International Inversion Initiative (UBC)
I3 — _I_nternational _I_nversion _I_nitiative SINBAD This project is a joint venture between the University of British Columbia (UBC), Imperial College London, the Universade do Rio Grande do Norte (UFRN), and BG group to develop and industrialize Full-Waveform Inversion technology with funding from the 1% R&D levy in Brazil.
Summary
We propose a major new initiative, designed to leverage opportunities for funding in Brazil, in order to advance full-waveform inversion into a robust transformative technology that is able to address problems across the entire spectrum of exploration and production.
The SINBAD consortium, led by Felix J. Herrmann at UBC Vancouver, and the FULLWAVE consortium, led by Mike Warner at Imperial College London, are each working to advance the capabilities of 3D full-waveform inversion and related technologies. The approaches taken by these two groups are largely complementary. Encouraged initially by BG, we believe that the synergies to be generated by much closer collaboration between these two groups could be hugely productive.
Several of the consortia’s members are exposed to the 1% R & D levy in Brazil, and more are likely to become so over the next decade. The intention of this levy is to build new capability and capacity within Brazil, and companies that are levy contributors will seek, in addition, to derive commercial benefit from their contributions. Our proposal aims to satisfy both of those requirements.
FWI on 3D field data is computationally demanding. This limits the degree to which innovative ideas in this field can be tested rigorously against field data, and it constrains the volume and rigour of parameter and uncertainty testing that can be contemplated even by the largest operating and service companies. This problem magnifies as we move to large datasets, deeper targets, higher frequencies, elastic inversions, and better approximations, and it is especially acute when we seek to hybridise FWI with other demanding inversion schemes – least-squares RTM or WEMVA for example.
Predicated upon these three observations, we propose a new collaborative research initiative, funded through the levy, centred at the International Institute of Physics, at Universidade Federal do Rio Grande do Norte, in Natal, Brazil. The project will support a major computational facility in Brazil, through which SINBAD and a new team to be developed in Natal, will collaborate to move FWI to a new level while simultaneously building Brazilian research and practitioner expertise in this and allied fields.
We believe that such a project would produce huge benefits to Brazil, potentially making them a world leader in an important emerging technology, and would provide equivalent benefits to its international sponsors and collaborators which are difficult to realise through any other mechanism. We would like the opportunity to explain our ideas in more detail, and to work with our existing sponsors and with UFRN, to bring this concept to fruition.
If you are interested in learning more about this project please contact
, SINBAD, SLIM, UBC, Canada
, FULLWAVE, Imperial College London, UK
SINBAD & FULLWAVE collaboration FAQ
Summary
With this FAQ, we hope to clarify some questions that may arise regarding the I3 project. This FAQ is not intended to be a legal contract or have any other force of law, but it does reflect our current best intentions, and we anticipate that it will form a basis for a constructive discussion between our various stakeholders. We are absolutely willing to listen to alternative views from all, and we look forward to hearing your feedback.
Frequently asked questions:
Why are the SINBAD and FULLWAVE consortia collaborating? The two research consortia are both working on full-waveform inversion and related technologies. The two groups are largely complementary in their expertise and approach, but their strategic aims are the same — to make FWI effective, efficient and robust for an increasingly wide range of targets, datasets, and acquisition systems. We believe that both groups will increase their productivity and their creativity through the rapid exchange of ideas, software, results and insights between them. And we believe that this will immediately and significantly benefit the sponsors and stakeholders of both groups.
Why is Brazil involved? While Brazil’s 1% levy is a critical catalyst for the collaboration between our groups and our Brazilian partners, giving us the opportunity to carry out academic research beyond usual constraints of finance and borders, the rationale for the core collaboration between our groups goes beyond this Brazilian opportunity. Our aim really is to bring FWI to a higher level by complementing our research efforts and by training the next generation of geoscientists. Undoubtedly the SINBAD-FULLWAVE collaboration will be enhanced and cemented by a successful joint enterprise within Brazil, but its benefits mean that we would seek to collaborate even without the Brazilian opportunity.
What does Brazil offer us? Petroleum operating companies with qualifying production in Brazil are under obligation to fund research and development to a value of 1% of their annual gross production. This is a large and increasing sum, and it provides a number of opportunities for research groups like ours that are based outside Brazil. It is able to provide direct funding and support costs to bring Brazilian nationals as PhD students and post-doctoral researchers to join our groups, it can fund international students and researchers from our groups to work and study within Brazil, it can provide a core research team within Brazil with which we can develop joint research programs, and significantly, it can provide a large high-performance computational resource based within Brazil that we can use as a research tool.
What do we offer Brazil? Advancements in computer architectures and performance combined with the joint development of new algorithms between FULLWAVE, SINBAD and our Brazilian partners, will make FWI a viable technology that has game-changing potential across the petroleum industry. Its benefits are only beginning to be realized and exploited. The strategic aim of our collaboration with Brazil is to develop world-leading expertise and capacity in FWI within Brazil — both in research and in production. Brazil is certain to be involved in petroleum production for the long term, and FWI shows every sign of evolving into a dominant technology over that term. Our aim is to play a significant role in helping Brazil to embrace, develop and exploit this technology to the benefit of the country and to the benefit of the companies that operate there — operators, oil-field service companies, hardware vendors, and others.
