In 1994, the Society of Petroleum Engineers gave L. Wietzerbin and Jean-Laurent Mallet the SPE Computer Application Best Paper Award for their article entitled Parameterization of Complex 3D Heterogeneities: A New CAD Approach. This paper proposed the Boolean simulation of channels and lobes represented as a smoothed backbone curve and associated cross-sections (this object is still known as GShape or Channel by Gocad users). That year, the average (inflation-adjusted) oil price was 26$ per barrel (source: 23 years later, the spot oil price is unfortunately much lower than most of us would like it to be, but it is still close to two times the 1994 amount! Both then and now, such descriptions of geological objects are highly relevant not only for thrill of advancing the science of quantitative geology, but also to optimize reservoir recovery. As a matter of fact, the paper by Wietzerbin and Mallet still reflects some essential characteristics of our research for integrative numerical geology: to improve the ability of numerical models to accurately forecast subsurface behavior, we need appropriate geometrical and petrophysical parameterizations of geological objects. Among the main evolutions since this seminal paper, we can mention the emergence of new ways to consider, explicitly or implicitly, the time dimension and the physics to integrate various types of subsurface data while considering uncertainties.

This meeting will, as each year, showcase some recent advances made by the RING team in addressing these difficult challenges.

Let’s now summarize some of the main striking results presented this year. In structural modeling, J. Renaudeau made significant progresses to better characterize the mathematics of implicit structural modeling. He presents a nice work that establishes some relationships between mesh-based and meshless smoothing criteria classically used in structural modeling. This work is a strong contribution to better explain and understand the essence of structural modeling engines and to propose new advances. As a matter of fact, Julien also presents a new idea and very promising 2D results to manage sharp thickness variations in implicit modeling, which has been a longstanding problem of these methods. Another challenging problem for which much progress was made recently through a Nancy-Monash collaboration is the management of folded terranes. This year, Laurent Aillères builds on this method to assess fold-related uncertainty consistently with field structural observations. This method is really powerful, as it is explicitly account for structural fold parameters rather than minimal Laplacian to constrain the parameter space.

Salt formations are also a tremendous source of structural complexity, for which new geomodeling methods are required. To handle this new topic of research, Nicolas Clausolles has started studying seismic attributes and their potential to analyze reflection seismic data and to detect geological objects. He has thus integrated the 3D unconformity attribute proposed by Wu et al. (2016) into the GoScope plugin, as well as a structure-tensor computation and a structure-oriented smoothing filter. The long-term goal of this work is to develop new modeling methods of salt geobodies that directly exploit seismic information and account for uncertainty. The last (but not the least important) structural objects are faults and fractures. In this respect, Francois Bonneau presents some ongoing results to characterize the relationships between pre-existing fractures, microseismicity and fluid pressure by analyzing experimental core-scale laboratory data obtained in the frame of the GEOTREF geothermal project. At a larger scale, Gabriel Godefroy proposes a new theoretical framework to look at structural uncertainty problems in the presence of data. He suggests to look at structural uncertainties using a graph whose nodes represent spatial fault evidences. This allows to shed a new light on previous stochastic structural modeling methods, and opens the path to more efficient implementations and ways to cluster structural model realizations. This new framework is also put into practice in the form of a stochastic structural modeling method which can integrate various structural rules.

In stratigraphy, Jonathan Edwards applied the stochastic correlation method on a North Sea case study, using sequence stratigraphic concepts and biostratigraphic constraints. His work illustrates not only the ability of stochastic stratigraphic correlation to generate plausible stratigraphic architectures, but also the role of this method to gain insights about borehole data interpretation and to highlight the value of biostratigraphic data in reducing subsurface uncertainties. Also on this front, Florent Lallier (Total) builds on the training-based correlation method and on advances in bioinformatics to significantly improve the applicability of the method, in particular for the multi-well correlation problem.

Significant sedimentological variability may exist within strata that are correlated between boreholes. This holds in particular in channelized (turbiditic or alluvial) environments, in which lateral variability can be large. Following the tracks of L. Wietzerbin (and of younger former RING researchers such as Sophie Viseur, Jeremy Ruiu and Guillaume Rongier), Marion Parquer presents recent advances to generate realistic models in these depositional environments. She introduces a new way to come up with possible ages of meander abandonment facies which may be visible in seismic stratal slices. She also presents some new work to simulate such facies consistently with available data during channel retro-migration simulation. These two complementary aspects are essential to improve the conditioning of realistic channel models to subsurface data. Channel modelling is a definitely hot topic this year, as Gavin Graham (Total) and Carl Jacquemyn (Imperial College) also present their work on improving process-like simulation methodologies.

Translating such realistic geometrical descriptions of geological objects down to physical subsurface forecasts is no easy task. It calls for defining appropriate petrophysical values at the scale of interest, and for applying numerical methods to solve the equations of interest. In this volume, you will find a paper by Guillaume Caumon which reviews and discusses the main existing possible approaches to reach this goal. One of the difficulties involve some local geometrical features which can make it impossible to numerically solve the physical problem. In these cases, making the required model simplification can be particularly difficult! Pierre Anquez addresses this very challenging problem by defining a general methodology based on geometric tolerances. The main idea of his contribution is to carefully analyze where the input model should be simplified before implementing the modifications. His first results on 2D models are very promising and shows the potential of this method to be successful in three dimensions. A second difficulty in relating geomodels and physical codes rests with the parameter field and the associated scale. As shown by Paul Cupillard in previous years, this problem can be analyzed and solved in the case of elastic wave propagation when the bandwidth of the wavefield is known. This year, Paul presents some practical uses of this homogeneization method to efficiently simulate wave propagation in complex geological media.

Another practical difficulty in running simulations on geomodels lies in varying software requirements on both ends. This is a reason why the RINGMesh project was initiated two years ago. In this frame, Margaux Raguenel has successfully established a first connection between RINGMesh and the CSMP++ physical solver, in the frame of the GEOTREF project. She shows some simple but concrete examples of this connection which demonstrate that the software infrastructure is now well under way to address complex problems such as geothermal reservoir simulation in the presence of supercritical fluids. In the frame of his LabEx Ressources 21 Postdoc, Gautier Laurent similarly describes another coupling between RINGMesh and the chemical simulation code PhreeqC. These two works open interesting perspectives to address coupled THMC problems on realistic subsurface representations.

Simulating physics on geomodels allows of course to make resource recovery forecasts. It can also help to reduce subsurface uncertainty by incorporating ancillary data, such as reservoir production data, borehole seismic experiments, or mechanical deformation data. Last, it allows to a priori design and test the value of such data in gaining subsurface knowledge. In the frame of seismic modeling, this important topic is addressed by Modeste Irakarama, who presents a method to appraise various stochastic structural interpretations by reflection or borehole seismic data. In his paper, Modeste defines a number of conditions to look for misfit functions that yield a successful model appraisal. He also proposes a new way to map data misfit into model misfits (i.e., to localize where and why two models produce different seismograms). In geomechanics, this topic is also addressed by Antoine Mazuyer who shows his latest progress to estimate the in-situ stress state in the subsurface consistently with mechanical heterogeneities and local observations along boreholes.

In addition to the above-mentioned presentations, we are delighted to have many external contributions of colleagues on the theory and the practice of integrative numerical geology. There is no doubt that these external contributions will contribute to a vibrant and memorable 2017 RING Meeting.

We would also like to deeply acknowledge all the sponsors of the RING team which make our research and the organization of this event possible not only through funding but also through collaborations. We wish you an excellent meeting!

Guillaume Caumon        
Pauline Collon  
Paul Cupillard.