Industrial Applications of High-Performance Computing

301

Chapter 22

Geophysical Imaging

in High-Performance

Computing Platforms

in Spain

Mauricio Hanzich, Josep delaPuente,and

FranciscoOrtigosa

22.1 COMPANY’S TRADITION FOR INNOVATION

Repsol, a world top 50 energy company, has a long history of research

and innovation. e Repsol Technology Centre (CTR) is a major R&D

center in specialties such as petrochemistry and energy eciency. Projects

such as Kaleidoscope (exploration geophysics), Sherlock (reservoir

characterization), or Repsol’s prominent role in providing racing fuels

and lubricants for World-Championship winning motorcycles have been

CONTENTS

22.1 Company’s Tradition for Innovation 301

22.1.1 Company and High-Performance Computing 302

22.1.2 Interaction with the National High-Performance

Computing Ecosystem 302

22.1.3 Case Study: 3D Full Waveform Inversion with BSIT 303

22.1.3.1 Critical Role of the HPC Community 303

22.1.3.2 Overview of Techniques Used 304

22.2 What’s Next and Conclusion 305

References 305

302 ◾ Industrial Applications of High-Performance Computing

renowned internationally. More than 400 researchers work at the CTR to

improve product performance and environmental impact, thus “redening

the world energy model by applying new technologies that help with

inventing the future.”

22.1.1 Company and High-Performance Computing

A milestone for Repsol has been the Kaleidoscope Project.* e project

started in 2007, uniting cutting-edge science and technology with the

goal of boosting the nding of gas and petroleum deposits while reducing

its environmental footprint. is has been possible thanks to exploiting

the latest generations of computer chips together with state-of-the-art

mathematical algorithms, which create accurate seismic images of the

subsoil. ese images enable us to “see” the earth’s interior and search for

hydrocarbons without drilling through the earth’s crust.

is proprietary Repsol technology can process seismic data 15 times

faster than other companies in the sector and greatly improves the

chance of nding oil and gas thousands of meters below the subsoil. is

technology has positioned Repsol at the forefront in the eld of exploration

of complex areas containing large undiscovered reservoirs, such as the

Gulf of Mexico or the Brazilian shelf, where an estimated more than 100

billion barrels of oil lie undiscovered. e Kaleidoscope Project has since

been named one of the ve most innovative projects in the world by the

American Institute of Electrical and Electronics Engineers. e Computer

World magazine also awarded Repsol for this project, which has been a

nalist in the Annual Creativity in Electronics Prizes and the Innovation

Prize awarded by Petroleum Economist magazine.

22.1.2 Interaction with the National High-Performance

Computing Ecosystem

e key to the success of the Kaleidoscope Project has been relying on

some of the best geophysicists, mathematicians, geologists, and engineers

from both Spain and the United States. is included researchers from

Repsol as well as 3DGeo, a top geophysical service company, and the

Barcelona Supercomputing Center (BSC), which operates Marenostrum,

one of Europe’s most powerful supercomputers. BSC provided hardware

as well as its best computing engineers and scientists for the task.

In 2010, and following the success of the Kaleidoscope Project, BSC and

Repsol created a joint research center: the Repsol BSC Research Center

*

http://www.bsc.es/projects/kaleidoskope_tmp/index.html.

Geophysical Imaging in HPC Platforms in Spain ◾ 303

(RBRC). e aim of the center is to tackle geophysical problems, which was

the goal of the previous project, and a broad spectrum of high-performance

computing (HPC) problems of interest for Repsol. e geophysical and

computational developments at the RBRC have resulted in a unique soware

platform called Barcelona Seismic Imaging Tools (BSIT),* a whole suite of

imaging applications, which includes state-of-the-art solutions for the most

challenging problems in exploration geophysics today, which is ready to be

used in the latest massively parallel computing infrastructures.

22.1.3 Case Study: 3D Full Waveform Inversion with BSIT

e most advanced geophysical imaging tool in production by the oil

industry is Reverse-Time Migration (RTM), which delineates geological

structures of the subsurface at a great computing cost (Araya-Polo et al.

2010). RTM was the target of the Kaleidoscope project, and its ecient

implementation in powerful computing platforms is the key to its success

(Figure 22.1). Full Waveform Inversion (FWI), on top of delineating

structures, gives quantitative insight on the properties of the rocks in the

subsurface, thus adding a whole new dimension to the insight that can be

extracted from the same data (Kormann et al. 2013).

22.1.3.1 Critical Role of the HPC Community

Already known since the 1980s, 3D FWI has not been adopted by the

geophysical exploration industry due to its enormous computing cost.

Set as an unconstrained inversion of a Terabyte-scale dataset containing

*

http://www.bsc.es/bsit.

FIGURE 22.1 Target model (top) and Inverted model (bottom) for a synthetic

industrial use case.