Intl. Geothermal Assoc.)

Session II: Framing Chair: Dick Plumb

11.30 The Value of Understanding: No Such Thing as the "Field Model" M. Bentley (AGR-TRACS)

11.50 Influence of Lithology on Fluid-flow, Geomechanical and Seismic Response in an Integrated 4D Earth Model A. Bottrill (MP Geomechanics)

12.10 Co-Visualization and Analysis of Multi-disciplinary Data J. Wheelwright (Dynamic Graphics)

12.30 LUNCH and Poster Session

Workshops Session Introducer: Jorg Herwanger

13.30 Icebreaker activity: “Draw this Story” Jorg Herwanger (MP Geomechanics)

13.40 The Art of Mixed & Augmented Reality M. Lato (BGC Engineering)



13.40 Co-Visualization of 4D Data: Barriers & Benefits J. Wheelwright (Dynamics Graphics)

13.40 Managing the 5th Dimension in Subsurface Modelling L. Johnson (Cognitive Geology)

13.40 Have Advances in Subsurface Modelling Enabled Higher Quality Decisions? C. Jacquemyn (Imperial College London)

13.40 What Challenges Drive Subsurface Modelling? A Discussion across Sectors C. Baxter (Seequent)

14.40 Workshop Summaries


Session III: Workflow Chair: Thomas Finkbeiner

15.30 Production-mode Geophysical Surveys & Geological Principles Applied to Shallow-Water/Earth 4D Models in Industrial harbours: Murphy's Mapping Method

B. Ward (e4 Sciences)

15.50 A Dynamic Model for Solution-Mining Caverns L. te Kamp (ITASCA/GuG)

16.10 Predicting the Future with Subsurface Models: How Well are We Doing Now? K. Heffer (Reservoir Dynamics)

16.30 Q&A and Feedback Session

Drinks Reception

17.45 Eccentric 19th Century Geologists & their Models L. Scales (Laurence’s Walks)

18.05 – 20:00

How Seequent found itself Modelling the Subsurface….

Sponsored by Seequent team

Thursday 21st February

Session III continued Chair: Thomas Finkbeiner

9.00 Introduction John Booth (Geological Society) & Glen Burridge (Conference Co-Chair)

9.10 National Surveys Keynote - Modelling the Solid Earth: A Geological Survey Perspective K. Royse (British Geological Survey)

9.40 Translating Compositional Earth Models into Realistic Computational Models with Consideration to the Complexity of Geological Constraints A. Long (Subterrane)

10.00 First History-Matched Full Field Coupled Geomechanics & Reservoir Flow Model of both the Valhall and Hod Fields, including Water Weakening & Re-Pressurisation J. Kato (Rockfield)

10.20 3D Geological Modelling of the Chalk Group in SE England for Engineering & Groundwater Applications: The Future is 4D? C. Cripps (British Geological Survey)


11.00 Development & Application of a Dynamic Fracture Growth Approach to Generating Discrete Fracture Network Systems in Sedimentary Rocks S. Baxter (Golder Associates)

11.20 Wellbore Instability & Deformation Risks Highlighted by 4D Seismic & Geomechanical Studies M. Calvert (Total)

Session IV: People Interface Chair: Wolfgang Hohl

11.40 Technology Keynote: 3D Holographic Visualization of 4D data M. Lato (BGC Engineering)



12.00 Data & Models: How do they Interact? An Example of how Gravity & Seismic Monitoring Constrain Subsurface 4D Models O. Eiken (Quad Geometrics)

12.20 Q&A and Feedback Session

12.30 LUNCH and Poster Session

13.30 Integrated Subsurface Modelling Framework for Unconventional Development R. Guises et al (Baker Hughes)

13.50 A Reliable Organisational & Procedural Framework for Maintenance of an Ongoing Mine Stability Safety Case in 4D D. Beck (Beck Engineering)

Workshop & Panel

14.10 Framing of Conclusions – Legacy from Event Glen Burridge (Glen Burridge & Assocs, Conference Co-Chair)

14.20 Have Advances in Subsurface Modelling Enabled Higher Quality Decisions? Chair: G. McKinley (Minerals Advisory)


Closing Panel Session

15.45 Wrap-Up Panel Session, Q&A and Feedback Chair: K. Royse (British Geological Survey)

16.15 Close

THE CONFERENCE ORGANISING COMMITTEE Originators & Co-Chairs Glen Burridge (Glen Burridge & Associates) https://www.linkedin.com/in/glenburridge/ Thomas Finkbeiner (KAUST) https://www.linkedin.com/in/thomas-finkbeiner-6166a023/

Richard Plumb (Plumb Geomechanics) https://www.linkedin.com/in/richard-plumb-a021177/ Planning Committee Jorg Herwanger (MP Geomechanics) https://www.linkedin.com/in/jorgherwanger/ Wolfgang Hohl (Hohlraum ZT) https://www.linkedin.com/in/hohlraumzt/ Katherine Royse (British Geological Survey) https://www.linkedin.com/in/katherine-royse-aa525821/ Benedikt Steiner (Camborne School of Mines) https://www.linkedin.com/in/benedikt-s-b20a7417/

Geological Society Science Committee Liaison & Support John Booth (Geotechnics) https://www.linkedin.com/in/john-booth-133a381/

https://www.linkedin.com/in/glenburridge/

https://www.linkedin.com/in/thomas-finkbeiner-6166a023/

https://www.linkedin.com/in/richard-plumb-a021177/

https://www.linkedin.com/in/jorgherwanger/

https://www.linkedin.com/in/hohlraumzt/

https://www.linkedin.com/in/katherine-royse-aa525821/

https://www.linkedin.com/in/benedikt-s-b20a7417/

https://www.linkedin.com/in/john-booth-133a381/



OUR OBJECTIVES FOR THE CONFERENCE

As geoscientists, we strive for an integrated view of the Earth beneath our feet.

This novel event examines what can be gained from building subsurface models through time and

seeks to explore lessons that can be shared across a number of industry sectors:

Mining - Oil & Gas - Civil Engineering - Geothermal

Equally, how challenges, such as sufficient framing, choice of analytical or deterministic methods,

composition of integrated team of experts, data availability etc. can be tackled to ensure maximum

value is derived.

We intend this workshop to offer a unique opportunity for creative minds within these industries to

present new perspectives, state-of-the-art technology and capabilities. In addition, a chance for

personal interaction between experts from widely different industry sectors in constructive, cross-

disciplinary discussions that will lead to fresh insights, collaborations and working relationships.

Key points we aim to address:

1. Purpose of a 3D Model through Time

• What do we seek to achieve by repeatedly building a model in 3D?

• What decisions does such a model impact upon in each industry?

2. The State-of-play in 4D Modelling

• How does each sector frame 4D modelling? What are the characteristic time-steps involved?

• What is the state of progress in each sector and where are the greatest advances?

3. Seeking Value from the Model

• How do we assure the right knowledge is generated, preserved & applied from a 4D model?

• How do we capture uncertainties, avoid cognitive traps and better assure value from 4D models?

4. Strength in Robust Inputs & Collaboration

• Where are the greatest bottlenecks in acquiring robust data and achieving reliable results?

• What ensures collaboration in building shared models that can be multi-scale and robust?



#Oil & Gas Keynote Modeling and monitoring reservoirs over their lifetimes M. Oristaglio (SEAM Project/ Yale Univ.)

Michael Oristaglio, Yale University, New Haven, CT

Shauna Oppert, Chevron Energy Technology Company, Houston, TX

Joseph Stefani, Chevron Energy Technology Company, San Ramon, CA

Jorg Herwanger, MP Geomechanics, Kent, UK

Peter Popov, MP Geomechanics, Kent, UK

Lijian Tan, AGT, Houston, TX

Vincent Artus, Kappa Engineering, Houston, TX

Large-scale simulation is a tool for understanding how underground reservoirs change during their

productive lifetimes and how these changes can be detected and quantified by geophysical remote

sensing. This paper describes a recent collaborative project between government, industry, and

academic research groups to design, build, and simulate an integrated geologic, reservoir, and

geophysical model during more than two years of fluid production.

