Evaluating the shale gas challenge6 May 2013
A probabilistic comparison of the investment risks of nuclear power and natural gas-based electricity generating plants has been carried out using a total-lifecycle power plant model. Although the cost of the gas plant (with carbon tax) is found to be slightly cheaper, that choice of fuel carries a far greater cost uncertainty, suggesting a greater long-term investment risk than nuclear power. By Rob Graber and Tom Retson
This study is intended to compare the cost of electricity from natural gas and nuclear power taking each technology's inherent risks into account. Since the cost of natural gas is a significant component of natural gas generating costs, and because the cost of natural gas is volatile, this study will also describe the natural gas forecasting process used in the generating cost studies.
There is investment risk inherent in both technologies, but from different sources. The key risk of nuclear power resides in uncertain capital costs. For natural gas, the risks are from the uncertain forward cost of natural gas and the potential for environmental compliance costs, primarily from the emissions of greenhouse gases (principally CO2). Because of these uncertainties it is more revealing to use risk-adjusted (probabilistic, or stochastic) forecasts of the comparative costs of electricity. These estimates show the probable range of costs for both technologies, given the uncertainties described above. The costs used are the levelized costs of generating electricity (LCOE), which are useful to compare generation technologies, and can be thought of as the equivalent annual cost incurred over the life of the generating technology having the same present value as actual costs which differ from year to year. The results were obtained using the EnergyPath Market Model EPMM), which simulates the operation of electric generating plants in order to calculate the LCOE (see box, p.16, for a description of EPMM). Since the newest technology nuclear plants are designed to be licensed for 60-year lifetimes, and natural gas generating plants have 30-year lifetimes, it was necessary to assume that the first gas unit (Unit 1) was retired after 30 years and a second unit (Unit 2) was constructed. Other key assumptions are shown in Table 1. Assumptions used for the risk assessment study are also shown in the box.
|Table 1: Key assumptions used in study ($2012)|
|Nuclear||Natural gas (CCGT)|
|Capital costs||$/kWe||$5000||Unit 1: $1107|
|Unit 2: $2045|
|Capital improvements||$/kWe||$20 per fuel reloading||$10 per year|
|Fuel costs (2012)||Uranium: $48/lb U3O8||Natural Gas: $4.35/MMBtu|
|Carbon tax||$/Ton CO2||$25|
EnergyPath Market Model (EPMM)
The EnergyPath Market Model (EPMM) is an Excel-based valuation model which simulates the construction, operation and decommissioning of an electric generating plant. The model also performs levelized cost of electricity (LCOE) calculations. For this study the model was configured to simulate both nuclear and natural gas plants over a 60-year lifetime. The model incorporates Oracle Crystal Ball© risk simulation software. A diagram of EPMM's modules is shown below. Also, assumptions used for the nuclear and gas plants, and financial assumptions, are summarized in Tables 3-5.
Table 3: Data used in EPMM simulation model for nuclear plant ($2012)
|Mean||Std Dev/Volatiliy||Type Distribution|
|Overnight capita costl||$5000/kWe||$500/kWe (std dev)||Normal|
|Uranium price||2012 Price: $42.50/lb U3O8||Volatility 11%/yr||Mean reversion|
|Long run mean: $33.03/lb U3O8|
|Reversion speed: 1.1%/yr|
|Real growth in long run mean: 0.5%/yr|
Table 4: Data used in EPMM simulation model for natural gas plants ($2012)
|Overnight capita costl||Unit 1: $1107/kWe||Unit 1: $110/kWe||Normal distribution|
|Unit 2: $2045/kWe||Unit 2: $211/kWe|
|Natural gas price||2012 Price: $4.35/MMBtu||Brownian Motion with growth|
|Drift rate: 1.57%|
Table 5: Financial data used in EPMM simulations
|Real cost of equity||9.27%|
|Real cost of debt||5.37%|
|Federal income tax||35%|
|State income tax||6%|
Figure 1 shows that the expected levelized generating cost of nuclear power over its 60-year lifetime to be about $87/MWh (all figures are in 2012 dollars). There is a 5% probability that the actual realized generating cost will exceed $99/MWh and a 5% probability that the realized generating costs will be below about $77/MWh. In other words, there is 90% probability that the realized generating cost will be between $77/MWh and $99/MWh -- a range of $22/MWh.
