Quote from: RE on Feb 09, 2026, 06:25 PM

Quote from: TDoS on Feb 09, 2026, 04:26 PMName anyone else who solved all future peak oils.

Nobody has "solved" Peak Oil, anymore than they have solved the problems of the Atmospheric CO₂, the Federal Deficit, Palestinian Genocide, Falling Birthrates or Donald Trump.

RE

You are quite incorrect. There were 3 up and running systems that I am aware one, one commercial, one available for a price out of Palo Alto, and the one I was asked to build. You can buy results from the first two if you know who has them, but not mine. These can't be the only 3, but are just the ones I have access to, have built, or been asked to stress test before commercial sales.

The first commerical system was deterministic in nature, working from central tendency inputs, delivering scenario based answers...."if I assume this price path, what is production potential, and for what length of time". Completely manual, price path of hydrocarbons being the only input variable. Maybe you could control the probability estimate in countries that supplied them under the PRMS system.

The second is an LP, working with proprietary economic assumptions and some really poor resource assumptions, simplistic resource cost curves, deterministic in terms of volumes. Lacking any decent engineering component, building stochastic outputs primarily based on variable price paths. Primrily econometric in nature, not enough detail on the resource side.

Mine is stochastic, based on the three main moving dimensions to the peak oil problem. Lets call them #1, #2 and #3. In order to solve it for any two of these dimensions, you assume a deterministic answer to the third. You iterate on the two until you have a 3 dimensional equilibrium. You change #1 slightly, and do it again. And again. And again. You develop an envelope of performance for the combination of a single changing variable. Then you make #2 dimension the changing variable, and iterate. And then #3. You then have a database of substantial size containing as many pertubations of the underlying data as you'd like...or are willing to wait for. It does take awhile.

The trick wasn't in the overall design, I use this technique for quite a few routines, the trick was in then assembling all this data into a matrix of outcomes for a given input that was possible to display to the user, and not screwing it up with cross correlations and weird dependencies. Once you ran it initially, you tracked down combinations that were ridiculous, and began to fine tune the working input ranges in order for a high price to intersect reserves and resources at a country level, not get too crazy over kerogen, stuff like that.

Call it teething problems. Took 6 months to figure out how to use it intelligently and display the statistical outcomes for any given input assumption. What does world oil look like with high prices, low resources, political instability in OPEC, wars, higher taxes in which countries or region, Venezuela comes off the board because no one wants heavy crude, how much oil can come from Russian LTO in the Bazhenov and under what geopolitical circumstances and resulting price consequences, and so on and so forth.

Took 2 years to design. Another year to build. Six months to debug. Cost mid 7 figures. After debugging and testing it with our own scenarios of whatifs, spewing out scenarions and liklihoods based on what WE thought were interesting scenarios, we built a control system on top of it. The user gave the control system a global volume path, and it would work backwards to figure out the price combinations necessary to figure out where the oil and gas comes from (it did 8 hydrocarbon products I believe). You put in a given price for all countries for all 8 products, and it would take everything it already knew and reverse calculate out all the volumes by country. And if you ran it bunches of times, you could develop a range of outcome for all products and prices based on given input ranges of prices or volumes.

Please. Didn't solve peak oil. And by the time it was solved, prices had crashed and nobody even cared anymore. I still run it on occasion, just for fun to see how it looks compared to what happened since its completion. Nearly all top level global run results did come up with an interesting answer, one in particular that tended to cross all scenarios, or all reasonable, non war, non political nonsense scenarios anyway. And these years later, we are approaching that point. If it pans out, I will be quite happy and consider the model validated.

Didn't even need the PhD Harvard mathematician involved, but he did pimp the idea to the right people to get the money.   

Makes the Limits to Growth models look like a Model T if I do say so myself.