The geologic model consisted of 2 billion cells representing a region 12.5 km by 12.5 km in horizontal

extent and 5 km in depth and including a 420-m thick reservoir, with upper and lower turbidite-fan

units separated by an impermeable shale layer and offset by faults. The model was based on a typical

Gulf of Mexico deep marine reservoir which can serve as an analog of turbidite fields around the

world. The reservoir simulation computed fully coupled three-phase fluid flow and linear

geomechanical responses in a production scenario involving 11 production wells and 6 water-

injection wells penetrating three reservoir compartments.

Seismic surveys were simulated with isotropic elastic wave modeling before the start of production

and after 27.5 months of production at a simulated rate of about 67 500 barrels per day, with about

32 500 barrels per day of water injection following a year of natural pressure drive. Rock properties

were updated by petrophysical models calibrated to turbidite systems in the Gulf of Mexico.

Analysis of the model and simulations improves understanding of the complex interaction of fluid

effects, pressure changes, and rock deformation. For example, compaction in the reservoir may

cancel time-lapse fluid effects, and compaction or dilation in the surrounding shales may override the

observed reservoir signal in the seismic bandwidth. Strain-induced velocity changes in the shales have

a much larger effect on estimated time-lapse time shifts than do strain-induced changes in path

lengths. The models and data created by the project are now publicly available through the SEG

Advanced Modeling project (SEAM).



#Civil Engineering Keynote 4D Ground Modelling in Civil Engineering: Recent Developments R. Talby & E. Russell (Mott MacDonald) Developing a 3 dimensional understanding of the ground is not new to civil engineering/engineering geology, as a 3D understanding is a fundamental principal of geological work. Conceptual models have long been used to illustrate our understanding of the ground and potential risks, and will continue to do so. Software developments, and the trickle down of the technology from the mining/petroleum industry is increasing possibilities for true ground models to be created. Rather than multiple 2D sections being used to convey changes around sites alone, 3D models developed by geologists, rather than CAD technicians, allow for a much better understanding to be conveyed and reduce chances of errors though the development process. Ground Model development is a 4D process as the available data increases with time typically through desk study, preliminary ground investigation, detailed ground investigation and finally through construction verification. The addition of new ground data with time is unlikely to show time related changes to the actual geology. As the design of a construction project develops so more geological data is gathered to inform and improve the understanding of the likely geology, potentially removing errors and misunderstandings and informing more clearly potential risks to the project team. The use of a 3D digital ground model, which develops with time and includes many sources of different data, requires careful management to oversee the transfer and inclusion of the new data. There are multiple other areas where the application of a 4th timeseries dimension to 3D ground models has the potential to greatly improve the understanding of geology and provide clarity of communication of issues such as: porewater pressure distribution; ground movements; development of contamination plumes. The main difficulties with these areas relate to establishing a suitable number of subsurface control points to provide enough information to make the addition of a 4th dimension useful. A number of examples from the Thames Tideway ground models are used to illustrate the benefits of use of 3/4D ground modelling will be provided. An example will include the development of a 3D ground model from a phased program of ground investigation to inform shaft construction (see image below).



Extract from 3D ground model for a DEPCS Shaft. Left – overall model (Chalk = green, Thanet Sand =blue, River Terrace Deposits = orange, Made Ground =grey) Right - Logged encounters with inclined fractures from borehole core indicted by red dots. Orientation of fractures from geophysical logging shown as blue discs. Conjectured fracture plane shown in orange. Offset of Shoreham marl marker horizons (blue and green planes) across fault from core and geophysical observations.



#Mining Keynote Talk 4D Data Management & Modelling in the Assessment of Deep Underground Mining Hazard W.J. McGaughey (Mira Geosciences)

A clear opportunity exists for earth scientists to take advantage of recent advances in machine

learning and related analysis tools that are taking place outside the geosciences. The challenge is to

formulate geoscience problems such that they can be addressed by these methods. One important

area ready for the innovation that may be afforded by these new methods is geohazard assessment, a

4D problem that has proven difficult to approach with quantitative methods that are sufficiently

general to consider the many natural and engineered factors at play and be applicable across a broad

range of hazard types.

Across the entire spectrum of rock engineering problems, the application of machine learning or

related methods of quantitative analytics is far from simple. The reason is that the condition being

predicted—for example the location and timing of a geotechnical hazard, or seismicity, or the many

operational challenges facing mines—may not be primarily related to measurable variables at the

location in space and time of the prediction. The condition to be forecast exists when and where it

does because of the properties of the complex, four-dimensional, spatial and temporal natural earth

and engineered mine system, which can only be partially known.

Our work has focused on geotechnical risk in mining, which is universally understood to depend on

many disparate factors, such as stress, stiffness, mine geometry, rock mass character and quality,

rock type, geological structure, excavation rate and volume, blasting, and seismicity. We have worked

on many case studies over the years in both underground and open pit mines with the objective of

discovering and documenting correlations of failure to these input factors, all of which can be

measured or modelled, and many of which vary in both space and time. Whether the failure being

analyzed is slope failure, fault-induced rock bursting, strain bursting, roof fall, or any other of many

possible failure types, quantitative correlations among many classes of data can be found, and

predictive rules based on them can be established. Such a data-driven approach requires application

of methods and avoidance of pitfalls that we have standardized into a generally-applicable workflow

of 4D modelling, formulation of data structures from the models that are appropriate as input to

machine learning or similar analysis, followed by representation of the outputs back in 4D model

space for validation and interpretation.

One important lesson from our work in integrating multi-disciplinary data for quantitative analysis is

that operations do not typically have the required data management infrastructure to readily provide

the required inputs. This situation is manageable where hazard assessment updates are only required

off-line on a one-off or intermittent basis. However, a more useful continuous assessment of hazard

requires up-to-date access to all of the data types required for calculation, which cannot be

adequately addressed without access to a multi-disciplinary 4D data management system. This is

because geotechnical hazard evolves in time, the underlying data are time-dependent, and the results

of analysis must be routinely updated. The need for systems to support, record, and communicate

multi-disciplinary geotechnical interpretation is pressing as mines become deeper, geotechnical

problems become more acute, and the flow of data available for analysis has ever-greater speed and

volume.



#Geothermal Keynote Talk The Geothermal Space: What's in it for me? C. Baxter (Intl. Geothermal Assoc.)

Geothermal energy has what it takes to drive the energy transition by utilising the naturally stored energy right beneath our feet to power up our grids and heat our homes. However, harnessing that phenomenal amount of stored heat and use it to its full potential in terms of both power and direct use, is still limited and fragmented. Geoscientists have a great role to play in driving geothermal as a renewable technology through their excellent knowledge of the subsurface, modelling capabilities, and integrative thinking. The presentation will focus on upcoming markets for geothermal energy, game-changing technologies needed for upscaling, and the drive to put geosciences in the heart of the energy transition.



Main Session Talk

The Value of Understanding: No Such Thing as the ‘Field Model’

Mark Bentley (Heriot-Watt University & AGR TRACS Training)

Ed Stephens (AGR TRACS Consultancy)

When we have a complex problem, we are often compelled to build a complex model to address it.

We often end up understanding neither the model (too complex) nor the original problem (still

unsolved).

This is a solution-driven approach to modelling. Models are built to answer a question, to find a

solution to some problem, yet often these models are too far from the truth to deliver useful

guidance or reasonably accurate solutions. If we run a suitable set of scenarios to explore our

uncertainty, we may end up illustrating that we simply have little forecasting power which prompts

us to then make a hopeful best technical guess, thereby invalidating the whole forecasting process.

This inevitably begs the question, “what’s the point of the modelling effort?” This presentation offers

two related suggestions:

1. the modelling need not be aimed at finding solutions, but to understand the problem, and

2. the modelling can be targeted at understanding the nature of the producing system: building

‘truth models’

Models for exploring problems are models for understanding and insight. We are attempting to

better define our questions and explore unknowns without necessarily building time-consuming and

complex full-field or full-lifecycle models. Armed with the insights gained we may go on to further

modelling efforts but may equally now have all the understanding we need to answer the question

initially posed. The answer may now be intuitively clear.

Attempting to model the truth means pitching the scale of investigation at the scale of the question

itself and driving model resolution down to the scale of the underlying heterogeneity and the scale of

the data in use. These so-called ‘truth models’ are an attempt to connect these scales and develop an

understanding of the producing system at the scale it is operating.

Depending on the nature of the system and the questions we are posing, suggestions (1) and (2)

above may align.

