Monday, November 12, 2012

Initial Production Numbers for GMXR Wells are Basically Irrelevant ($GMXR)

The initial production (e.g. "peak 24 hour") numbers of wells are touted by oil companies because they are supposed to imply something about future production.

Using initial production as a metric is predicated on the assumption that if one well has an IP of X bbl/day and another has an IP of 10X bbl/day, then in six months the better well will still be producing 10 times as much per day, that the better one will produce 10 times as much over its lifetime, and that it is therefore worth 10 times as much.

Is that true? For GMXR wells in the Bakken, the answer is empirically NO.

GMXR drilled a well in the Bakken called the EVONIUK 21-2-1H, which they reported had an IP rate of 637 BOE/d. It was actually 521 bbl/d according to the state of North Dakota database. Anyway, after four months, according to that database, the production had decreased to 110 bbl/d.

They also drilled a well called the AKOVENKO 24-34-1H, which they reported had an IP rate of 1483 BOE/d. It was actually 1409bbl/d, and decreased to 117bbl/d after four months.

Woah!! The ratio between the initial production rates was 2.7x. After four months, the ratio between daily production numbers was only 1.06x - essentially no difference. At first glance, this well looked like it was going to be a huge improvement, almost 3 times more production and therefore more valuable for roughly the same cost. But now there's almost no difference?

Does that happen with other wells? If it does, it means the initial production numbers are a useless metric; telling us nothing about how profitable a well is going to be. We can answer this question by looking at the seven GMXR wells in the Bakken for which we have at least four months of data: the WOCK 21-2-1H, FRANK 31-4-1H, EVONIUK 21-2-1H, TABOO 1-25-36H, JOHNSTON 31-4-1H, AKOVENKO 24-34-1H, and LANGE 11-30-1H.

The following scatterplot shows GMXR's claimed initial production rate on the x-axis (BOE/d) and the actual bbl/d number from the state of North Dakota for the period when the well is four months old.

This is on a log scale to make it easier to look for order of magnitude differences. You see immediately that, while wells with higher IPs do have higher production after four months, the production numbers after four months are compressed into a tight band. Meaning: a 10x difference in initial production implies a month four production difference of less than 2x.

Those seven wells had an average IP of 1182bbl/d with a standard deviation of 799bbl/d. However, by month four, the average daily production was 123bbl/d with a standard deviation of 15bbl/d. That's the tight band - the higher initial production numbers aren't translating into higher production down the road. In fact, the wells quickly decline as we can see in the following chart:

Might this be why crude oil production for the third quarter 2012 was 53,305 Bbls, a decrease of 17% over the second quarter of 2012? The company said that the quarter-over-quarter decrease was "a result of a delay in new Bakken wells coming on line and ongoing drilling operations within our Bakken leaseholds."

That's probably literally true, but I think you need to read between the lines. These wells have very steep decline rates. In order to be able to report respectable production numbers, they have to continually drill new wells to get those big bursts of initial production included in the statistics.

No comments: