tag:blogger.com,1999:blog-5822805028291837738.post3574782575538852089..comments2020-01-16T11:19:18.557-05:00Comments on Various Consequences: Uncertain Rate in FFT-based Oil Extraction ModelJoshua Stultshttp://www.blogger.com/profile/03506970399027046387noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-5822805028291837738.post-21981827253459245792010-03-06T20:30:40.136-05:002010-03-06T20:30:40.136-05:00I think using fatter tail distributions would like...I think using fatter tail distributions would likely slow down the convergence of the stochastic collocation method for sure, since the 'edges' are the last thing to be added as the order increases.jstultshttps://www.blogger.com/profile/03506970399027046387noreply@blogger.comtag:blogger.com,1999:blog-5822805028291837738.post-19418672256891103622010-03-06T18:15:14.126-05:002010-03-06T18:15:14.126-05:00Good eye on the 0.9 confidence level result.
The ...Good eye on the 0.9 confidence level result.<br /><br />The main uncertainties on the extraction model seem to be (1) statistical variation on the sizes of input discoveries -- as shown in the MC sim (2) uncertainties in the interpreted size of the discoveries as the field matures -- which is the reserve growth problem and (3) the uncertainty in the actual rate.<br /><br />The rate of extraction I propose as a greedy model which is just proportional to the interpreted size of the discovery. Based on what I observe the ranking of the importance of the uncertaincies goes 1,2,3. <br /><br />Both 1 and 2 are fat-tail PDF's so they deserve the most attention. I can do #2 automatically if I use the entroplet form instead of the damped exponential in the maturation phase. The uncertainty in #3 could only be fat-tail if I put a maximum entropy variation in the rate. I would need some justification for this as I sense that oil companies want to pump pretty close to flat-out (as much as technology allows), with variation of this allowed, but mainly as a perturbation, ala the Oil Shock Model.<br /><br />You know what would be intriguing is to do the FFT on the entroplet form a/(a+t)^2, t>0 instead of the damped exponential exp(-a*t). The entroplet generates the max entropy uncertainty for reserve growth. This would quite obviously show the broadening of the peak. The FFT of the entroplet might be touchy because it has the fat tails which give a large low-frequency component.WHThttps://www.blogger.com/profile/18297101284358849575noreply@blogger.com