This approach is very different from the pursuit of full-field ‘life-cycle’ models involving full history

matching of production data, assisted or otherwise. Current computing power is diverted from the

goal of including just a little more detail in a big model (usually by reducing cell sizes) to refocussing

the modelling effort to a smaller scale: the scale of the data, the scale of the question, the scale of the

operative process in the reservoir, or all three: ‘truth models’.

This style of modelling lives for the season to help develop understanding when it is needed. The

models are simpler in design and nimbler than full-field, life-cycle models, and they change as

understanding develops and as the questions themselves change. In this sense they fill the fourth

dimension of time. In the world of dynamically changing truth models, there is never such a thing as

the ‘full-field model’, only a full-field database from which the truth is extracted and pursued as and

when necessary to promote understanding.



Main Session Talk

Influence of Lithology on Fluid-flow, Geomechanical and Seismic Response in an Integrated 4D Earth Model

Andy Bottrill, Jorg Herwanger, Peter Popov (MP Geomechanics)

Lithology or rock type has a strong influence how the Earth responds to fluid extraction and to

geomechanical processes. It also has a strong impact on the seismic signal recorded in reflection

seismic surveys. I will argue that rock type (e.g. sandstone, shale, carbonate and similar descriptions)

is the most important parameter to focus on during model creation during 4D subsurface modelling.

The argument will be supported by results from an integrated 4D Earth modelling study. The study

demonstrates the first order control of lithological stratigraphy on the coupled flow and

geomechanical response from oil production and water injection into a complex deepwater turbidite

reservoir. During reservoir production the seismic response is recorded after 3 years of production,

again demonstrating the strong influence of the seismic signal on the lithological boundaries.

According to Darcy’s law of fluid flow, the hydraulic permeability is the rock property which

determines at which rate fluid flows in the subsurface. Accurately gridding different lithologies, which

can have orders of magnitude differences in permeability, is therefore vitally important to fully

characterize pressure and fluid fronts in the model. In building our simulation model we employ a

novel workflow which tracks lithology boundaries in the geological model using a level-set method,

thus maintaining the lithology. Pressure and fluid fronts propagate within sandstone lithologies, and

thin shale baffles act as baffles for fluid flow.

Lithology also strongly influences geomechanical processes though both the initial mechanical

properties and the changes in pore pressure induced through production. The preservation of

lithologic boundaries in our model improves the geomechanical predictions by not only accurately

representing the distribution of stronger (quartz rich) with weaker (shale rich) lithologies but also

containing pressure changes to the correct lithologic units.

The reflection seismic response in the Earth is determined by the contrast in elastic impedance (the

product of density and propagation velocity of an elastic wave) across a layer boundary. Lithology or

rock type is again a major indicator for the magnitude of elastic impedance. Within each lithology, the

porosity and shape of pore space will determine any further variations in elastic impedance.

This model represents a step toward a full 4D Earth model utilizing a consistent representation of

flow, geomechanical, and seismic properties. Only by accurately representing the lithology can we

produce truly predictive models.



Main Session Talk

Co-Visualization and Analysis of Multi-disciplinary Data Jane Wheelwright (Dynamic Graphics Inc.) There will always be limitations in fully understanding the subsurface, but, by assimilating all available data eg gravity, magnetic and seismic surveys, borehole wellpath and logs files, and applying robust model building algorithms to interpolate and extrapolate this information over the area of interest, planning and development of engineering strategies can be optimised. A key part of these workflows is to generate shared earth models which are as representative of the input data as possible, and also to provide a framework to enable predictive forward modelling to understand the implications and uncertainty of any processes being conducted. To fully comprehend the models being built requires a sophisticated co-visualization system which is capable of simultaneously viewing and interrogating multiple data sets, across a range of disciplines, with the option to also easily access non spatial data where it can be vital to the decision making process. In addition to building static models, it is also, for many applications, of considerable importance to understand changes over time which are brought about through excavation and drilling. The advantages of incorporating the 4D or temporal element into a visualization system enables changes in each of the input data types to be understood, not in isolation, but in regards to the status of all relevant data. If an observed seismic survey, for example, is not matching a predicted outcome, having all information in a common environment leads to a rapid understanding of why there may be inconsistencies and what steps need to be taken to resolve any potential issues. In areas of hydraulic fracturing, any seismic events can be correlated with the timings of drilling events to accurately determine cause and effect. Understanding the effects of engineering and drilling, both in the zones where activity is taking place, and in the surrounding areas, is key to ensuring any problems will be minimised and the effect on infrastructural integrity can be monitored. When multi-disciplinary data is available in a common 4D visualization environment, in addition to the power of qualitative analysis, it is also the natural place to statistically compare overlapping data sets, at different time points, by back interpolating and running quantitative workflows. This enables predictive models to be rapidly updated and improved. Other 4D workflows, such as the generation of synthetic seismic from simulation data and the prediction of changes in the overburden can also be accomplished. This talk will look at some examples of subsurface modelling using a variety of input data sources and emphasise the value of co-visualizing all available data over time. Examples are primarily from the Oil and Gas industry but we will also look at an open cast gypsum mine in Switzerland and the surface deformation encountered as an effect of water production in the Las Vegas area.



Workshop Session

Draw this story – an icebreaker activity

Jorg Herwanger, MP Geomechanics

“A picture is worth a thousand words”. I am suggesting an experiment to test this hypothesis for the

icebreaker (or an introductory session) at the 4D GSL conference. The participants will be given a

story describing a model. The participants will then ask the participants to draw their own figure

depicting the model. Subsequently, a picture will be shown detailing the model underlying the

description. I suggest placing the drawings of the participants on a wall during the conference. In the

abstract below, the verbal description as well as the graphical description are given. During the

conference, the participants will initially be given only the verbal description from which they will

create their own graphical representation. Subsequently the “true” image will be shared.

The expected outcome and benefits of this experiment includes:

• The activity will stimulate discussion on models, the importance of visualization, and the

difficulty in accurately describing models using words only;

• Participants are active themselves, will have fun doing so, and I expect a lively discussion

afterwards – and the proverbial ice is broken;

• We can collect the drawings and get a range of outcomes of models based on one

description. This allows to frame further a session on uncertainty.

Model 1: Stresses around a wellbore

Your task: Please create a graphical representation of near-wellbore stresses from the following description The presence of a wellbore (or a tunnel with a circular profile) perturbs the stress state from the reference stress state in the rock mass without the wellbore. At the wellbore wall, the principal directions of the stress tensor align in directions perpendicular and parallel to the wellbore, forming the axial, radial and tangential stresses with respect to the wellbore geometry. Moving away from the wellbore wall, the principal directions of the stress tensor rotate (within roughly 2 to 3 wellbore radii) to eventually align with the principal directions of the reference (regional or far-field) stress state. In the wellbore wall, the magnitude of the effective radial stress is given by the difference between the pressure of the fluid in the wellbore and pore pressure in the adjacent formation. Thus, if the wellbore pressure and the pore pressure are matched the magnitude of the effective radial stress is zero. Note furthermore that the radial effective stress is the same for the entire wellbore circumference. The tangential stress with respect to the wellbore circumference varies markedly as a function of azimuth angle. For a vertical wellbore, the maximum tangential stress is observed in the directions given by the minimum horizontal stress of the reference stress field; and the minimum tangential stress is observed in directions given by the maximum horizontal stress in the far-field. In directions given by the minimum horizontal stress, the difference between tangential and radial stress becomes large. This large differential stress can be the cause of wellbore breakouts.



Figure 1: Principal stresses in the vicinity of a well of a tunnel with a circular profile

Model 2: 3D model of an oilfield

Your task: Please create a graphical representation of an oilfield from the following description The reservoir is formed by an elongated double-dipping anticline. The long axis trends NNW-SSE. The field extends 12.5 km along the major axis of the anticline and 4 km along the minor axis of the anticline. The crest of the field has a post-depositional graben structure, with the axis of the graben being aligned with long axis of the anticline. The field is developed by extended-reach horizontal wells from a central platform. Four production wells ae interleave with three water injectors. There are two high porosity oil-bearing reservoir intervals. At the crest, the reservoir thickness is approximately 40 m for the upper reservoir 60m for the lower reservoir. The two reservoirs are separated by a low porosity zone. Reservoir thickness increases as one moves downslope along the shoulders of the reservoir. The highest porosity is found on the crest of the anticline, and porosity decreases as one moves downslope on the shoulders. Top reservoir forms a strong impedance contrast with the overlying shale rocks, creating a strong top-reservoir reflection in 3D seismic data.