Figure 2 is the same comparison for a high efficiency natural gas plant using a combined cycle technology (and including a carbon tax). Because a second natural gas unit was assumed to be constructed after 30 years, this, introduces the prospect of the second plant having a higher capital cost than the first unit. This was accounted for by assuming that that the capital cost of a natural gas plant grows by 2% per year (in real dollars).
In the case of natural gas the expected value of generation is about $84/MWh, lower than for the nuclear plant. However, the range of uncertainty is higher for the natural gas plant. In this case the 90% probability range is greater than $38/MWh, or nearly twice the range of the nuclear plant. This is the result of the volatility of natural gas over a long time frame and implies a greater investment risk if a natural gas plant is chosen over a nuclear plant (as will be discussed below).
This is one key result for this study; but perhaps more important, not only is the investment risk higher, all the risk occurs after the build decision is made. Thus, natural gas plant investors are in the position of having to manage fuel and potential environmental compliance costs for 60 years after the plant is constructed. To illustrate this point more dramatically, Figure 3 shows the risks associated with a nuclear plant in the post-build decision period. The uncertainty range now has been reduced to about $4/MWh, which represents the risk of uranium price increases. The reason for this result is that uranium costs comprise only about 2-3% of the levelized cost of generating electricity from a nuclear plant. For natural gas, the cost of natural gas comprises 60% or more of the levelized generating cost.
From an investor's standpoint all the risk of a nuclear plant is in the build decision, and can be managed with contractual arrangements between investors and the plant suppliers before any major costs are expended. Unlike the previous generation of nuclear plants which experienced significant cost overruns due to a flawed licensing process (particularly following the Three Mile Island nuclear accident in 1979) and led to major rate implications for electric utilities bearing the financial risk, arrangements in today's nuclear markets place the majority of risk on the plant supplier, providing investors with greater certainty about final construction costs they will bear. In addition, there is a new US licensing process that combines the construction and operating license into a single process, further enhancing investor security.
This brings into play the ultimate risk management tool: withdrawal or delay the project. As a risk management tool this option is unavailable to natural gas plant investors as nearly all the risk occurs after plant construction costs are sunk.
Table 2 illustrates typical results obtained by forecasters using only the static (non-simulated) LCOE. This chart clearly shows the tradeoff between capital costs and variable costs (fuel and environmental compliance) between nuclear and natural gas plants. But more important, it illustrates the risks of relying on static LCOE results. The contrast between the wide range of the risk-adjusted results in Figures 1 through 3 and the point values of the static results in Table 2 is stark. It is particularly dangerous when making generating technology decisions owing to their dependence on a commodity with a market-derived price over a very long time. This is particularly true for natural gas exposed to not only supply and demand; but also the potential for climate change initiatives directed at carbon-emitting fuels.
|Cost component||Nuclear||Natural gas||Natural gas|
|($/MWh)||(No environmental cost)||(With $25/ton CO2)|
|Capital||$ 59.44||$ 12.90||$ 12.90|
|O&M||$ 10.03||$ 3.46||$ 3.46|
|Fuel||$ 5.55||$ 46.99||$ 46.99|
|Taxes (sales and income)||$ 9.79||$ 10.39||$ 10.39|
|Waste disposal||$ 1.00||-||-|
|Environmental compliance||-||-||$ 9.80|
|TOTAL||$ 87.27||$ 73.74||$ 83.54|
The price of natural gas delivered to US electric utilities in 2012 was approximately $4.35/MMBtu. However, this price is unsustainable as it is below the average cost of producing shale gas -- currently the major source of new drilling in the US -- estimated at between $5-8/MMBtu . While the prospects that shale gas will extend the supply of natural gas are positive, like any commodity the cheapest and most-easily mined supply will be produced first. Further, LNG facilities in the US, once constructed to import LNG, are being converted into export facilities as natural gas prices measured in US dollars are as high as $16.50/MMBtu in Japan and $9.00/MMBtu in the UK .
