tag:blogger.com,1999:blog-5552647839482265343.post7397648032915969137..comments2019-08-04T20:46:36.888-04:00Comments on peak oil climate and sustainability: Future Bakken Output and the Average Well ProfileDennis Coynehttp://www.blogger.com/profile/17688179289708969899noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-5552647839482265343.post-33324315955030369372013-09-25T11:09:35.440-04:002013-09-25T11:09:35.440-04:00WHT,
Sorry it took so long for me to reply. Than...WHT,<br /><br />Sorry it took so long for me to reply. Thank you.<br />I had attempted to use the OU diffusion model for the Bakken and Eagle Ford, but I was having difficulty matching the data of the first 24 months, the OU diffusion model ramps up very quickly over the early time period. I was contemplating using a hybrid hyperbolic/OU Model because I like the fact that the OU Model has a physical basis where the hyperbolic is more of a curve fitting exercise. I decided just using the hyperbolic would be easier, especially because my understanding of the underlying physics of the diffusion process is pretty sketchy and I don't really have any data or parameters for these geological formations which I could use to verify if the various parameters I am using in my OU Model make any physical sense. Essentially It was a choice between two different curve fitting problems, each with three parameters, and the hyperbolic seemed to fit the data better over the first 24 to 36 months (which is all the data I have so far for the Shale plays). <br /><br />I do understand however that using the physical model has advantages because I could create a scenario using OU diffusion even without a good understanding of the underlying physics and then some one like you could look at it and say no, parameter x, y or z in your model are not even in the ball park, that x would need to be between a and b, y between c and d, and z between e and f, for the model to match the real world.<br /><br />DCDC78https://www.blogger.com/profile/17688179289708969899noreply@blogger.comtag:blogger.com,1999:blog-5552647839482265343.post-57835685807733386002013-08-02T11:01:36.074-04:002013-08-02T11:01:36.074-04:00DC, here is the fitted Eagle Ford curve
http://img...DC, here is the fitted Eagle Ford curve<br />http://img163.imageshack.us/img163/3246/i4m.gif<br /><br /><br />Cumulative = 10000/(1+SQRT(105.75/t'))<br /><br />where t' = (1-EXP(-0.012*t))/0.012<br /><br />Also shown is a hyperbolic<br /><br />Cumulative = 255/(1+24/t)<br /><br />Interesting how a strong O-U diffusive model is close to a hyperbolic. There are subtle differences in the initial slope and in the asymptote. <br /><br /><br />WHThttps://www.blogger.com/profile/18297101284358849575noreply@blogger.comtag:blogger.com,1999:blog-5552647839482265343.post-62043884511626079752013-06-27T23:03:30.833-04:002013-06-27T23:03:30.833-04:00DC, you were asking about the fit to this curve:
...DC, you were asking about the fit to this curve:<br /><br />http://img441.imageshack.us/img441/8193/p8w.gif<br /><br />Cumulative = 1850/(1+SQRT(150/t))<br /><br />where t in years.<br /><br />It was straight dispersive diffusion, no Ornstein-Uhlenbeck<br /><br />Your Eagle Ford profile has a strong O-U character<br />https://sites.google.com/site/dc78image/images/efwellpro.png<br /><br />Cumulative = 2350/(1+SQRT(50/t'))<br /><br />where t' = (1-EXP(-1.6*t))/1.6<br /><br />So the EUR cumulative is about the same in both cases.<br />The effective diffusivity is 3x higher in the second case, but is tempered by a strong reversion to the mean.<br /><br />Interesting how they differ.<br />WHThttps://www.blogger.com/profile/18297101284358849575noreply@blogger.com