What does this offer companies with Brazilian production? Qualifying companies have to pay the 1% R&D levy in Brazil. At least half of that spending must occur within Brazilian universities and associated research labs, and many Brazilian universities are already heavily engaged within this program. Our FWI initiative aims to turn the commitment that operators have to spend, into an opportunity to generate value. To achieve that, we plan a world-leading research program, coupled to a unique research facility, leading to practical outcomes. The unique research facility is to be provided by an HPC system within Brazil that will be substantially dedicated to FWI R&D.
How does this benefit our existing sponsors? FWI is computationally expensive – particularly when it is applied to large and sophisticated field datasets. And it is evolving in new directions — multi-parameter, full-elastic, WEMVA, broad-band, uncertainty analysis — that challenge even the largest HPC systems. Major commercial HPC systems are necessarily largely dedicated to production and immediate problem solving rather than to strategic R&D. The pace of development within both SINBAD and FULLWAVE is significantly compromised by the limits to our computational facilities, and that is likely to be the case across almost all academic and commercial research groups active in this field. The initiative in Brazil will represent a huge uplift in our computational ability and efficiency. We will be able to test ideas on field data in hours or days that currently take us months. The pace of our research will increase, and some areas will become accessible to investigation that are currently beyond our computational reach. Our existing sponsors will benefit directly and immediately from this uplift.
How will this be organized? Imperial College London in the UK, the University of British Columbia in Canada, Universidade Federal do Rio Grande do Norte in Natal Brazil, the International Institute of Physics in Natal Brazil, and SENAI Cimatec in Bahia (the HPC-hosting Institution), are combining to form the International Inversion Initiative, I3. BG Group, who have significant production in Brazil, are facilitating the formation of this group, and the purchase and operation of a secure approximately 500 TFlop HPC system that will be based at SENAI Cimatec in Salvador Bahia. A new FWI research team will be built up at UFRN, and the IIP will facilitate national and international workshops and exchanges. The HPC system will only be used for R&D, not for production, but it can work on datasets from outside Brazil and from outside BG Group, and it will be accessible by university researchers from London and Vancouver. Once I3 is launched, and the HPC system is operational, it is our hope and expectation that other operators will wish to join I3
How will we handle IP? SINBAD, FULLWAVE and I3 will all generate IP. The direct sponsors of these three projects will have, as now, access to the IP generated within the projects that they fund. It will be a condition of membership of I3 that member companies also become and remain members of both SINBAD and FULLWAVE. Brazilian partners in I3 who are not operating companies will also need to become paying members of both consortia. If we operate like this, then IP generated within either SINBAD or FULLWAVE can flow to all participants within I3 on exactly the same conditions as currently apply to members of those consortia who are not members of I3. Existing members will benefit because the additional consortium fees will fund expanded research programs, and because the Brazilian HPC system will be available to the Imperial and UBC research teams with the results available to their sponsors.
What about IP generated by I3? If IP is generated by members of the SINBAD or FULLWAVE research teams, whether or not they use the Brazilian HPC system, then the IP generated will be available to SINBAD and FULLWAVE sponsors. This IP will also be available all members of I3 because all I3 members will belong to both consortia. IP generated by the team at UFRN, or by other personnel funded directly by I3 inside Brazil, will not automatically flow to SINBAD and FULLWAVE members, but will flow to all I3 members who may request a reasonable period of exclusivity to the results in so far as that is consistent with the rules that govern the Levy. Where there is a clear intellectual contribution from FULLWAVE, SINBAD or I3 Imperial College London, UBC and UFRN will own a share of the IP to which they have contributed.
How can I protect my proprietary data? The HPC system will operate to the same levels of security that pertain within commercial seismic systems. Data can only leave the HPC system through a defined and controlled route and process. Proprietary data within each of the consortia cannot be released to members of the other consortia, or used by researchers in the other consortia, or results shared across the consortia, without the express permission of the original data owner.
What about IP generated using proprietary data? Data-specific results generated in any of the three consortia that involve proprietary datasets will not flow to members of other consortia, however they were generated, without the explicit permission of the owners of that proprietary data. Consequently, for example, a member of I3 can provide a dataset to that consortium, have that dataset worked upon intensively on the HPC system, and receive back the results of that research, secure that these results may not be released to others without their explicit permission. Resulting generic algorithmic developments however would normally be shared.
What about flow of IP between SINBAD and FULLWAVE? It is our hope that all sponsors will wish to belong to both consortia, however we recognize that this may not come to pass. The two groups will exchange software between them internally, and will cross license to each other all associated IP. Sponsors of either consortium will have access to any results generated by this cross licensed IP, but they will not have access to the source code, software or other licenses unless they are also members of the second consortium. Members of only one consortium therefore obtain partial benefits from the work of the other consortium, and in exchange for this, they allow members of the other consortium to benefit reciprocally. Except where there are patents in hand, genuine commercial software products, or other special commercial circumstances, the IP generated within one consortium will normally be licensed to members of the other consortium after a delay of three years. That is, members get three-years exclusivity to the results of the consortia to which they belong.