A single metric can be useful in summarizing these results: the coefficient of variability. The coefficient of variability is defined as
COV = Standard deviation/Mean
That is, the coefficient of variability measures the amount of risk (standard deviation) that an investor has to bear in order to get the expected levelized costs (mean). The higher the coefficient of variability, then, the riskier is the project to investors. The COV results for the cases described above are shown below in Figure 4. As shown, nuclear power represents a significantly smaller financial risk relative to natural gas, and particularly so after construction.
It may be argued that decommissioning also represents a higher risk for investors in the case of nuclear power. However, this is already accounted for in Figures 1 and 2, and moreover, investors have possibly up to 80 years before the decommissioning decision must be made, resulting in an annuity that is easily managed.
These results are being validated in real life. In March, 2012 the US Nuclear Regulatory Commission awarded combined construction and operating licenses (COL) to two privately-owned nuclear plants in the southeast US: the Vogtle 3&4 units owned by Southern Company and the Summer 2 & 3 units owned by South Carolina Electric and Gas. (First nuclear concrete has recently been poured for Summer 2 and Vogtle 3). Both utilities obtained regulatory approval from their respective state regulatory bodies, largely on the basis of fuel diversity. It was precisely a reluctance to develop additional gas resources on the very basis that it left both companies vulnerable to increased fuel and regulatory compliance costs that was instrumental in choosing nuclear -- in spite of considerable opposition from parties opposed to nuclear power. While coal would have been an option, both utilities already have substantial coal capacity and there is warranted anticipation that coal will be increasingly targeted by the US Environmental Protection Agency for stringent emissions controls, including, potentially, controls on CO2 emissions. A third utility -- Florida Power and Light -- is likely to also be granted a COL for the construction of Turkey Point 5 & 6, and FPL has made exactly the same fuel diversity case to the Florida Public Service Commission. All of these regulatory agencies feel strongly enough that nuclear power is essential that they granted the utilities the ability to place their construction costs in the rate base for recovery prior to actual plant operation -- a first for U.S. electric utilities.
In conclusion, while natural gas currently has lower generating costs, there is a significantly higher investment risk in natural gas that does not appear to be reflected in the current "bandwagon effect" that natural gas is enjoying owing to very low current natural gas prices and no environmental compliance costs.
The natural gas markets
This section is intended to show how the price of natural gas was forecast for use in the generating cost comparison between nuclear power and a high-efficiency natural gas plant in the previous section.
In 2009, electric power consumed 30% of natural gas supplied in the USA; within the electric power industry, it generated 18% of electricity, second only to petroleum (EIA Annual Energy Outlook 2009) . Unlike petroleum, natural gas markets are regional rather than global and there can be significant price disparities between markets. Unlike oil, gas is transported through pipelines and delivery costs represent a large fraction of the total costs in the supply chain. This makes, for instance, European gas more expensive than US gas. The evolution of LNG import and export capability may make gas more of a global commodity than it is currently. There are two noticeable historic trends in natural gas prices (shown in Figure 8, p. 17). First, the price of natural gas was generally rising over the period 1976-2011 with an annual average growth rate of about 3.5% (in constant 2011 dollars). Second, there is significant volatility -- about 30%/year -- in natural gas prices, especially following the year 2000, when natural gas supplies were unable to adapt quickly in the short term to changes in demand during a period when a perception of natural gas scarcity prevailed. The initial onset of volatility corresponds to the repeal of the Power Plant and Industrial Fuel Use Act (FUA) in 1987 which ended the ban on the use of natural gas use in power plants (originally instituted in 1978 following the first energy crisis).
More recently, owing to technological advances in drilling, very large reservoirs of natural gas found in shale deposits (shale gas) have been discovered in the US. Shale gas differs from conventional gas deposits in that it is much less permeable than conventional gas and requires more effort and cost to recover. In addition to shale gas, the unconventional gas category includes coal bed methane, tight gas sands and methane hydrates. In all cases more technology and lower recovery factors distinguish the unconventional natural gas category.
It is likely the large and growing reservoirs of shale gas are contributing to current US natural gas prices remaining depressed. US conventional gas production (including associated gas from oil drilling) has been steadily falling since 2000 while unconventional gas resources have been increasing, as shown in Figure 6.