Figure 2: Porosity model and wells for parts of the South Arne oilfield.



Workshop Session

The Art of Mixed & Augmented Reality M. Lato (BGC Engineering)

Holographic visualization allows people to interact with, and control their experience, which facilitates the generation of a deeper understanding of 3D environments and their changes over time. Holographic visualization is enabling clearer communication and more confident decision making. It will also help identify, anticipate, and mitigate challenges earlier within the project lifecycle by being able to visualize the site in 3D on a boardroom table or standing in the landscape. It creates a shared vision, a common reality that allows meaningful input and decisions to be made based on science, facts and evidence. In this workshop we will explore and understand how holographic models are developed and utilized.

We will begin with an overview of AR, VR and MR and what technologies are typically used in

different applications from model building to public engagement. We will explore the fundamentals

of holographic visualization, why some models are easy to build, why some are hard, and when the

investment in a hard to build model is worthwhile.

Multiple models from various projects will be available for demonstration. Models range from

geophysical data to boreholes, mine plans, and orebody models.

Be prepared to discuss a project of interest and work through how it could become a holographic

model.



Workshop Session

Co-Visualization of 4D Data: Barriers & Benefits J. Wheelwright (Dynamics Graphics)

This Conference aims to address the benefits of incorporating temporal information in the model building process and to investigate workflows to achieve this. An integral part of the challenge is getting different 4D earth models, and associated data, into the same visualization space. It’s relatively simple to view the changes over time of a single attribute in a single model, but this doesn’t take into account other ancillary time dependent data which may be key to understanding the processes at work. It’s difficult to know what you don't know without time-variant co-visualization. This workshop aims to set out what attendees understand by the term co-visualization, investigate the range of data which requires assimilation on a common platform, discuss the difficulties to achieving this and, hopefully, to come to some agreement on what such a system should provide. As a starting point to discussion, I would like all attendees, in groups of 3, to fill out the questionnaire to highlight the major 4D data types being used, the difficulties integrating data and to understand where there is commonality across the disciplines. This is an ideal platform to share experiences, both positive and negative and to come to some agreement about what would constitute a valid system. The benefits of co-visualization extend beyond the correlation of data at a particular time point, understanding relationships can lead to unforeseen benefits as well as providing a communication tool for professionals working on the data and for a wider audience. The tools that are used to gather, process and interpret data to build earth models may not always be the same tools required to unite data to explore implied relationships, evaluate opportunities, and communicate with peers, decision-makers, partners, stakeholders and regulators.

Q&A

What types of 4D data do you have available? Simulation, Seismic, Production etc What do you consider the term co-visualization to mean? What are your current tools for integrating and visualizing temporal data? PowerPoint etc What issues do you have integrating data? Access, Cleaning, Coordinate Reference System consistency, Depth/Time domains, formats, issues of scale etc Where do you see the key advantages of 4D Data visual integration? Sharing data across teams and disciplines, saving time, understanding relationships between datasets etc What would the ideal be?



Workshop Session

Managing the 5th Dimension in Subsurface Modelling L. Johnson (Cognitive Geology)

The acquisition of new field data helps hone our understanding of the subsurface, but our

interpretations are never certain.

Unfortunately, by the time we have production data, contemporary modelling workflows attempt

to produce a single, perfect "history match" - from which all uncertainty is ignored, until the

prediction model is proven wrong by production data 6 to 12 months later.

Throughout this workshop, we will explore how ensemble modelling methods can be used to

progressively incorporate new knowledge, without forcing a path to uncertainty blindness. The

future of modelling and flow simulation will see the term "history matching" retired, as we move

towards weighted scenario management: Come and see why.



Workshop Session

What’s the Difference Between the Models We Build for O&G vs Geothermal? C. Jacquemyn (Imperial College London)

Modelling of subsurface reservoirs is common practice across different applications (O&G,

Geothermal…) to derive useful information from available data, concepts and their uncertainties.

Many similarities exist between such models, and probably at least as many differences can be found.

We will explore how models compare between different applications, from the question models are

trying to answer, over the workflow and modelling emphasis, to the input data that is being used.

This comparison will help to understand where overlap and differences exist and show opportunities

for transfer of ideas, technology and people between applications.

We will go through the modelling workflows in reverse order (<10min on each of the below) and get

input for each what is common practice and how it is different or similar:

1. What is the end-product (volumetrics, production profile, …)? What information does the

model provide?

2. Dynamic behaviour information: well placement, production schedule, model initialization

(OWC, natural state…), boundary conditions

3. Static model information: volumetrics, heat capacity

4. Petrophysical, geochemical and thermal properties distribution

5. Geology representation (facies modelling, fracture modelling, faults…)

6. Input data

7. Uncertainty handling and back-of-the-envelope sensitivity ranking



Workshop Session

What Challenges Drive Subsurface Modelling? A Discussion across Sectors C. Baxter (Seequent)

Representatives from Seequent’s Geothermal, Mining & Minerals and Civil & Environmental teams discuss their experiences in developing the sector specific Leapfrog geological modelling products, instead of a One-Size-Fits-All solution. This will lead into an open conversation to explore the challenges that can accelerate or delay the adoption of geological modelling. Set up Given Seequent has experience across Mining and Minerals, Geothermal and Civil and Environmental industries we will look to have a panel of technical representatives from each of the sectors and a chair person to host it. All representatives are technical people within their field being geologists or engineers. The Civil and Environmental representative will be on a live link. Schedule One of our objectives is to have as much interaction with the audience and generate discussion. We will do 30 minutes panel discussion that will help in providing some context and 30minutes open to the audience for questions or comments about their experience. It could also work for the chair to ask the audience questions as well. Possible Questions to Discuss

• A little history on how Seequent has entered each market and the impact we have seen o Introduces what Seequent is as well as how the company as developed to give

context to audience.

• Why do we have separate products?

• What approaches to we have for each market to fill gaps in modelling requirements both now and looking to the future?

• What pan sector challenge trends do we see?

• We see acknowledge differences across sector, what regional differential trends are there?



Main Session Talk

Production-mode geophysical surveys and geological principles applied to

shallow-water shallow-earth 4D models in industrial harbors: Murphy’s

mapping method applied in northeastern United States

William F. Murphy III, W. Bruce Ward, A. Beckett Boyd, Matthew B. Art, Daniel

A. Rosales Roche, James R. Trotta, Salvatore Triano, Kurt Schollmeyer, Lisa M.

Stewart, Dillon L. Sparkman, Hui Long, Alex Bishop, Coleen K. Murphy

(e4sciences)

For decades William F. Murphy III (1951-2017) was impatient with the rate that geological data was produced, interpreted and utilized. For geology to influence decision making, Murphy envisioned teams mapping and measuring geological phenomena and producing in real-time 4D digital geological models. His goal was to combine the routine and immediacy of medical imaging and diagnosis, oil-field technology, and geological principles. For the last two decades Murphy guided e4sciences in focusing his vision on production-mode geological and geophysical surveys to create shallow (<30m) 4D earth models in shallow (1-17m) waters of active industrial harbors. Much of this focus has been on New York-New Jersey and Boston Harbors. 4D earth models can and should influence the design, engineering, construction, inspection, and monitoring of marine structures, such as bridges and docks; dredging, environmental assessments; and tunneling and directional drilling beneath waterways. e4science’s measurement-based earth models have guided key economic and engineering decisions for many aspects of harbor-based projects. For such projects e4sciences routinely maps the 3D distribution of geological and manmade materials. This includes the 3D distribution of industrial sediments, top of bedrock, top of competent bedrock, submerged and buried infrastructures and historic structures. Results include geological 3D surfaces, forms and features at a given time.

e4sciences integrates into a single 4D reference frame single-channel and multi-channel seismic

reflection, side-scan sonar, bathymetry, magnetic field, borehole measurements, sediment and rock

sample measurements and descriptions, data. From this e4sciences produces orthosonographs (sonar

images of the seafloor), bathymetric and historical maps, magnetic contour maps, 3D acoustic point

cloud data, seismic, single-channel seismic velocity cross sections, subsurface maps, geological cross

sections and geotechnical boring and sediment core logs.

