Is A Natural Gas Strategic Reserve for the US Necessary?
A System Dynamics Approach
James Ellison
Andjelka Kelic
Thomas Corbet
Critical Infrastructure Modeling & Simulation
Sandia National Laboratories
PO Box 5800, MS 1137
Albuquerque, NM 87185-1137
Tel: (505) 284-7811 / Fax: (505) 284-3850
jelliso@sandia.gov
akelic@sandia.gov
tfcorbe@sandia.gov
Abstract
The large volume of shut-in natural gas production in the US Gulf of Mexico following
the 2005 hurricane season led some US policymakers to consider whether creating a
Natural Gas Strategic Reserve (NGSR) might be beneficial. This paper uses a system
dynamics-based approach to analyze whether a NGSR is needed, and what having one
would mean for the US natural gas infrastructure. Analysis shows that the infrastructure
is likely resilient in the face of a more stringent test than the 2005 hurricane season
provided. Moreover, as the infrastructure is essentially a closed system, any
replenishment of the NGSR would compete for gas with other users, and depending on
the rate of replenishment could cause a disruption as large as that it was created to
prevent.
Keywords: natural gas, strategic reserve
Problem Statement
The goal of this paper is to examine whether establishing a Natural Gas Strategic Reserve
(NGSR) for the US would be justified. We divide this problem into two issues: 1) how
resilient is the current system, and 2) how would having an NGSR impact the system as a
whole.
We define resiliency as the ability to supply gas to customers willing to pay the clearing
price, even in the face of supply constraints. If gas were unavailable to an area at any
price, we would say that the system is not resilient.
Moreover, we do not attempt to measure the ability of a NGSR to reduce natural gas spot
price increases after a production shut-in, and calculate an economic benefit to gas
consumers due to the existence of the reserve. We believe the primary concern of the US
government, which would presumably finance the creation of an NGSR, should be that
gas be available to paying customers in the aftermath of a disaster — not whether
customers pay more after a disaster than they would like.
We will first present an introduction to the US natural gas system, proceed to examine the
model created to assist in evaluating system resiliency and the effect of an NGSR on the
system, and finally discuss the pros and cons of an NGSR.
Introduction
Natural gas has become a key energy source for the US. About 24% of all energy
consumed in the US is from natural gas. 17% of electrical power in the US is generated
by natural gas, and natural gas supplies 30% of US industrial energy consumption (EIA
2005).
Natural gas also happens to be the cleanest of all fossil fuels, producing the least amount
of greenhouse gases and combustion byproducts of all fossil fuels. Environmental
considerations and stricter air pollution laws are not the only reason for the attractiveness
of natural gas. It is projected that most new power plants to be built in the coming
decade will be gas turbine plants, as these have the lowest capital cost and highest energy
efficiency of all fossil fuel plants.
Natural gas is unique among fossil fuels in another way: as a gas, it cannot be easily
transported in the way that solids or liquids can. It must be transported either by
pipelines, or cooled down to -162C (where it becomes a liquid, called Liquid Natural Gas
— or LNG) and transported in special thermally insulated vessels.
Below we will discuss the key elements of the US natural gas system: production,
consumption, storage, transmission, and imports. We will also discuss the overall
impacts of the 2005 hurricane season on the system.
Production
Production throughout the year in North America is fairly constant. The rate at which
new wells come online and old wells become depleted is roughly the same. Wells are not
adjusted for output throughout the year based on demand — they are designed to operate
at maximum efficiency at all times.
In 2003, 24 Tef was withdrawn from US wells (with 17.8 Tcf from gas wells, and 6.2 Tcf
from oil wells). Of this, 19 Tcf of dry gas remained to be sent to customers (EIA 2005).
44% of this amount was produced in the Southwest and 15% in the Central region, with
the Southeast, West, Northeast, and Midwest producing 3%, 2%, 2%, and 1%,
respectively. Alaska produced 15% of the total, while Federal Offshore areas (mainly in
the Gulf of Mexico) produced 19%.
