IC Traditionally, a geological model is built in the early days of a field’s life. After a few years, these models are often forgotten, as operational focus shifts to individual well production and test data. This move to quick-look, point analysis means that folks lose contact with the big subsurface picture and the fact that all surface issues originate in the subsurface.
Reservoir modeling is well-established field. What can there be left to discover?
Previously the reservoir could be modeled as a tank or a box, which makes for fast but overly-simplistic calculations. On the other hand, a million cell numerical model may take too long to produce a result. A full scale reservoir simulation exercise may be done once every six months or once a year. In fact, most reservoirs are never simulated!
We hear a lot today about data-driven models. Is your approach data-driven or forward modeling?
We are definitely in the physics-based, forward modeling camp. Paradigm/k is not proxy-model based.
Ok but again, this is a well-studied field, where does the breakthrough come from?
From the guy who wrote the book, Michael Thambynayagam! His 2,300 page ‘Diffusion Handbook*’ is painstaking compilation of the math behind fluid flow in porous media. Our system embeds his work, figures out the flow regime and applies the appropriate equations. Paradigm/k is 100 times faster than current systems and no special knowledge is required of the user. There are no complex decks to build before running a model.
This is a major leap for Paradigm which has to date been more in the geoscience arena. Did you do a deal with Mr. Thambynayagam?
Yes, we have an exclusive arrangement with him and his colleagues to use and develop their techniques.
Paradigm has been closely involved with standards in the past particularly ResqML. Will this and maybe ProdML be part of the new solution?
Sure we can read these formats and also grab data from other vendors’ tools. But are true focus is production optimization, nodal analysis and applications such as artificial lift. Another great facet of Paradigm/k is its collaboration platform, a kind of Facebook for wells. The collaboration tool lets users work in channels. Wells too can have a ‘persona’ and contribute to the ‘discussion.’
Does Paradigm/k talk to scada/DCS systems directly or does it operate downstream of the historian?
It is more downstream of the historian. Actually, Paradigm has its own historian but we can also read data from others such as OSIsoft’s PI system.
We have noticed that ‘shale’ is often an assortment of stringers and shales. How can you model such complexity?
We picked unconventionals as a first target because this complexity is solvable with our approach. Paradigm/k is robust in the face of such complexity. We also layer on top of these models a capability to model complex completions like branched and oriented fractures. Next year there will be more conventional.
Optimization is often divided up into fast, medium and slow loops. Where is k?
We are upending this approach. k makes it possible to do the slow loop every day!
Does this mean conventional geological models (like Skua) are left on the shelf?
No. It means that you can feed the results of k into a complex model like Skua, linking production back into the model. This is an exciting foray for Paradigm as we extend our pioneering, science-based approach into the production arena.
How do you propose to take this forward, are you planning a consortium?
We have shown it in beta to several clients and there is a lot of interest. We are open to how this may progress. We will be working with our initial clients although there may not be a formal consortium.
~
* Thambynayagam’s oeuvre won the 2011 Prose/RR Hawkins Award and he featured in a short film, ‘A Holy Curiosity: The ‘poignant story of one man’s quest to create an epic work of scholarship.’ The film is available on YouTube and includes enthusiastic endorsement of Thambynayagam’s work from Schlumberger’s Michael Prange. At the time of writing the video has had a measly 21 views. Maybe Oil IT readers can move the dial.
As Samhita Shah points out in this issue’s ‘Letters,’ we mangled Thambynayagam’s previous job title in our last issue. He was MD of Schlumberger Cambridge Research, not CTO.
© Oil IT Journal - all rights reserved.