Reihan Salam posted this piece the other day at National Review, asking if natural gas and oil prices are diverging for good, and if so, what that means for startup Bloom Energy, which uses natural gas to power “Bloom Energy Servers” (100 kilowatt [kW] Solid Oxide Fuel Cells). First, I want to address the issue of diverging natural gas and oil prices, and then I’ll take a look at Bloom Energy’s claims about their fuel cell and what a changing natural gas pricing landscape might mean for them.
Natural Gas v. Oil
Salam’s post starts by pointing to analysis from Donald Marron which argues that the while the price ratio between oil and natural gas has historically been fairly stable, it has risen sharply in the past two years, currently hovering around 20 (i.e. 1 barrel of oil costs the same as 20,000,000 Btus [20 MMBtu] of natural gas). Marron attributes this to (1) “a dramatic expansion in domestic natural gas supplies” and (2) the “limited opportunity for energy users – utilities, businesses, and homeowners – to switch from oil to natural gas.” He also predicts that they will return to somewhere closer to their historic ratio in the long term. I think that he is correct in his identification of two important drivers of this decoupling leading up to now, but ignores a third driver of natural gas prices that will be more important going forward (replacing fuel switching between oil and natural gas): climate legislation.
The chart above shows the composition of the US electricity sector from January 1979 to May 2010 plotted against the left Y-axis, with the price ratio of a barrel of Crude Oil to an MMBtu of Natural Gas plotted on the right axis. As Marron notes, the price relationship between Oil and Natural Gas is relatively constant right up until about 2006, when the ratio starts to climb. This climb coincides with the near disappearance of oil-fired generation as a share of the total, to the point where it makes up slightly less than 1% of total generation by kilowatt-hour [kWh] in 2009. This supports Marron’s view that the substitutability between the two fuels has declined to the point where it is largely irrelevant
I would like to focus, however, on the two significant dips after January 2006: the first is a drop from 9.53 MMBtu/barrel to 7.22 MMBtu/Barrel from October to November of 2006 (bottoming out at 6.67 in February 2007), and the second is a drop from 13.00 in September 2008 to 11.19 in October 2008, down to 8.33 in November 2008 (bottoming out at 5.47 in January 2009).
So in the months following both the Democratic takeover of the legislative branch in 2006 and the executive branch in 2008, natural gas quickly became more valuable relative to oil, but then separated again as time passed. I would argue that these changes reflect perceptions of the probability of climate change legislation.
The Effect of Climate Legislation on the Electricity Sector
It may seem counterintuitive at first that an increased likelihood of climate legislation would cause an increase in the relative value of natural gas, because natural gas-fired generation emits CO2. But the key here is the big black area at the bottom of the chart above: King Coal.
For the last 30 years, coal has made up about half of the U.S. energy supply, because coal-fired generation is the cheapest way to produce electricity. However, it’s also the dirtiest: coal emits about 2 pounds of Carbon Dioxide (CO2) per kWh, while natural gas emits about 1.3 pounds of CO2 per kWh, a 35% reduction. So if climate legislation were to pass, coal plants would be the first to go. And that’s exactly what the electricity sector saw in following the Democratic takeover in 2006. Utilities started suspending or canceling their plans to build new coal-fired generation, while state goverments (like Kansas’) started to block the construction of coal plants over concerns about the environmental impact and the impact on ratepayers were climate legislation to go into effect. As the appetite for climate legislation lessened in the face of recent economic troubles (see, for example, Prop 23 in California), coal plants started to make their return to utilities’ planning portfolios.
In the electricity sector, though opportunities for short-term fuel switching between oil and natural gas are mostly gone (sometimes because of environmental restrictions rather than economic concerns, see page 6 [PDF]), opportunities for longer-term fuel switching from coal to natural gas are plentiful. Utilities need generating resources which they can control (as compared to intermittent wind and solar generation), and if climate legislation prevents or discourages them from using coal, natural gas plants will be their next choice. Permitting issues will likely prevent nuclear plants from making any meaningful contribution in the coming years, and carbon capture and sequestration technologies have yet to be shown feasible on a large-scale, much less economical.
I think the coal-natural gas link is likely to be stronger than the oil-natural gas link in the future because of the consumption patterns by sector across these fuels. The electric power sector made up ~30% of the total natural gas consumption and ~94% of the total coal consumption in 2009. In contrast, it made up only 1% of oil consumption during the same time, with the transportation sector taking the lion’s share at 71% (see Tables 5.13a-d). Changes in the electric power sector’s demand for natural gas and coal will have a much more significant impact on the market than changes in its demand for oil. Similarly, the oil market will be much more responsive to changes in the transportation sector, while this sector’s influence on the natural gas and coal markets will be negligible.
So instead of the link between natural gas and oil that we’ve historically seen, I think that legislation regarding emissions from the electricity sector (if it ever comes about) and the extent to which it forces out coal in favor of natural gas will be a major driver of natural gas prices in the years to come.
So what does this mean for Bloom Energy?
In short: nothing. I think without major structural changes in the way that energy is priced, fuel costs will not be a concern for Bloom, while getting their capital costs down to their target levels will be their primary focus.
According to the Bloom Energy datasheet, a Bloom Box requires 6,610 Btu of natural gas to produce 1 kWh of power when operating at capacity. In contrast, the average combined-cycle gas turbine needs an average of 6,924 Btu per kWh over its lifetime (see ‘Heat_Rate’ tab of this file [XLS]). The higher fuel requirements mean that, all else held constant, the costs of a CCGT will be more sensitive to fuel costs than Bloom Boxes. Since electricity is priced based on the marginal resource, and that marginal resource is usually a CCGT (or at peak demand times, a combustion turbine, which uses fuel even less efficiently), electricity prices on the whole will be more sensitive to changes in fuel prices than Bloom Boxes, meaning that increasing fuel prices will actually improve the competitiveness of Bloom Boxes (again, unless there is a major shift in the composition of the electricity market).
This doesn’t necessarily mean you should rush out and buy a Bloom Box with that $700,000 you have lying around. Bloom employee Stu Aaron told Lux Research that the average cost of energy from a Bloom Box over the course of its lifetime would be between $0.08 and $0.10 per kWh, very attractive when compared to the $0.1248 per kWh average electricity price in California,* home to all but one of the existing Bloom Boxes. However, this number assumes a $2500/kW rebate from California, and a federal Investment Tax Credit (ITC) of 30%. I’ve not seen it explicitly stated anywhere, but I would assume it qualifies for accelerated depreciation benefits as well.
To get a better idea of the cost of the unit without incentives, I modified the resource costing module in this model [XLS] in an attempt to get a clearer picture of the costs of the Bloom Box. I started with Stu Aaron’s numbers regarding cost and performance metrics, and made optimistic assumptions for any unknown parameters (no operating or maintence costs, no fuel cost escalation, 100% of costs eligible for the ITC, 100% capacity factor)**. Under these assumptions, I calculated a price between $0.09 and $0.10 per kWh (using $700K and $800K as the bounds), similar to the cost that Bloom claims.
For all sensitivity cases, I assumed that the cost of the Bloom Box was $750K ($7500/kW) before incentives.
Taking away the ITC increased the price by 47% to about $0.13/kWh.
Taking away the California rebate increased the price by 32% to about $0.12/kWh.
Taking away both the ITC and the California rebate increased the price by 94%, to $0.17/kWh.
In a world without ITC or the California rebate, the Bloom Box would have to get down to about $3,000/kW to reach a levelized cost of $0.10/kWh. (This result matches Bloom’s target for individual home-sized systems, which I would say is a good sign for the accuracy of the costing model.)
So, moral of the story: given their efficient use of gas relative to the resources driving electricity prices from utilities, the success of Bloom Boxes isn’t likely to depend too much on the fluctuations in the cost of gas. What will be crucial to their long-term success in a world without government support is whether they are able to achieve their cost reduction goals while scaling down to household-sized installations. If that happens, they could have a few ready-made markets in California, Hawaii, Alaska, and the Northeast, where electricity prices are highest.
The modified costing model I used to do my analysis can be found here [XLSM] (limited downloads, so if it doesn’t work, contact me at adcerebrum at gmail dot com).
[Disclosure: The models used in this analysis are publicly available models developed by my employer, Energy and Environmental Economics, Inc (E3). All information used in writing this post is publicly available, and the views here are mine and mine alone, and do not represent the views of E3.]
* – This comparison may not be as straight forward as it seems, however. Levelized per-kWh numbers can be manipulated to tell a story, as shown on slide 44 of this presentation [PPT]. Lacking better information, I will assume Aaron’s number can be compared on an apples-to-apples basis with current electricity prices.
** – I assumed that the project was financed with 60% debt at 6% and 40% equity at 11.8%, based on information from thatswacc.com.