Consumption
In stark contrast with the constant rate of gas production, gas demand is highly seasonal.
Residential and commercial demand peak in winter, and are at a low point during the
summer. Power generation from natural gas has a peak in the summer. And industrial
consumption is fairly constant throughout the year.
The net effect is that gas demand is much higher than production during the winter, and
much lower than production during the summer (with a slight summer uptick due to
power generation demand). With production constant and demand seasonal, the role of
storage capacity is to bridge that gap.
Storage
As of May 2004, there was roughly 8 Tef of gas storage capacity in the US, with about 4
Tcf needed to always remain in storage as base gas. Base gas is needed in order to
provide enough pressure for gas to be withdrawn on demand. Therefore, total working
storage capacity is about 4 Tcf. This is equal to approximately 17% of annual
consumption.
Most gas storage is in depleted gas or oil fields, with the second largest storage capacity
being in aquifers, and the smallest in salt caverns. Depleted gas or oil fields are the
cheapest to commission, as they take advantage of existing wells, internal distribution
systems, and pipeline connections. They are also widely available. Aquifers are suitable
for gas storage if the water-bearing sedimentary rock is overlaid with impermeable cap
rock. The use of aquifers requires more base gas than does depleted gas or oil fields.
Last, salt caverns provide high withdrawal and injection rates, and have low base gas
requirements. The commissioning of cavern storage is more expensive than depleted
field conversions, but this type of storage allows several withdrawal-injection cycles per
year (EIA 2004).
Gas storage is heavily clustered in the consuming Northeast / Midwest, with most of the
depleted field storage in Pennsylvania, West Virginia, Ohio, and Wisconsin, and most of
the aquifer storage in Illinois and Indiana. Storage is also clustered in the producing
Southwest, where there is mainly depleted field storage, but also a growing amount of
salt cavern storage (primarily in coastal areas).
Typically, gas storage reaches its peak volume in about October, and gas is drawn down
throughout the heating season. About March, gas stocks reach their low point, after
which the buildup for winter begins. Last, many salt formation and other high-
deliverability sites have been developed by independent storage service providers, who
cater to customers that require quick response times, such as electricity generators and
gas marketers. The seasonal role that storage plays can be clearly seen in the graph
below.
Figure 1: Total Natural Gas in Storage in the US
9,000
8,000
7000 oe
@ 6000
3 a
5 91000
3
2 4000
3
5 3,000
o — = Base Gas
2,000
= = =Total Gas in Storage
+000 —Total Storage Capacity
°
s$ 8 8 8 8 8 8 8 8 8 8 8 8 8
2 = 5 8 2 3 2 = 3 & 2 5 2 &
‘Source: Energy Information Administration, Natural Gas Monthly (OCEVEIA-0130), May 2002-July 2004
Source: EIA, Natural Gas Monthly, May 2002 — July 2004
New storage facilities are continuously being constructed. As of May 2006, FERC lists
130 Bef of new storage projects that are on the horizon (projects where a permit has not
yet been requested of FERC) (FERC 2006).
Transmission
The largest capacity pipeline route is from Gulf Coast production (onshore Louisiana and
Texas, as well as offshore Gulf of Mexico) to the Midwest and Northeast.
The Western part of the country uses much less gas than does the East. It is served,
however, by multiple sources — namely, by pipelines from the Southwest (connecting into
Southern California), pipelines from Canada, and pipelines from the Rocky Mountain gas
fields.
Pipelines from gas-producing western Canada connect to the Northwestern, Central,
Midwestern, and Northeastern parts of the US. A small amount of gas is exported to
Mexico.
The current US interstate natural gas pipeline consists of over 200,000 miles of
transmission lines with an estimated daily delivery capacity of about 119 Bef (billion
cubic feet) (Tobin 2001). The average daily consumption rate in 2000 was half this
amount. This results in a capacity utilization factor of roughly 50%.
International Connections / