Oil & Gas activities

Oil & Gas fuels the Irish economy by providing over 75% of total energy requirements (Energy in Ireland 2016 Report, SEAI). As Ireland's national High Performance Computing (HPC) centre with a decade of Oil and Gas experience, ICHEC is uniquely positioned to provide services in consultancy, pre-commercial innovation activity as well as research partnerships with industry and academia.

ICHEC has links to the industry with a long lasting relationship with Tullow Oil plc. Academic services are through our close collaboration with colleagues in Lero and iCrag. The establishment of a dedicated Oil & Gas activity at ICHEC and lead by Ruairi Short has allowed us to bring new visibility and focus to our activities.

If you are interested in further information about our services in this domain or wish to propose new ventures, please contact us at consultancy@ichec.ie.

Introduction

ICHEC is currently providing consultancy services in this domain with dedicated projects from development to research and innovation. ICHEC has a team of computational scientists who are also skilled in modern collaborative software engineering practices with certification in agile development methodologies (Scrum). Leveraging our domain knowledge and software development experience, ICHEC develops software for, and in close collaboration with, industry partners.

Activities

There are currently 5 active topics within the activity, 2 of the topics are performed under consultancy contracts while 3 are performed through the ExSeisDat project:

Seismic Migration Consultancy

As part of the consultancy services we provide to Tullow Oil plc, we develop seismic migration applications with a particular emphasis on performance and scalability. This is a long term project to deliver production codes to the client. The first instalment was a reverse time migration code, TORTIA, which has since been the subject of a paper and a basis for further novel technology investigations. Kirchhoff migration work is ongoing and showing promising signs of outperforming competing solutions.

Machine Learning

Based on our experience in the Oil & Gas domain as well as our technical skills in machine learning, we are also performing a targeted report on the applicability of machine learning to seismic exploration workflows with a cross-functional team of machine learning specialists and geologists.

Extreme-scale Seismic Parallel I/O Library (ExSeisPIOL)

Through our seismic migration consultancy, we saw first hand the importance of replacing existing parallel I/O code distributed across various code bases with a single unified and optimised library called ExSeisPIOL which reduces redundant code development while providing comparable I/O performance without manual optimisations.

We presented the ExSeisPIOL and our work on the DDN IME at the Rice Oil & Gas HPC 2017 conference in Houston, Texas

Extreme-scale Seismic Workflows  (ExSeisFlow)

To overcome limitations within proprietary seismic applications requires that restrictions in the execution model be relaxed and the additional complexity managed. ExSeisFlow provides a flow control tool which is I/O focused and with multiple execution models supported. This allows support for operations which do not fit well into an embarrassingly parallel paradigm. Although this product does not aim to replace proprietary applications, it does allow for increased time spent out of application, thus freeing up valuable licence resources.

Seismic Codes on Next-Generation Hardware

ICHEC profiles seismic applications for DDN Storage, the largest independent supplier of data-storage solutions, on their next-generation Infinite Memory Engine (IME), a burst buffer and write-performance optimiser, that has been installed on world class systems. We have showcased our results at internal Oil & Gas conferences such as the Rice University Oil & Gas HPC conference 2017 in Houston, Texas. This work complements early stage profiling also performed by ICHEC on Oil & Gas codes.

ICHEC's experience with novel technology is in the domains of next-generation hardware and also in machine learning / big data.We have considerable experience with compute-accelerators such as NVIDIA GPUs (CUDA Research Centre), Intel Phi (Intel Parallel Computing Centre), Xilinx FGPAs (Xilinx Alliance Program member) and data-accelerators such as the DDN IME in direct collaboration with the technology vendor. This allows us to leverage expert knowledge to perform cutting edge performance optimisations within the Oil & Gas domain. In particular the ICHEC developed code TORTIA was ported to the Intel Xeon Phi showing performance improvements and was presented at an ISC'15 workshop.

Other Activities

ICHEC also has a number of activities we have performed in the past, including code optimisation for Saudi Aramco, one of the worlds largest upstream producers, where we identified and resolved a system-level bottleneck

Expertise