Stratigraphic layering and uncertainties , how knowledge can be inferred from the simulation of depositional processes ?

Daichi Yoon and Jonathan Edwards and Florent Lallier. ( 2017 )
in: 2017 Ring Meeting, ASGA

Abstract

Stratigraphic well correlation and associated stratigraphic layering define the lateral continuity of sedimentological and petrophysical information measured along well bores. Over the past few years, uncertainty on stratigraphic correlation has become an increasing concern in the subsurface characterization domain. Uncertainties are handled either by developing multi-scenario approaches or by using stochastic stratigraphic correlation algorithms. We review and discuss the latest advances in automatic and stochastic stratigraphic well correlation methodologies. The first concern is the stratigraphic correlation rules that are implemented in these algorithms. Multi-scale and multi-component heuristics used by geologists to build stratigraphic correlation models cannot be integrated in numerical methods. Instead of mimicking geologist's thinking, the latest developed methods propose to infer stratigraphic correlation rules from stratigraphic forward models (SFMs) used as a training models. By doing so, we assume that a stratigraphic correlation is likely to be valid if the physics implemented in SFM are able to produce it. However, the validity of SFM derived training models can be questioned. The second concern is the development of multiple sequences alignment algorithms. Building a stratigraphic well correlation model requires to minimize the objective function that implements the correlation rules, i.e. to find the optimal alignment between the multiple sequences that are the wells to be correlated. To date, due to the complexity of the numerical problem, this optimal alignment can only be built for few wells. Different strategies have been developed to tackle this issue. Among them, a sequential and iterative multi-well correlation has been proposed. However, its validity hasn't been demonstrated either mathematically or experimentally. We discuss the validity of this approach through its application to various test cases and investigate the potential of alternative strategy that has been proposed in the bioinformatics field for the simultaneous alignment of multiple amino acid sequences.

Download / Links

BibTeX Reference

@INPROCEEDINGS{Yoon2017,
    author = { Yoon, Daichi and Edwards, Jonathan and Lallier, Florent },
     title = { Stratigraphic layering and uncertainties , how knowledge can be inferred from the simulation of depositional processes ? },
 booktitle = { 2017 Ring Meeting },
      year = { 2017 },
 publisher = { ASGA },
  abstract = { Stratigraphic well correlation and associated stratigraphic layering define the lateral continuity of sedimentological and petrophysical information measured along well bores. Over the past few years, uncertainty on stratigraphic correlation has become an increasing concern in the subsurface characterization domain. Uncertainties are handled either by developing multi-scenario approaches or by using stochastic stratigraphic correlation algorithms. We review and discuss the latest advances in automatic and stochastic stratigraphic well correlation methodologies. The first concern is the stratigraphic correlation rules that are implemented in these algorithms. Multi-scale and multi-component heuristics used by geologists to build stratigraphic correlation models cannot be integrated in numerical methods. Instead of mimicking geologist's thinking, the latest developed methods propose to infer stratigraphic correlation rules from stratigraphic forward models (SFMs) used as a training models. By doing so, we assume that a stratigraphic correlation is likely to be valid if the physics implemented in SFM are able to produce it. However, the validity of SFM derived training models can be questioned. The second concern is the development of multiple sequences alignment algorithms. Building a stratigraphic well correlation model requires to minimize the objective function that implements the correlation rules, i.e. to find the optimal alignment between the multiple sequences that are the wells to be correlated. To date, due to the complexity of the numerical problem, this optimal alignment can only be built for few wells. Different strategies have been developed to tackle this issue. Among them, a sequential and iterative multi-well correlation has been proposed. However, its validity hasn't been demonstrated either mathematically or experimentally. We discuss the validity of this approach through its application to various test cases and investigate the potential of alternative strategy that has been proposed in the bioinformatics field for the simultaneous alignment of multiple amino acid sequences. }
}