While the increase in production in unconventional gas, and primarily shale gas, is impressive, the potential supply of unconventional gas is even more so. The US maintains the third largest supply of natural gas globally, behind the Middle East and Russia. Perhaps as significant the countries with the largest need for energy, India and China, have nearly no natural gas resources, which could very well have a large bearing on the price trajectory of natural gas. With 2100 trilion cubic feet (Tcf) of estimated resources (including 630 Tcf of shale), the US represents about 13% of the global natural gas resource base. At current rates of US demand, this purportedly represents about 100 years of available natural gas; there is little concern over running out of gas. However, in spite of what Figure 6 shows, this does not mean that natural gas is going to remain a cheap energy resource.
The history of US natural gas wellhead prices is shown in Figure 8.
Forecasts of natural gas prices
With this background, natural gas price forecasts from five different reputable sources [3-7] have been compiled. The forecasts have all been converted to $/MMBtu in 2012 dollars and cover the period 2011-2050, although in most cases forecasters used five-year intervals and may not have provided full forecasts to 2050. Nevertheless, the trend is very clear, as shown in Figure 7: natural gas prices will almost certainly be increasing over the next 60 years, as would be expected given their low prices today. In 2050 the high and low forecasts are, respectively, $18/MMBtu and $9.40/MMBtu. The average growth rate is 1.57%. Using this average growth rate results in the forecast shown in Figure 8 in 2012 dollars along with the 90% confidence interval. While the upper confidence interval appears to be quite wide, this reflects the high volatility of natural gas. The lower confidence interval, by necessity, is much narrower as natural gas prices can never be negative.
As is evident from Figure 8, the projected 2050 price of natural gas is projected to be within a wide range of $19/MMBtu and $3/MMBtu with an average expectation of $8/MMBtu in 2012 dollars.
For the simulation model (which is stochastic) a Brownian Motion model was used to forecast prices. This is the basis for the forecasts shown in Figures 1 through 3. This model used the historic natural gas volatility of 0.31 and a drift rate of 1.57% (from the reputable forecasters' results shown in Figure 7).
 Berman, A.E. and Pittinger, L.F., "U.S. Shale Gas: Less Abundance, Higher Cost", The Oil Drum, August 5, 2011 (www.theoildrum.com/node/8212).
 BP (British Petroleum), "Energy Outlook 2030, Statistical Review of World Energy 2012", Natural Gas Prices http://www.bp.com/sectiongenericarticle800.do?categoryId=9037181&contentId=7068643.
 EIA (Energy Information Administration), Annual Energy Outlook 2012 with Projections to 2035 (June, 2012), Washington D.C, Reference Case Table 1 http://www.eia.gov/forecasts/ aeo/data.cfm
 FPL (Florida Power & Light), In Re: Nuclear Power Plant Cost Recovery For the Years Ending December, 2012 and 2013, Testimony & Exhibits of Steven R. Sim, Docket No. 120009-EI (April 27, 2012), Exhibit SRS-2 (Tallahassee, Florida, USA)
 PacifiCorp, 2011 Integrated Resource Plan, Volume 1 (March 31, 2011), Figure 7.6 "Comparison of Henry Hub Gas Price Forecasts Used for Recent IRPs", Portland, Oregon, USA http://www.pacificorp.com/content/dam/pacificorp/doc/Energy_Sources/Integrated_Resource_Plan/2011IRP/2011IRP-MainDocFinal_Vol1-FINAL.pdf
 EIA (International Energy Agency), World Energy Outlook 2011: Are We Entering a Golden Age of Gas: Special Report, Paris, France, Table 1.1 "Natural Gas Import Price Scenarios By Scenario" http://www.iea.org/weo/docs/weo2011/WEO2011_GoldenAgeofGasReport.pdf
 MIT (Massachusetts Institute of Technology), The Future of Natural Gas: An Interdisciplinary MIT Study (June 6, 2011), Cambridge, Massachusetts, USA, See Figure 3.2 (Scenario 1), Figure 3.3 (Scenario 2), Figure 3.8 (Scenario 3) http://mitei.mit.edu/system/files/NaturalGas_Report.pdf
Rob Graber, and Tom Retson, EnergyPath Corporation