Monday 28 August 2017

A sort of AI Quiz!

Whydriverless cars might not hit the road so fast” is an article on LinkedIn by Scott Nyquist who is a Senior Partner at McKinsey & Company: we are both on the Advisory Board at Kimmeridge Energy LLC.

 

I guess you can only access it if you are registered on LinkedIn but basically he is predicting that the growth in autonomous vehicles (and electric cars too) might not go as fast as everybody is currently prognosing.


It reminded me of when I pitched up in Houston from Denver a couple of months ago. My flight was 3 hours late and by the time I got to the car rental car park there was one car left which thankfully turned out to be my Jeep automatic. But suppose the guy had said "Sir, there's just one car left and it's this autonomous vehicle!" Through my mind would have flashed – “find the Hardy Toll Road, do 610 West and then 610 South with an 18-wheeler either side, off at Woodway, sharp right, sharp left...."

Would I have said "That's great, thanks!" or "You know what, I'll just take the shuttle bus back to the Terminal and take a cab!"?

Straw poll of acquaintances is 100% the latter option.....


So here is the Quiz question: what would you have done???

Using AI to improve E&P performance

You may have noticed this news item a few weeks or so ago, talking about BP’s plans to introduce AI to seek drilling performance improvement.

It seems these plans are part of an effort to embrace new technologies, which Bob Dudley, BP’s chief executive, said last week were “rewriting the rule book for E&P”.

This set me thinking (again) about where are the E&P problems – the opportunities for performance improvement that could have such an effect?

I had a few ideas and here are some of them – not in order of importance but roughly in the order in which they occur in the value chain:

Using satellite- and air – borne sensors: there has been an explosion both in the number of vehicles flying around above us and in the bandwidth and resolution of the sensors they carry, Can we analyse these potentially huge data sets to reveal geological variation, map micro-seepage etc?  Make onshore exploration more successful……

Interpreting 3D and 4D seismic: huge data volumes - essentially we use such seismic to “do geology”. Geology is a rules-driven science; can a machine learn from previous interpretations and outcomes how to do this much better than humans?

Drilling: the point of the article about BP. Can a machine learn from the thousands of wells that have been drilled (in any one basin onshore USA; North Sea etc) how to optimise drilling performance? A useful performance metric is days/10k ft; is a focus on reducing Non Productive Time(NPT) the only key to performance improvement or are there others, for example the real-time adjustments that the BP article mentions?

Completions: again, from the thousands of wells drilled in any one basin or region, can a machine learn how to optimise/maximise IP (Initial Production)?

Preventative Maintenance: can we do as the airlines seem to do routinely and learn to predict equipment issues and failures before they actually happen? In turbines and compressors for example: I find it only slightly ironic that GE who sell such things are at the forefront of trying to sell software to tell us when they will break……

Actually, as a pointy-toed geoscientist I am underselling the impact that analytics could have on production operations, in fact on the whole of the engineering that that takes place between the well head and the storage facility. For a much better account, I refer you to Karl Jeffrey’s summary of a recent Finding Petroleum event on Transforming offshore operations – with better use ofdata” which you can access by clicking on that title. There are so many things that could be done to improve efficiency, cut costs, prevent incidents, improve operational integrity, improve safety…….


Monday 21 August 2017

"Winners" and "Losers" with a LFE* oil price

When discussing future oil prices, some CEOs seem to have moved on from ‘Lower for Even Longer’ to ‘Lower into the 2020’s’ to ‘Lower For Ever (*LFE)’.

Compared with the late 1990’s, this period of low oil prices has not yet seen a huge amount of consolidation – debt re-structuring, yes – but not outright acquisitions.

Exploration Performance in the 1990’s as a cause of ultimate demise!

It is said that banks and funds invest at least as much in the management teams of small exploration companies as in the actual prospects they are shown.

And therefore it is still common to read, in a company’s prospectus, that “X and Y were part of Z Oil’s widely respected and highly successful exploration team” or similar.

It is worth recalling that exploration is in general a game in which there are a few “winners” and very many “losers” and that in fact those that “lost” in exploration in the 1990’s were generally on the receiving end of consolidation at the end of the decade; this has been thoroughly documented by consultants such as ADL, Wood Mackenzie and others.

An example of these real differences in performance is shown in Figure 1 (a ‘classic’ exploration performance benchmarking display, showing reserves replacement as a percentage on the vertical scale and Finding Cost in $/boe on the horizontal).

Figure 1

What happened to Arco, Amoco, Mobil, Phillips, Texaco and Enterprise? Burlington, Kerr McGee, Unocal, Norsk Hydro? Gone!

What is my point?
It is that it is performance that ultimately decides the destiny of companies, not conspiracies - although I hear some Discussion Boards (e.g. the one on Premier Oil) believe something different!

And that over time, companies (have to) reveal enough about their actual, as opposed to promised, performance for investors to be able to see the truth. And for Executives to figure out their position relative to their competitors – and, if necessary, do something about it before it is too late.