What about service companies? Service companies belong to both SINBAD and FULLWAVE. They are not however liable to the 1% levy. We would welcome the participation of one or more service companies in I3 and we are open to suggestions as to how that might be sensibly achieved. Service companies can undertake R&D that can be paid for under the levy, and it is possible that we can make use of that route to involve service companies to the benefit of all parties. More specifically, contractors (and others) involved in Brazil but who do not have obligations under the one-percent Levy can have the opportunity to join I3 with ‘in-kind’ contributions towards
support for the research group in Brazil with practical aspects of field data acquisition, design of industrial processing flows and other practical issues related to wave-equation based inversion;
field data acquisition and the design of new (e.g. ocean bottom nodes) field data acquisition schemes in collaboration with the industrial and academic partners of I3;
cash contributions to meet the (indirect) costs of I3 research incurred by the SINBAD Consortia. This will directly benefit the current members of the SINBAD/FULLWAVE Consortia;
joint research projects, processing as part of case studies, software donations, and software development. This benefits all sponsors.
In return, contractor companies may seek to negotiate a period of exclusivity over their competitors that is commensurate with their level of contribution to the project.
Will the Brazilian effort distract, dilute or otherwise interfere with SINBAD and FULLWAVE? I3 is an unique opportunity to bring rapid advancements in FWI research by bringing more resources and more people to both consortia, it facilitates compute activities that currently we are unable to carry out, and it speeds up dramatically the real-time turn-around of FWI applied to 3D field datasets. Currently we have far more ideas, untested hypotheses, untried algorithms and ambition that we can undertake using only our consortium funding and our limited local HPC facilities. We will go faster and do better once I3 is launched than hitherto, and all stakeholders will benefit from that outcome.
Will SINBAD and FULLWAVE concentrate upon Brazilian datasets? Absolutely not. The I3 consortium may chose to do that, steered by its members interests, but the SINBAD and FULLWAVE consortia will continue to work on whichever datasets and problems appear most likely to advance FWI generically, and as encouraged and facilitated to do so by their members. Most companies that are involved in Brazil have worldwide production, and many of the problems in Brazil have worldwide analogues and will benefit from generic solutions of wide applicability.
I am in FULLWAVE and SINBAD already — what would I get by joining I3 as well? Foremost, you get the opportunity to spend the 1% levy in an area that is clearly important to you. You get the opportunity to put your datasets through the HPC system in Brazil. You get the opportunity to set the strategic agenda within I3. You get early access to those results and IP that are generated entirely within I3. You get early access to the Brazilian PhD students, researchers and others working on FWI, and these are people that you will often hope to employ subsequently in Brazil. More strategically, you contribute to advancing FWI globally and in making Brazil a centre of excellence in this field. You have Brazilian production, and you have data that is acquired, processed, imaged, and interpreted in Brazil. You want the smartest and most experienced people involved in that work, and this helps to bring that about.
What will be the governance structure of I3? The partners in I3 have yet to establish this in detail, however the principles that will be adopted are clear. Commercial sponsors of I3 Brazilian academic partners in I3 Imperial College and UBC will all be represented on I3’s steering committee. To guarantee representation and transparency for SINBAD and FULLWAVE members, the steering committee will invite observes from SINBAD and FULLWAVE representing consortium members that are not direct sponsors of I3. The intention will be to make the governance, actions and decisions of I3 as transparent as possible to all stakeholders and interested parties.
OK, I am convinced — how do I join up? BG Group expects to be the founding sponsor of I3. Their contributions to I3 will set up the initial HPC system, launch the Brazilian research team, and facilitate long-term collaboration with SINBAD and FULLWAVE. Their investment under the levy will be significant. They will consider welcoming other sponsors as co-founders or as late participants. The expectation is that the HPC system will expand as new sponsors join I3 and that the research program both inside and outside Brazil will also expand in volume and ambition. There is also recognition that different sponsors may have different levels of liability under the levy, and that their contributions to I3 may in some way reflect this. Although there will be no formal minimum level of contribution to I3 contributions that do not represent a substantial commitment to the project, having regard to the level of activity in Brazil, are unlikely to be accepted.
OK, I am convinced but I cannot yet join because we are not yet liable under the 1% levy. Late-joining sponsors in Brazil, who are now or who become liable in the future to the 1% percent levy, can join I3 at any stage subject to the agreement of existing members and stakeholders. Negotiations to do this will be on an individual basis having regard to the individual circumstances of each potential member. Companies that become liable for the levy in the future might reasonably expect to be able to join I3 as late participants without incurring a substantial late penalty, whereas those that seek to join I3 as late participants while having been already liable to the levy for some extended period might reasonably expect to incur some additional costs resulting from their delayed membership.
Where can I get more information? Contact:
, UBC
, Imperial