The future output from the light tight oil (LTO) sector of the US oil industry is the subject of much speculation. Above I present some possible future output scenarios based on a simple model of US LTO, the scenarios are compared with the EIA’s 2017 Annual Energy Outlook (AEO) reference scenario with cumulative output of 82 Gb from 2001 to 2050. The cumulative output of the model scenarios is for the same period (2001-2050).
The models all use the same well profiles from 2006 to 2016 and are based on data gathered from Enno Peters excellent blog, shaleprofile.com. A preliminary hyperbolic profile was fit to the average LTO well data and then the parameters were fit using least squares and solver in Excel so that the model matched the data for output and number of wells added each month over the period from 2011 to 2015. The data for 2016 is incomplete and this leads to an under report of wells added for most of 2016 (from March through October). For this period the wells added were adjusted so that the model matched the output data from the EIA (which is more complete than the data reported at shaleprofile.com.)
The well profiles used are shown below, two were used, a lower profile for the early period and a higher well profile for the later period. The vertical axis is output in barrels per month and the horizontal axis is months from first output.
The well profile in red (219 kb) is the basis for all the scenarios. In every case it is assumed that the estimated ultimate recovery (EUR) or total output from the well over its life starts to decrease in Feb 2017, but the rate of decrease varies from model to model, based on underlying assumptions and the number of new wells added each month.
The 36 Gb, 44 Gb, and 60 Gb scenarios use the oil price scenario in the chart below along with $6 million well cost, transport, OPEX, and other costs of $13/b (combined), royalties and taxes of 27%, an annual discount rate of 10%, and down hole maintenance costs of $42,000 per year, all dollar figures are in constant 2016$. A discounted cash flow analysis suggests these wells would be profitable on a point forward basis under the above assumptions and the oil price scenario below. The 50 Gb scenario would require higher prices as the EUR decrease assumptions for that scenario are identical to the 44 Gb scenario (profitability was not considered when constructing the 50 Gb scenario.)
The maximum rate that new wells were added for the 36 Gb scenario was 870 new wells per month after June 2021 (the completion rate was below this from Jan 2017 to June 2021). For comparison I show in the chart below for all 4 scenarios the annual rate of decrease in new well EUR when 870 wells are completed (the rate of decrease of EUR will vary depending on completion rate).
To evaluate the probability that any of these scenarios might be realized I considered a maximum entropy probability distribution, which only requires that we assume that the probability distribution has a mean. It can be shown for this case that the lowest information (or fewest assumptions) coincides with a negative exponential probability distribution whose probability density function(pdf) has the form
where x is a number greater than or equal to zero and a is a positive constant.
x in this case represents the URR and 1/a is the mean URR or the expected value. The cumulative distribution function(cdf) has the form
I assumed 1/a=60 Gb and the cumulative distribution function is shown below. The curve represents the probability that the URR will be less than the value on the x axis.
My reason for choosing a mean of 60 Gb was that the EIA clearly thinks the likely value is about 80 Gb, my estimate is about 40 Gb based on USGS estimates of 24 Gb for the Permian and 10 Gb for the Bakken/Three Forks and an assumption that Eagle Ford and other LTO plays will produce at least 6 Gb combined. The 60 Gb URR is just the average of these two estimates.
I also assumed based on about 75,000 wells completed through Dec 2016 that at least 15 Gb will be produced from the LTO plays in the US (8 Gb have been produced so far). So my probability distribution is for the 45 Gb yet to be produced and then 15 Gb is added back.
Based on this analysis there is a 33% probability that the URR will be between 33 Gb and 64 Gb (with a 33% probability that URR will be above this range and a 33% probability that it will be below this range.) This range is fairly close to the 36 Gb (37% probability that URR will be lower) and 60 Gb (37% probability that URR will be higher) scenarios. A URR of 46 Gb has a median probability with a 50/50 chance the URR will be higher or lower than this. An intermediate estimate between the mean and median is 53 Gb which has a 57% probability that the URR will be lower.
For this reason my best guess for US LTO URR is between the 44 Gb and 50 Gb scenarios, as the average of the two (47 GB) is close to the median (F50) probability estimate. Finally, there is about a 50% probability that the URR will be between 28 Gb and 77 Gb, with a 25% probability that it might be above or below this range.
A number of side cases are presented below without much comment, note that total wells in these charts includes the 75,000 wells already drilled through Dec 2016 and that in all cases the decrease in new well EUR begins in February 2017.
The 15 Gb model assumes no wells are completed after Oct 2016.
Note the estimate for 616,000 wells to reach 82 Gb by 2051. Also the spreadsheet was only set up to go to Oct 2041 so the steep decline in 2041 is due to no wells being added after that date another 200,000 wells added from 2041 to 2050 might allow another 13 Gb to be produced, note the low productivity 65 kb per well from 2041-2050 vs 167 kb/well from 2001-2041.