FOR GEO-IMAGING

Led by scientists at the Technical University of Delft

My Image
091-algorithm-1

Delphi's Closed-loop approach

Central to all research within Delphi is the closed-loop approach. At the small scale in e.g. Full Wavefield Migration, where the modeled data from the current reflectivity image is continuously compared to the measured data in order to improve the final image. But also at a larger scale, where the output of reservoir characterization feeds the design of new acquisition geometries or even acquisition methodologies.

Full Wavefield Technology

Our industry is in demand of imaging results beyond today’s resolution limits in order to progress in the next decades. We believe that exploiting all complex propagation effects in the seismic data is an important way to achieve this ambitious goal. This means the end of ‘linear’ imaging methods, based on primaries only, and that multiple reflections, transmission effects and wave conversion are the key components of retrieving detailed, reliable information on both complex subsurface structures and the reservoir area.

Currently, Full Waveform Inversion (FWI) concepts – based on directly solving the two-way wave equation – is very promising methodology, however also a highly nonlinear method, prone to local minima. Furthermore, the output of FWI – being the velocity model – is often used in traditional migration procedures, losing all the benefits of exploiting the full wavefield.

Therefore, in addition to such solutions, within Delphi we propose the Full Wavefield Migration (FWM) approach, which combines the best of imaging and inversion methods and exploits all higher-order scattering effects in one integrated framework.
Extending this concept, Joint Migration Inversion (JMI) also includes the estimation of the background velocity model. Because our inversion parameters – being propagation velocity and reflectivity – connect more directly to the observed data (being arrival time and amplitude), we aim at reducing the non-linearity of the inversion problem.

Furthermore, FWM and JMI are very much suited for application in a time-lapse fashion, where all vintages of seismic data are inverted simultaneously in order to obtain detailed, dynamic properties of the reservoir, taking into account geologic scenario information.
Finally, the FWM and JMI framework allows to provide high-resolution input to reservoir characterization and beyond, as all information on the elastic reflection properties are fully maintained.

Delphi's Background

Encouraged by the success of John Claerbout's consortium at Stanford in the seventies, professor Berkhout decided in the early eighties to set up a seismic consortium at the Delft University of Technology (TU Delft). Particularly with the help of one of his students, Paul van Riel (co-founder of Jason Geosystems), he started in 1982 the so-called PRINCEPS-consortium. The objective was estimation of acoustic impedance from seismic data by constrained trace inversion. PRINCEPS started with 5 companies.

From the research in the PRINCEPS-consortium it became clear that the extraction of in-situ rock information beyond acoustic impedance would require analysis of pre-stack seismic data, preferably after removal of the down- and upward propagation effects. Therefore, it was decided to set up a second consortium in Delft that was aiming at distortion-free, angle-dependent input for PRINCEPS. Particularly with the help of Kees Wapenaar (who had just successfully completed his Ph.D. thesis), professor Berkhout founded in 1987 the so-called TRITON-consortium. The objective was target-oriented pre-stack migration, using data-driven multiple removal and true-amplitude redatuming as preprocessing steps. TRITON started with 13 companies.
From the inversion research in PRINCEPS and the migration research in TRITON it became readily clear that both consortia would significantly benefit from a closer interaction. In 1989 it was decided to merge PRINCEPS and TRITON into one consortium: Delphi. The objective of Delphi was an integrated approach to multiple removal, velocty-independent migration and reservoir characterization. Delphi started with 21 companies.

From the integrated research in Delphi it emerged that the success of seismic imaging is largely determined by the way data acquisition is carried out. This particularly applies to the geometrical distribution of the sources and the detectors. Therefore, it was decided to start a third initiative aiming for an acquisition consortium that would investigate the influence of source and detector geometries on the quality of imaging and characterization results. Particularly with the help of Dr. Leo Ongkiehong (a former colleague of professor Berkhout in Shell), the so-called DOLPHIN-consortium was founded in 1995. The objective was to introduce a new approach to the analysis of acquisition geometries by making use of the focal beam concept. An important tool in DOLPHIN was, and still is, the downward- scaled acquisition system that has been built with the group's knowledge on medical imaging instrumentation. DOLPHIN started with 6 companies.

With the help of Dr. Gerrit Blacquière and Dr. Eric Verschuur (who both had successfully completed their Ph.D. thesis), DOLPHIN was fully integrated into the Delphi program as the Acquisition & Preprocessing (A&P) Project. In 2003, Delphi was further strengthened by adding the Reservoir Characterization & Monitoring (C&M) Project in order to make a better connection to the geologists and reservoir engineers.

Copyright 2019 - Delphi Consortium I Contact us I Disclaimer I Login