Thursday, July 5, 2012

Rising Energy Output/Input Ratio for US Oil and Gas

I discovered something interesting while poking around this morning on the website of the Bureau of Economic Analysis (the agency that produces GDP statistics for the United States).  It turns out they have a spreadsheet with estimates of the amount of energy inputs to different industries.  Since they also have estimates of the outputs of different industries, and for the oil and gas industry "output" means oil and gas, it's possible to form estimates of the ratio of energy output to energy input.  That ratio is above for 1998-2010 - all the data available.

This is very, very roughly like an energy-return-on-energy-invested (EROEI) for US oil and gas - a familiar concept in energy analysis.  It isn't that for several reasons:
  • We are measuring both inputs and outputs in dollars, not joules.  Since the figures mix oil and gas which have had substantial variations in their price ratios, it's not straightforward to turn the dollar ratios into energy ratios.
  • We are comparing the inputs and outputs for a single year - not comparing all the outputs for projects with all the inputs required for that particular project integrated over time.
The last point, for example, should probably be invoked to explain the spike in 2009 when new investment collapsed in the low price environment following the financial crisis, yet output was still fairly high based on prior investments.

This is also a narrow boundaries analysis - we are looking just at energy used directly in operations by the oil and gas industry, not at embodied energy used in their materials or by their suppliers.

All that said, it's hard to believe that if the EROEI of the US oil and gas industry were plummeting (due, say, to the excessive investment required to frack oil and gas out of tight resource plays) that this ratio would be rising.  But rising it appears to be: if the numbers are to be believed, over the last decade or so, the industry is producing more dollars of energy output per dollar of energy input.


Pierre said...

It's because the value of output is dominated by oil output (rising market price), while the inputs are natgas (falling prices over the decade in north america) and coal (mostly stable).

The $ price of a joule in oil or gas has always been very different, and the ratio has spiked even more in the recent past. It's just not possible to infer the EROEI from the $ value.

But it does tell something: liquid fuel are so convenient that people will keep drilling for oil long after it has EROEI<1 (as long as the energy inputs have lower convenience value: natgas, nuclear...). The only limit to drilling is the energy yield of direct conversions like Fischer-Tropsch, and that's something like 50% EROEI.

Stuart Staniford said...

Pierre: it's not really accurate to say that NG prices fell in the US over this period - you can see a chart here for example:

They largely rose from the late 1990s to 2008 and only fell after that.

KLR said...

Interesting - there has to be some simple artifact at work here, though. Probably nothing more than sale prices rising faster than production costs, right?

Missed your post of yesterday; strangely enough, I've been attempting to extend annual production series for the big producing states back further than the EIA's curiously arbitrary date of 1981. So far I have TX/CA/LA/OK/ND, back 110 years in OK's case. Probably dig up just a few more, the rest are pretty piddling. Might build it all into a TOD article, or provide the numbers to Dave Summers, who can blog that blog far better than I.

Do a YOY - TX and ND went through the roof last year, with all that crazy fracking. ND was in the triple digits; far as I can tell, CA only did that once, in 1968 - the year they seemed to reach absolute peak, and they seemed to crash the same amount the next year, too. For that era I had to reverse engineer a graph, though...CA's not very forthcoming with this data.

rjs said...

can that data possibly capture all the inputs, i wonder...for instance, it takes around 2 million gallons of water to frack each well in the bakken...that water is trucked from the missouri river; is the diesel fuel for those trucks included as an energy input into the bakken output?

it would seem that EROEI cant be on a yearly basis either; for a conventional well, say, you invest your energy up front, and yield energy over a 40 year life span; for a fracked well, the maximum flow is measured in weeks or months rather than years; how does the BEA data account for those differences?

here's a graphic i was looking at: but there is too much noise to determine anything

Anonymous said...

Hi Stuart-

Just a thought, but it seems reasonable to say that much of the $ expenditure for any given $ of energy happened a while in the past -- exploration, drilling, pipelines, etc. So in an era of rising energy prices, c.p, you'd expect this number to go up.

How would one try to model that?


Justin Ritchie said...

This trend could also be occurring because of the falling value of the US dollar over the same time frame. Here's a look at the 25 year DXY:$DXY&t=BAR&size=M&v=0&g=1&p=MO&d=X&qb=1&style=technical&template=

As the DXY has fallen since the late 90s and early 2000s, the amount of $ output in energy terms has increased.

Some of the spikes/dips in the energy ratio above even line up with dips/spikes in the DXY. Though I'd have to do a bit more work to see how strong the correlation truly was and would need a longer time frame for the energy ratio too.

jhm said...

I'm not competent to judge the importance of this, but my information is that the NatGas industry (at least in America, and lately) is heavily leveraged (some even see it akin to the mortgage bubble, if much smaller). Could this borrowing affect the data one way or the other?

Mike Aucott said...

This is interesting and good news that there's a source for these data, as limited as they may be as you note.

However there is probably too much variation in these data so far to indicate a trend either up or down. I estimated the numbers from your graph and ran a little statistics program; it indicates the slope is not significantly different from zero.

buck smith said...

Horizontal drilling may be a significant enough innovation to account for the improvement. Say we have a producing formation 40 feet thick located at 10000 feet. We can get 2000 feet of producing well bore by drilling 50 wells 10000 feet deep. That is 500000 feet of borehole to be drilled. Or we drill one well 10000 feet deep with a right turn that goes 2000 feet. As a rough a approximation energy input equals feet drilled. So horizontal drilling can improve EROI by 50 to 1.

buck smith said...

Another big driver of efficiency is fewer dry holes drilled. This primarily driven by innovations 3D sesimic. Look at thsi graph from eia.

The number of wells drilled has gone way up. Dry holes are flat to declining. This is a direct improvement in EROI.

Stuart Staniford said...

buck - I'm inclined to agree with you that this is likely due to improved technology in the oil and gas industry including better seismic and horizontal wells. Your specific comparison on the horizontal isn't a fair comparison though - you could efficiently drain a lot more of the field with the fifty vertical wells than the one horizontal well because the oil won't be able to get through the rock to the latter. So while there's going to be some improvement it won't be anything like 50 to 1.

Ian Langdon said...

Hi Stuart - after being surprised to see this and thinking about it I think it is due to the investments being made. Prior to HMF (horizontal multifracs)most operators were infill drilling their fields; increasing production and cashflow but reducing RLI (reserves life index). Very few reserves were being added.

Since HMF; industry is cycling money on high rate, high decline, and high F&D cost efforts. High cost, and relatively few reserves being added relative to the capital inputs. Lots of short term deliverability though.

Both items could look good on the numbers you've discovered but would look horrid on a true EROI plot if someone could ever construct one. It's almost impossible to construct true EROI estimates on micro or macro basis.