Simmons critique-

Simmmons technical editors killed him.

The whole bubble point/reservoir pressure dew point think- he couldn't have missed the mark any further.

The critique is a bunch of drivel that doesn't get any closer to the actual problem- which is 4 MMBOPD at Ain Dar/Shedgum and Uthmaniyah resting on the precipice.

Thanks for the clarification.

Would you begin to have doubts if, when fed this information, it actually predicts a URR and a % Qt?

How is that? I observe an asymptotic function that never intersects the x axis.

I do have strong reservations about the information content in a period of flat production, though.

Increasing QT.

The whole point of the model is that production does not increase every year it peaks. So HL has some predictive power pre-peak certainly open to debate. Next post peak it seems pretty good until you get into the long tail where production is constrained by the nature of the recovery method. Water handling nitrogen handling steam etc. Enough oil is left in the ground from primary and secondary extraction that given enough money you can produce the fields for a long time at low production rates and EROI using advanced methods.

I've said a few times you have to bracket the HL using other information if you don't accept this then you can prove HL is wrong all day long. Since it is a empirical model the connections between HL and the real geology are complex at best.

And finally if you read my post about stat thermo all I'm saying is that the real geology can be treated in the same manner as stat thermo treated using statistical methods to derive higher order equations that describe the bulk this can be extended to groups of fields etc. It would be nice to have a solution developed from first principals but for now the logistic curve is good enough. We know from other processes that the bulk behavior is statistical.

To use another example consider a pile of sand the forces that effect a pile of sand vs a bucket of water are quite different and operate on different scales. But they both average to overall fluid equations for flow. HL is simply a proxy for these unknown fluid equations which are themselves only valid withing a given range.

Now with all that said if the bracketing of HL is simply to contrived then I'd say drop it. If a reasonable set of methods can be used to bracket HL then its useful.

I don't know what your going to write by my interest is in HL applied within say +/- 20% of peak. I could care less about its behavior outside this range. On the front side the data may be too noisy on the backside you have obviously peaked so you don't need HL nor does it apply as production moves to technically constrained.

The trick and the part your rejecting is that you have to have other information to determine if you think your within 20% or so of the peak. And even after this you need additional information to narrow down the actual peak year.

Now if someone comes up with the equivalent of stat thermo for the problem and can derive the real equations from first principals I'd drop HL in a heartbeat but I suspect they won't be all that different. Without this HL is the only method we have does not result in parameter explosion and bias.

I've said several times that most people modeling peak oil get sucked into the long tail region where a lot of oil will be produced at low flow rates. Just about everyone that starts doubting HL seems to focus on this region. We do have a lot more oil to produce and advanced methods may increase URR but its not going to change the fact that post peak we will consistently have demand exceed supply and thats all that matters. Once the economics of oil change the party is over.
So I could care less about 75% of the oil left to produce its the first 25% or even 10% post peak thats useful the rest is irrelevant since the geopolitical/social issues will determine if this oil is produced and what it is used for. Basically once we have a supply/demand wedge develop greater than say 4mbpd forget about the models since above ground factors will control oil production.

BTW and I think you know this in chemistry we have the theoretical plate model to normalize chromatography.

http://teaching.shu.ac.uk/hwb/chemistry/tutorials/chrom/chrom1.htm

A similar sort of model should hold for oil extraction.

Not sure about reversed :) Historically we had thermo first then figured out why it worked much later.

The right statistical model for Oil fields needs to include or derive from the field geometry factors and should include at least some way to model changing well technology. HL does neither actually thats why its really a empirical model semi derived from real data. My point is that I don't see how HL is all that different from the "real" statistical model done from first principals since using real production data tends to take into account these factors. If and this is a big if you feel you have interpreted the data correctly. I'd hesitate to make any real connection between HL and physical production.

The macro relation of produced OIP in place and production rate could be fitted to any Gaussian like distribution so the correlation is weak. We could easily find another macro model that works better than HL but I don't think considering the noise in the data it makes a big difference.

How to say this ...
From first principles we know something like HL will result if we modeled the system correctly and the data has a lot of error so why not just use HL ?

My point of contention if any with Robert is that such a model does and must exist I'm not overly concerned that we have not developed the exact model since it does not change the ramifications of peak oil and the bottom up methods tend to agree within the error range with HL. Picking the exact year of peak is simply and art if we are right then fine if not no biggie we can't confirm world peak until we know the world decline rate is hight enough that we will never bring projects online fast enough to recover. This is at the minimum two years after peak. I've said a few times that even then it will be controversial but its important to build a strong case that peak is behind us. HL may or may not be useful for this case. Post peak its seems pretty good so I'd say its useful.

To repeat myself its the decline rates that come out of HL that I like foremost and the URR estimate of what I call easy oil second which is the oil recovered from beginning production till say 20% post peak or the stuff we can get at high production rates.

Not overall URR which I have zero interest in KSA may be producing 1mpd in 100 years for all I know I don't see how that effects us today.

Yes Guassian would probably work also. Several different curves have been used. In general we don't have a strong reason to choose one method over the other. In this case the data has enough error that the detailed shape of the curve is probably not all that important. If someone comes up with good reasons for using another model and it is in agreement with the data from known regions that are in decline switching models is probably the easiest thing we can do :)

The point I've come down to is pretty simple regardless of the shape of the curve the date of the peak and URR are pretty much constants. How you produce the fields and other factors are simply changing the shape of the curve but the total URR i.e the area under the curve and the position of the peak are fixed.

A simple triangle might even give good results.

I might add that if you think about it picking the peak year is not that onerous its basically picking the peak date within 1-2% if the field is produced for 50-70 years.

Now picking the peak month ....

Damn near made more sense to me :)
Actually from my chemistry degree we do similar stuff to perform separations. A oil field has a lot in common with HPLC. High Pressure Liquid Chromatography. But we actually want breakthrough while the goal with oil is to not let the system separate or fractionate. It seems everything I was taught to try to do in chemistry is wrong for oil.

As a chemist I'd probably de-pressurize the reservoir let the gas bubble out first then extract then oil with a strong water drive causing the oil components to fractionate. Completely and utterly wrong :)

That's what I thought he said. :-(

kaufst du diese Schätzung... 105?

Entschuldigen Sie mich. Tief ist nicht genug.