For example in dredging projects, e4sciences measures top and bottom surfaces of layers of dredge

material and calculates dredge volumes used in design. Because high-frequency shallow seismic

measurements interrogate earth volumes on a scale appropriate to dredging, single-channel seismic

velocity cross sections help identify bedrock that is dredgeable without blasting. In dredging,

measurements and interpretations are tested by dredging and removal of the material within

minutes to a couple of years.

Fundamental aspects of Murphy’s mapping method include:



a. adhere to the principle of equipresence.

b. place data in a single xyzt reference frame,

c. make redundant measurements,

d. ground truth by physical sampling,

e. integrate all data, different data types and at different scales,

f. know resolution and uncertainties of all data

g. base interpretation on cross-cutting relationships, superposition, geological models and forms

h. geology is predictive, use this in mapping

i. test geological hypotheses with geophysical surveys, test surveys with sampling

j. treat infrastructure as geological features

k. incorporate historic maps and bathymetric data

l. design maps and cross sections to be as self evident as possible

m. use survey grid to evaluate error form interpolating between 2D data to create 3d surfaces and

volumes

n. adapt scale and frequency of measurements to capture geological phenomena under

consideration

o. keep measurements in larger context



Main Session Talk

A Dynamic Model for Solution-Mining Caverns L. te Kamp (ITASCA/GuG)

When designing a cavern field, the size of the single caverns and their location is set. Very often, the

design shape is a cylinder with a cupola, and the design locations follow an equilateral triangular

scheme. But due to various reasons, caverns are very often not cylindrical, and in cavern fields, there

will be an interaction, which depends not only on the shape, but also on the distance.

Numerical modelling and stability analysis for caverns is usually done based on single design caverns,

i.e. perfect cylindrical cavities. But because of the deviations from the design, this is not sufficient for

most of the real cavern fields.

Therefore, a dynamic model has been developed for the simulation of solution mining cavern fields.

The internal shape of the caverns is controlled by annual or bi‐annual sonar measurements and is

considered in the simulation.

The dynamic tool considers the real shape of the caverns and their growth over time, the real

geology, topography, geological intrusions, filling with air, brine or mud, and the closure of caverns.

The results of the numerical modelling are compared with measured displacements in drifts and show

a very good match in the magnitude of the displacements as well as in their distribution.

The dynamic model is used for annual stability analyses as well as for prediction simulations, which

are used to optimize the future solution scheme from the safety and the economical aspect as well.



Main Session Talk

Predicting the Future with Subsurface Models: How Well are We Doing Now? K. Heffer (Reservoir Dynamics)

A focus of this workshop is the ability of models to accurately predict the future evolution of the subsurface in industrial contexts.

This talk examines the current skill of modelling forecasts of hydrocarbon reservoir performance. Much effort now goes into estimating the uncertainties associated with simulation forecasts of production from fields, but those estimates are generally produced a priori and are seldom tested post facto. It is contended here that reservoir simulation can progress much more readily if its predictions are widely subjected to measured and publicly-documented tests against subsequent field reality, preferably blind tests conducted under the auspices of a disinterested party. By ‘improvement’ is meant achieving both a reduction in bias (loosely speaking, on average in an ensemble, predictions are neither too optimistic nor too pessimistic), and also a more concrete estimate of uncertainty (such that actual production rates do indeed fall within predicted confidence levels in the statistically appropriate number of cases).

Whether or not the ability to improve also requires the incorporation of other means of ‘improvement’, such as more simulator grid-blocks, different schemes of discretisation of the differential equations of fluid flow, better reservoir descriptions, extra coupled physics, improved history-matching, calibration against new monitoring methods etc., can only soundly be judged once we know quantitatively how well current technology is generally faring, and can also quantitatively compare performances with the incremental technique(s) added.

In an analogous context, the UK Meteorological Office does continually monitor the accuracies of its weather forecasts encapsulated in a set of indices; as a consequence, it can quantitatively demonstrate a continually improving forecasting skill over past years. Such a culture is likely to also assist in improving the forecasting skill in the oil industry.

An illustrative analysis has examined pre-existing forecasts and the actual production levels over the same periods to assess and compile errors and their trends. The database derived from 45 individual forecasts involving data from 19 fields or regions with a large range of characteristics. Forecasts had been made at various times of start and finish over the period 1971 to 2008 and involved a range of methodologies (simulation, trend analysis, material balance etc). Some results from the analysis are:

• The overall mean proportional error = -11% (i.e. a pessimistic bias)

• The overall standard deviation of proportional error = 38%

A related issue is whether the physics in conventional reservoir simulators is adequate: this may well be a prime influence on the skill of simulation in forecasting; it also affects the reliability of simulations for their other purposes: viz. (i) to gain a better understanding of the physical mechanisms behind reservoir behaviour; (ii) to compare the predictions arising from implementations of various field development schemes, assuming that the differences between predictions remain meaningful despite absolute errors in the basic simulation model. Field data that point to geomechanics playing a strong rôle in the behaviour of most reservoirs, including recent concepts and analyses, are used to illustrate this issue.



#National Surveys Keynote Talk Modelling the Solid Earth: A Geological Survey Perspective K. Royse (British Geological Survey)

This talk will track the development of geological modelling at the British geological survey since 2000. It will discuss the difficulties of modelling using imperfect and incomplete data (often third-party data), development of a national digital framework for modelling in the UK, supporting a large and diverse stakeholder community, and the use of novel visualisation technology from the gaming industry in order to communicate subsurface geology. Finally, we will look at the future of modelling at the survey, including the development of digital twins, process modelling, use of structured and unstructured data within models, Machine learning, coping with uncertainty and the impact of new digital technology on the horizon.



Main Session Talk

Translating Compositional Earth Models into Realistic Computational Models with Consideration to the Complexity of Geological Constraints A. Long (Subterrane)

There are two main Earth models we use to consider the subsurface structure, that is compositional

and rheological. In geophysics, we utilize residual anomalies to bear some semblance to the structural

changes in the subsurface that can be constrained by other independent observations. Geophysical

modelling may be 1D (a series), 2D (profile or plan), or 3D (parameterized, hybrid, or regularized

mesh). Sometimes we classify a forward or reverse (inversion) model dependent on the purpose.

Forward models are generally used to test geophysical responses, for example, in survey design.

Reverse, or inversion models attempt to fit observed geophysical measurements to a model of

equivalent geophysical properties constrained by known hard data, whilst minimizing the mis-fit

error, to solve a geological question.

In order to acquire measured data meaningful to the long wavelength structure in the crust and

mantle layers of the Earth our measurements are limited to those that can achieve global coverage.

We are therefore limited to satellite gravity and magnetics, and the networks of permanent

seismometers such as published by IRIS, and regional survey networks, for example, the West African

and SAMTEX MT, and the legacy Kenyan Rift International Seismic Project (K.R.I.S.P.). Deeper

structure relies on seismic tomographic models of p and s-wave velocities. There exist other hybrid

mantle composition models based on density conversions, but all the tomographic models rely on

some form of regularized 3D cell inversion of spatially under-sampled 1D vertical profiles of velocity

structure.

In order to achieve better unifying theories of the Earth’s structure, we must also consider qualitative

methods of understanding the structure. Therefore our geophysical models should conform to the

geological surface exposure and geological inference of bedrock subcrop structure. If we abide by the

compositional model of the Earth, then the structure we see from the measurement of geophysical

properties and their anomalies should bear good correlation to the mapped surface geological

expression and rock exposure.

Here we present several examples of the correlations that can be made from different geophysical

methods employing different quantitative and qualitative methods that help us understand the

longer wavelength structure and its impact on regional stress fields, using satellite potential fields

data, seismic tomography, seismic refraction and MT station networks. We will explore the

correlation with geology at a variety of scales (from continental to prospect level) and how this helps

us understand the fourth dimension, time, that defines the evolution of geological processes. Utilizing

data from the African continent, consideration will also be given to different theories we

commercially use and their influence on our present day soft and hard rock 3D models in basin

exploration, and mineral ore modelling respectively.



