Why model the subsurface?

In the months leading up to the "4D Subsurface Modelling: Predicting the Future" conference in London in February 2019, the co-organizers, (@glenburridge, @thomasfinkbeiner,@richardplumb), will be posting several short articles exploring the event’s themes. We hope these articles will stimulate some discussion among geoscientists, engineers and managers working in the Civil Engineering, Mining, and Energy sectors on the justifications, purposes and pitfalls of subsurface models and spur them to participate in the conference. We’d love to hear your feedback!

In our first article, we pose the question of why we try and model what’s happening below the Earth’s surface during a project’s life and discuss the role and benefits of subsurface modelling in three important industrial sectors geoscientists work in - Civil Engineering, Mining and Energy.

First, What is a subsurface model?

For our purposes, a model is a representation of soil and rock formations below the Earth's surface that cannot be seen or experienced directly.

Typically, models are constructed from knowledge of: 

·     the geologic history of the region,

·     interpretation of geophysical data, e.g. well logs, seismic, gravity, magnetics

·     soil mechanics and/or rock mechanics theory,

·     laboratory measurements on direct core or sediment samples,

·     experience from past operations, and

·     field measurements.

The objective of this model is to capture current understanding about:

·     the geometry and spatial extent of subsurface formations,

·     the formations’ material properties, geochemical and mechanical behavior, and

·     the contemporary stress field acting on them and how it changes in response to human activity, such as drilling, excavation, production, injection etc.

A model may start as a "cartoon" in a PowerPoint presentation, but to be most useful it soon evolves into a numerical representation in a software modelling package. Such models may be in one to four dimensions:

·     One dimensional models are linear collections of information that come, most commonly, from a vertical wellbore. These will have a limited zone of validity but will be sufficient for some purposes during exploration.

·     2D and 3D models are generated through correlation in additional dimensions and contain spatial information about soil and rock formations and their physical and transport properties. Their aim is to inform engineers about the subsurface environment of interest.

·     4D, or time-dependent, models use repeated iterations of geometrical and material property information in a geomechanical simulator to predict if, where, and when subsurface deformation may occur. 

A classic example of the highest order, 4D, modelling is in the Oil and Gas sector. Such models are used to study the effects of compaction of a reservoir and surface subsidence due to production of groundwater or hydrocarbons or the effects of hydraulic fracturing and fault movement due to fluid injection.

Minimize uncertainty and costly surprises

In Civil Engineering, models typically extend to depths of 100m or less in urban environments and to 2.0-2.5 km for deep railway tunnels. In the Mining, Coal, Oil and Gas, and Geothermal sectors, models typically extend from 1-5 km or more. Whether the models are shallow or deep, it should come as no surprise that there can be considerable uncertainty associated with models.

Where there is uncertainty lurks the risk of costly surprises. So, one of the main reasons for modelling is to help engineers reduce uncertainty in the design and construction of surface and subsurface structures.

Think of the massive tunneling machines like Big Bertha employed in the Alaska Way Viaduct project or in the Crossrail project under London. Civil engineers want to increase the use of underground space in an efficient and sustainable manner-they don't want to run into another subway tunnel or damage structures above ground. Similarly, mining engineers want to locate and economically extract minerals without causing collapses trapping miners and equipment. In the Energy sector, subsurface professionals want to reach and produce oil and gas in deep offshore reservoirs without their wells collapsing or blowing out like the Macondo well in the Gulf of Mexico. In the world of geothermal energy, power companies want to extract heat energy without inducing earthquakes that shake up their customers. 

One dimensional, 2D and 3D models help reduce operational risk during a project's design, construction and operational phase. Four dimensional models help to manage risk during an asset’s operational phase when minerals, oil, gas or heat are extracted or underground structures need to be kept operational. Accurate models have the potential to help all sectors through the project’s life and enable asset managers to maximize asset value by reducing operational risk and costs. The more accurate the model the fewer surprises and cost overruns.

Realizing the value

However, realizing this value requires a willingness to acknowledge some uncomfortable factors including that:

·     knowledge of the subsurface is uncertain,

·     there are risks inherent in perturbing its equilibrium,

·     additional information and resources are often required to mitigate operational and financial risk, and

·     upfront investment in data acquisition and interpretation is needed to design and build these models –before subsurface difficulties are experienced.

On a positive note, the very decision to build a model, while acknowledging the foregoing caveats, leads to improved communication, knowledge sharing and informed decision making across project teams. Even simple models can be highly effective for communicating the knowns and unknowns of a subsurface volume of interest (VOI). For example: what is the VOI, where is it, what are its geometry and properties, what is the best path to reach it, what are potential hazards, where are they, where do we need more information? What happens to the VOI, and the region surrounding it, when voids are created, fluids are extracted or injected? What happens if the model is not accurate enough? How accurate does it have to be? What is the trade-off with time and cost? What do we do in the event of an unexpected event? Get the picture!?

Experience has shown that significant value from modelling is obtained if an iterative approach is adopted in a project's planning phase. That is, one builds an initial model early in the planning phase and enables/promotes model updates as new information becomes available. Early modelling informs a project team where critical information is lacking in time to include data gathering/processing in the project budget. Iterative modelling fosters continuous organizational learning about a subsurface asset. This in itself is a risk mitigation strategy against the unexpected.

Avoiding a classic trap

Such a learning cycle of iteration helps projects avoid a classic trap, experienced in all three sectors: namely, starting a model only after some un-planned and expensive event has already occurred, e.g. a tunneling machine gets stuck, a mine shaft collapses, or a well blows out. Experience has shown that in such situations there is not enough time to mobilize staff, to find, process, and interpret data, and to build a model to optimally affect the outcome of the incident. The most valuable models are ones started before the first test boring or the first geophysical survey is ordered.

Bottom line

So, why do we model the subsurface? Because holistic modelling done by a truly integrated and informed project team, using all available data-streams and covering all project stages:

·     improves operational safety,

·     improves project success,

·     helps to protect , the environment, and

·     reduces overall project costs.

Finally, fundamental challenges faced by modelers often lead to development of new geoscience knowledge, new measurement technologies, improved software platforms and more effective engineering practice.

Possible Future topics: Modelling as a risk mitigation strategy, When to model, Who models, Communication tools, Sector case studies.....