Main Session Talk

First History-Matched Full Field Coupled Geomechanics & Reservoir Flow Model of both the Valhall and Hod Fields, including Water Weakening & Re-Pressurisation J. Kato (Rockfield)

Valhall-Hod is two connected large chalk fields in the Norwegian sector of the North Sea. The reservoirs consist of highly overpressured, (0.84 psi/ft) chalk. The high overpressure and early oil migration has resulted in very well preserved porosity, exceeding 50% in parts of the field. For more details see [1, 2, 3]. The compaction is providing an excellent source of reservoir energy, accounting for 50-60% in certain areas of the field and need to be properly accounted for in the reservoir model, which can be a challenge without a geomechanics model. The subsidence is currently approaching 7 meters below the central platform complex at Valhall and is around 1 m at Hod.

Water injection started in the field in 2004 with a gradual step up in rate. Chalk is known to be weakened by seawater injection [4], so this aspect needs to be included in the model as well. There has been several geomechanics models published from Valhall in the past, this paper will discuss the results of the previous publications and compare them with the most recent model applying technology not used in previous studies. This paper also presents the first numerical prediction of compaction and subsidence in the Hod field, since the Hod field is included in the same model.

This paper documents the development of a coupled geomechanics model (ELFEN) based on the finite element method coupled with a multiphase reservoir flow model (NEXUS) based on the finite difference method. The paper covers the work to couple the two codes for this project. The paper covers the calibration of a suitable and effective constitutive model to account for strain rate dependent reservoir compaction during depletion, re-pressurization and waterflooding.

The constitutive deformation model is driven by pore pressure and watersaturation calculated in the reservoir flow model and passed to the geomechanics model which is calculating the change in pore volume which is then passed back to the reservoir flow model. The initialization requires an initial stress state and it will be presented how this was defined.

We also included faults in the model and check for re-activation and amount of slip on them. They are entered as double sided surfaces. One special requirement for the model was to honor thin, down to 20 cm thick very stiff chalk (hardgrounds). Dealing with such thin layers in a big elasto-plastic model exceeding 3 million finite elements is challenging. An explicit formulation including mass scaling and parallelization was used to make the model as effective as possible. The Valhall-Hod model can also be constrained by a relative large amount of field and surveillance data, like GPS, time lapse seafloor bathymetry maps, radioactive markers in the reservoir and overburden as well as 4D seismic. There are also other high impact business applications of the model prediction, besides seafloor subsidence forecasting, related to well planning in terms of drilling and well survivability.



Main Session Talk

3D Geological Modelling of the Chalk Group in SE England for Engineering & Groundwater Applications: The Future is 4D? C. Cripps (British Geological Survey)

Catherine Cripps1*, Joanne Thompson1, Rowan Vernon1, Mark Woods1, Andy Farrant1

The Upper Cretaceous Chalk Group (‘Chalk’) underlies much of SE England, and understanding its 3D

spatial heterogeneity is important for both engineering and hydrogeological applications. It supplies

75% of groundwater for public consumption, and is the host bedrock to many major infrastructure

and engineering projects (e.g. HS2, Thames Crosslink).

Traditionally, the Chalk has been divided into a tripartite scheme of Lower/Middle/Upper Chalk,

based on biostratigraphic zonation. However, during the last 30 years or so, there has been a Chalk

stratigraphic revolution and currently the Chalk in Southern England is divided into nine formations.

Crucially, these formations are based on gross lithologic and secondary characteristics, and which

therefore follow changes in hydrogeological, geotechnical and engineering properties. The British

Geological Survey (BGS) over the past 20 years has incorporated this ‘new’ scheme into a detailed

mapping programme covering much of SE England. As such, it has allowed variations in structure

(such as enhanced fractures in particular horizons) and lithology (i.e. flint horizons, marl seams and

hardgrounds) to be mapped out. This has particular relevance to the engineering and

hydrogeological communities as this geological level of detail has huge implications in geotechnical

characteristics (e.g. flint bands in tunnelling) and groundwater flow.

At the BGS, the new Chalk maps (incorporating the new stratigraphy), and in combination with

lithological and geophysical boreholes, and remotely sensed data, have been developed into a series

of 3D models, using GSI3D and Groundhog software. These are implicit modelling tools that not only

allow for the input of these data types, but also capture the geologist’s inherent understanding of the

strata. Lastly, the Chalk models are built at various scales according to the need of the end user.

Against a background of increasing stress in the SE in terms of population size and growth, aging

infrastructure and accelerated anthropogenic climate change, these models are of increased interest

to engineering/geotechnical companies, water companies and the Environment Agency (in its role as

the environmental regulator). Currently, the Environment Agency have commissioned a series of

catchment scale models in the Chilterns, and these are being used to inform groundwater models.

4D modelling has always been of importance in the groundwater community, and it is the feeling that

this is just the beginning with regard to Chalk 4D modelling approaches. This presentation will further

look at how we potentially evolve and integrate static implicit 3D geological models into the fourth

dimension.

1 British Geological Survey, Keyworth, Nottingham



Main Session Talk

Development & Application of a Dynamic Fracture Growth Approach to Generating Discrete Fracture Network Systems in Sedimentary Rocks M. Cottrell (Golder Associates)

The development and use of the 3D Discrete Fracture Network (DFN) approach provides an accepted structure-based framework for the characterisation and analysis of both static and dynamic properties of naturally fractured rocks. It has become a sound basis for describing fracture attributable performance across different industry areas. DFN fracture networks have become a trusted structural representation for developing oil and gas reservoirs, providing integrated site descriptions for nuclear waste repositories, through to geotechnical based assessments for the civils & infrastructure areas.

The philosophy for building DFN models has historically been a static one, in that the present-day system of fractures are a combination of deterministic and stochastic structures generated individually from direct observation or from statistics of key fracture geometrical attributes taken from measurements. These attributes are spatially employed for populating the key DFN fracture parameters of orientation, intensity, size and shape.

The stochastically driven DFN approach is routinely able to produce large scale DFN models, often containing several 10’s - 100’s millions of natural fractures, that honour the observed statistics of the physically fractured rock. Although the fracture systems may produce several matched statistical characteristics of the measurements, other indirect properties of the fracture network can often be neglected or ignored. Such models do not honour geomechanical principles of how the fractures formed, evolved and interacted with each other and the intact rock.

It has been shown by various researchers that termination relationships among fracture sets are very significant for hydraulics, since fracture systems with non-zero termination percentages generally result in more well-connected than non-terminating systems with identical fracture intensities P32 values. Network characteristics such as fracture cross-cutting relationships are often poorly reflected in large scale DFN models. Failing to describe phases of brittle deformation and fracture terminations can lead to poorly constrained descriptions of fracture size and importantly network connectivity. These in turn are known to be very significant controls on fracture permeability, anisotropy and channelling of fluid pathways.

This paper presents an evolution of the flexible geocellular DFN generation approach whereby mechanistic based rules are used in conjunction with grid-based maps of the geometrical fracture attributes. This results in a new DFN approach where natural fractures are dynamically “grown” with time, in full 3D, until they meet the predefined fracture characteristics as determined from well data, outcrop maps, high resolution seismic, geophysical lineaments or tunnel mapping. The incremental growth of the natural fractures provides opportunity for fracture connections to be monitored, and the intersections of pairs of fractures be resolved in a more consistent manner. The approach permits growth, termination and linking of multiple fracture sets such that the DFN fracture network is more geologically founded whilst still honouring the observed statistics. Fracture size distribution and network connectivity are then emergent properties, that can be further constrained by calibration on well tests.

This paper will provide an overview of the approach and provide demonstration of the generation approach for several problem types taken from different industries.



Main Session Talk

Wellbore Instability & Deformation Risks Highlighted by 4D Seismic & Geomechanical Studies M. Calvert (Total)

Monica Calvert, Frederic Bourgeois & Frederik Ditlevsen (Total)

In the Danish sector of the North Sea there has been an increased focus on integrating 4D seismic and geomechanical studies to highlight areas in the overburden at risk of wellbore instability and deformation due to depletion-induced reservoir compaction and seafloor subsidence. Wellbore deformation due to overburden subsidence can limit the ability to pass beyond a certain measured depth in a well (Figure 1). This limit is known as the hold-up depth (HUD).

By 2017 several wells in the Tyra field had undergone wellbore deformation and the delivery of new wells in the Tyra SE field had also proved challenging. In each of these cases, there was a spatial correlation between the HUD locations and the time lapse seismic attribute considered to be a proxy for shear strain in the overburden (Figure 2).

Present day seafloor subsidence in Tyra is approximately 5m with geomechanical modelling suggesting a total of 6-8m total seafloor subsidence in 2042 when the license expires. Due to the seafloor subsidence, the Tyra field will be shut-in from 2019 until the new facilities are in place and production can resume in 2022.

An integrated study combining the 4D seismic, reservoir model and geomechanical model was initiated to predict which wells, if any, would be at risk of deformation or potentially failure in the future and hence have an impact on the project economics. Results from this study and additional examples showing the correlation between the 4D-derived high shear strain zones and drilling challenges in the Tyra SE field are presented here.



#Technology Keynote: 3D Holographic Visualization of 4D data

M. Lato (BGC Engineering)

Robust and effective outcomes on geotechnical projects emerge when the design is based on a thorough understanding of the geology and the environment, and the interaction of these systems over time. Traditionally, a significant challenge faced by geo-professionals is our ability to observe, interpret, understand and visualize the physical environment, particularly as it applies to changes over time, and the effect of those changes on the as-built environment. To address this challenge, we brought complex projects to life by developing a process that converts information and data traditionally reported as 2D drawings, to be visualized in true 3D on a holographic visualization platform. The platform allows participants to visualize 4D data in 3D at scales from high level overviews to true life boots on the ground viewpoints. Visualizing information 4D data in 3D, on a holographic visualization platform, provides project participants with an unprecedented view of a project’s scope, scale, and technical details. Environments can be animated and presented at true scales to immerse the user in a project, from anywhere in the world. Furthermore, users can share environments and explore a design together. Holographic visualization allows people to interact with, and control their experience, which facilitates the generation of a deeper understanding of 3D environments and their changes over time. This is especially important for the comprehension of technically diverse project teams and external stakeholders. Holographic visualization is enabling clearer communication and more confident decision making. It will also help identify, anticipate, and mitigate challenges earlier within the project lifecycle by being able to visualize the site in 3D on a boardroom table or standing in the landscape. It creates a shared vision, a common reality that allows meaningful input and decisions to be made based on science, facts and evidence.



Main Session Talk

Data & Models: How do they Interact? An Example of how Gravity & Seismic Monitoring Constrain Subsurface 4D Models O. Eiken (Quad Geometrics)

4D subsurface models are no better, or no more valid than the data which constrain them, one may argue. As science itself, they are founded on observations. In spite of this, the professional disciplines of “modelling” and “measuring” seem to have a lack of overlapping individuals and practices. High-quality time-lapse data can help facilitate the communication between the disciplines. 4D seismic has had some effect of breaking down the silos in the petroleum industry, with conceptual data interpretation and automated history matching as two distinctly different workflows.

A North Sea CO2 storage case will be briefly described, with the starting point: How can we learn from a full-scale injection for predictions in this and later projects? In spite of excellent quality 4D seismic and gravity monitor data, “modellers” have been slow at picking up the wealth of information inherent in the data, and “measurers” have been slow at interpreting the details of the observations. Who shall decide the most important questions to ask and the model matches to accept? After more than 15 years of international research, flow models which include necessary temperature effects have finally been matched to the data.

Gravity changes are straightforward to interpret, and also such data can work as “enablers” for cross-disciplinary communication. Many subsurface processes, whether natural or man-made - for exploitation of resources, involve mass changes or mass redistributions, which result in changes in the surface gravity. Gravity changes may be monitored by accurate instruments and careful measurement and survey design. Today, changing water level in shallow aquifers, geothermal and gas reservoirs at several km’s depth are monitored by gravity measurements. Water and air/gas have particularly large density contrasts and can hence cause large gravity changes. Precision has improved steadily over decades, with 1 µGal relative precision now achievable in aerial surveys, and with some 10s of nGal relative precision for stationary measurements over days or weeks. This has widened the range of applications, and new ones are yet to be identified.

Mass redistributions or changes may also cause subsurface rock strain, which propagate to the surface, via complex geomechanical mechanisms. Surface movements will affect surface gravity, and strain and gravity change observations are thus coupled. There exist a variety of methods for measuring both vertical and horizontal surface strain, like optical leveling, GPS, InSar, seafloor pressure and tilt changes. Vertical movements of a few mm can be monitored both on land and at the seafloor. It may vary from case to case whether modeling uncertainties are larger than the observational uncertainties of strain measurements. In our experience, valuable subsurface compaction mapping is achievable from surface vertical movements by using relatively simple geomechanical models.

Further technology developments are likely to give even more precise measurements in the future, and also less costly data acquisitions. This will require improved modeling capabilities, if the modelling algorithm precision shall not become the weaker part of the workflow. The interaction between data and models will likely continue be a fruitful arena for future improvements.



Main Session Talk

Integrated Subsurface Modelling Framework for Unconventional Development R. Guises et al (Baker Hughes)

Unconventional resource development has grown significantly in many basins worldwide. While unlocking hydrocarbons from low permeability formations requires the use of new technologies, its economic viability is strongly linked to reliable characterization of the subsurface. Typical unconventional plays are complex, exhibiting spatial heterogeneities, natural fractures, varying in-situ stresses and challenging pore pressure predictions, all of which leading to elusive production models. In this context, an inter-disciplinary approach connecting seismic, geological, petrophysical, geomechanical and stimulation models is proposed in order to produce a model capturing the mechanisms driving the subsurface response during and subsequent to the stimulation phase. The holistic approach to describe the subsurface allows development of robust extrapolation strategies from calibrated wellbore geomechanical models to the field scale. Subsequently, a state of the art hydraulic fracturing numerical simulations, capable of capturing the hydraulic vs. natural fractures interactions or rock strength anisotropy, are performed in order to assess the formation response to various injection plans. The presentation will share the methodology developed for the preparation of the 3D subsurface model and present a study showcasing the value of cross discipline integration. Additionally, the numerical simulations constitute a rigorous basis for optimizing well placement and reservoir contacts, improving stimulations programs, hydraulic fracturing efficiency and ultimately the field economics. Finally, capturing the complex interactions between hydraulic, natural fractures and laminated bedding provides a better understanding of potential risks associated with unconventional developments.



Main Session Talk

A Reliable Organisational & Procedural Framework for Maintenance of an Ongoing Mine Stability Safety Case in 4D D. Beck (Beck Engineering)

Mines must be able to confirm at all times that the mine is stable and safe and that forecasts of deformation are sufficient. For large scale hazards including mine scale instability, large seismic events or inrushes the analysis and decision making process during design and operations is however hard to prescribe. Too often, comprehensive databases of real time measurements are left unused by engineers until after a serious none failure. As many accidents have occurred in part because of failings such as these, it is the authors view that mining jurisdictions will eventually legislate that a global stability 'safety case' be properly maintained as a condition of ongoing operation. The guiding principle would be, that if at any time the mine cannot disconfirm adverse scenarios, the mine will stop and evaluate the causes and update the plan, performance forecasts and safety measures. This paper discusses one reliable organisational and procedural framework for maintenance of an ongoing mine stability 'safety case'. The structured method involves sufficient measurements, 4D numerical analysis and ongoing appreciation of forecasts versus measurements to confirm the stability state of the mine.



Workshop Session

Have Advances in Subsurface Modelling Enabled Higher Quality Decisions? Geoff McKinley (Mineral Advisors)





Poster Programme

Natural Cosmic Radiation as a Density Probe of the Subsurface T. Botto

Cornish Lithium: A New Metal from An Old Mining Area. Lithium exploration in SW England A. Matthews (Cornish Lithium)

Explicit Modelling at the BGS for the Infrastructure Sector S. Thorpe (British Geological Survey)

The Augmented Reality Sandbox V. Lane (University of Leicester)



POSTER Natural Cosmic Radiation as a Density Probe of the Subsurface T. Botto In this paper we wish to give a brief overview of the applicability of so-called cosmic rays as a passive source for both 3D or 4D bulk-density mapping of the subsurface. Naturally occurring cosmic radiation is a well-known, penetrating nuclear probe that, analogously to X-rays, can be used to infer a density map of the subsurface by using suitably designed and well positioned nuclear detectors. In most practical scenarios, however, one must deal with the fact that this flux, while free, is not constant in time nor in space, and that the measured data-sets are typically incomplete, i.e. they cannot be easily inverted to provide a voxel by voxel density map. This has typically limited the application space to the detection of gross features, such as large voids in ancient pyramids. Nonetheless when more properly viewed as an aid to benchmark geological models or to monitor dynamic density or saturation changes in the underground, cosmic-ray density measurements appear to be well suited for a number of practical applications in mining, civil engineering as well as oil production. Historically, another practical obstacle has to do with a lack of detector configurations suitable for borehole deployment, however we will show that some are indeed possible at a relatively low cost by integrating off-the-shelf components. The necessary trade-offs between detector and installation costs, observation time and spatial/temporal resolution will be highlighted and results from a few exemplary test-case studies will be presented.



POSTER Cornish Lithium: A New Metal From An Old Mining Area - Lithium exploration in SW England Adam Matthews (Exploration Geologist, Cornish Lithium)

Cornwall is a polymetallic district that has produced a variety of metalliferous products since the Bronze age. From the 16th century, the county became a major European producer for tin and copper, fending off numerous changes in metal price, until the international tin market crash of 1985. Mines in Cornwall eventually closed due to the falling metal prices (rather than a lack of ore), leaving vast amounts of data remaining in archives, private collections and county records. Many of the district’s mines were very wet and required constant pumping. At United Mines (~4 km east of Redruth), hot springs (temperatures of up to 51 °C ~500 m below surface) ingressing from mineralised faults and dykes prohibited the miners progress. Historic geochemical analysis of the hot springs (Miller, 1864) identified that lithium was present in elevated concentrations. Lithium was also identified within hot springs encountered in many other mines across Cornwall. As the world transitions towards a low carbon future powered by renewable and green energy and significantly by the demand for electric cars, the global demand for lithium is predicted to increase to 785,000 t Lithium Carbonate Equivalent (LCE) by 2025 (Roskill), from 217,000 t LCE in 2017 (Reuters). The industry must discover new sources of lithium to satisfy demand, and so understanding the distribution of these lithium bearing geothermal fluids, and the structural features that control them in 3D, is vital for Cornish Lithium’s exploration success. Due to the extensive mining history in Cornwall, there are vast archives containing underground mine plans, cross-sections and long-sections. Academic papers and journals produced at the height of the mining boom also give detailed insight into the positioning and chemistry of the hot springs. Cornish Lithium is compiling this data into their 3D model, which provides accurate, high-resolution information on lithium bearing fluids and geology. Cornish Lithium’s model is constantly expanding to create a representative fracture model for Cornwall, validated via correlation with existing 3D mine plans and boreholes. There are 2 significant fault sets in Cornwall which trend approximately E-W and NNW-SSE. The E-W faults host the Sn and Cu mineralisation. Hot spring data is mainly documented within these mineralised structures, as the springs were encountered whilst the structures were being mined. Despite the lack of data within NNW-SSE structures, they may be the best conduits for fluid flow due to the orientation of the current tectonic stress regime and reactivation history. Cornish Lithium’s current exploration model predicts that NNW-SSE structures will have naturally enhanced permeability as structures parallel to the maximum compressive stress will dilate. Cornish Lithium’s next challenge is to assign additional parameters which are more commonly used within geothermal and oil & gas industries (e.g. porosity, permeability) to their geological model. Leapfrog Geothermal, combined with hydrogeological modelling packages, will assist with their interpretation of fluid flow and the distribution of lithium enriched fluids. Understanding how the fluids move and interact with the host rock will greatly inform their exploration programme.



POSTER Explicit Modelling at the BGS for the Infrastructure Sector S. Thorpe (British Geological Survey)

This poster will discuss the technological advances made by the BGS in the last 2-3 years particularly highlighting work done to connect BGS with industry including civil engineering, site/ground investigation organisations, and infrastructure projects. Much more emphasis is being placed on working closely with partner organisations in order to align the research and modelling done at BGS with the needs of our stakeholders. This has never been more necessary and poignant as BGS continues to aim for more data availability to a wider audience through organisations like the Geospatial Commission. A model is only as good as the data that is put in, and we at the BGS have seen an increased level of use of our website features such as the GeoIndex and Geology of Britain viewer, which has specifically opened up the borehole data collection to the industry on a scale never seen before. The explicit modelling methodology employed by the BGS over the last 20 years is reliant on two things: good quality data that is easily available, and the knowledge of geologists and engineers. The explicit method allows our clients to understand the model and be more engaged and therefore able to update the model post-delivery. The BGS modelling teams are investing more in ensuring closer links with industry software. We have developed a free piece of software for anyone to use called Groundhog Desktop, which is a borehole and cross-section drawing package, and we continue to develop it further by looking towards a grid/surface calculation and also a 3D viewer. Groundhog can import/export a range of different formats, which allows greater flexibility for integration into other workflows and processes. We want to promote the idea of sharing further by encouraging people to think about their data more widely, and about how industry and our stakeholders can create a much more improved service of data provision by supplying their data to us in our remit as the National Geological Repository. Our most recent project called Dig-To-Share is aiming to do just that, by investigating the perennial issues associated with sharing data, and trying to break down any barriers to preventing data being shared such as liability and legal issues and time/resource issues. The poster will discuss several examples of how working together has paid off for industry and BGS alike. From projects like Farringdon Station, to the electrification of the Leeds-York railway by TSP and T&P Regen’s Ashton Gateway modelling of a contamination plume, the data supplied by BGS has enabled a more efficient design through enhanced geotechnical knowledge and certainty, which has produced savings in time and money.



VIDEO

The Augmented Reality Sandbox V. Lane (University of Leicester) Dr Victoria Lane is a research associate seismologist working for the NERC funded seismic facility, SEIS-UK, based at the University of Leicester. SEIS-UK provides equipment, software, training and support for onshore seismic projects anywhere in the word. Loans are free of charge to universities and other institutes engaged in environmental research within the NERC remit (http://seis-uk.le.ac.uk). Victoria has a keen interest in outreach and engagement projects and has been working with the National Youth Agency, National Space Academy, Leicester City Football Club Community Trust and the Schools Seismology project at the British Geological Survey on the ‘Geophysics in a Box’ project. This initiative, funded by the Royal Astronomical Society, aims to engage community groups and school children with STEM science by providing links to real-world, sport-inspired data, which in turn can improve their understanding of the physical properties of the earth. The augmented reality sandbox, presented here, is a new display at the University of Leicester. Set up by Victoria & colleagues and using code developed by University of California, Davis W.M. Keck Center for Active Visualization in the Earth Sciences, the sandbox will be used by the School of Geography, Geology and the Environment to improve understanding of the 3D subsurface and its 2D representation.

http://seis-uk.le.ac.uk/

http://seis-uk.le.ac.uk/



Burlington House

Fire Safety Information

If you hear the Alarm Alarm Bells are situated throughout the building and will ring continuously for an evacuation. Do not stop to collect your personal belongings. Leave the building via the nearest and safest exit or the exit that you are advised to by the Fire Marshall on that floor. Fire Exits from the Geological Society Conference Rooms Lower Library: Exit via main reception onto Piccadilly, or via staff entrance onto the courtyard. Lecture Theatre Exit at front of theatre (by screen) onto Courtyard or via side door out to Piccadilly entrance or via the doors that link to the Lower Library and to the staff entrance. Main Piccadilly Entrance Straight out door and walk around to the Courtyard. Close the doors when leaving a room. DO NOT SWITCH OFF THE LIGHTS. Assemble in the Courtyard in front of the Royal Academy, outside the Royal Astronomical Society. Please do not re-enter the building except when you are advised that it is safe to do so by the Fire Brigade. First Aid All accidents should be reported to Reception and First Aid assistance will be provided if necessary. Facilities The ladies toilets are situated in the basement at the bottom of the staircase outside the Lecture Theatre. The Gents toilets are situated on the ground floor in the corridor leading to the Arthur Holmes Room. The cloakroom is located along the corridor to the Arthur Holmes Room.



STAFF ONLY

MAIN RECEPTION

MAIN ENTRANCE TO GEOLOGICAL SOCIETY

Documents

4D Subsurface Modelling: Predicting the Future/media/shared/documents/Events/2019… · 4D Subsurface Modelling: Predicting the Future – Burlington House, 20-21 Feb 2019 20th-21st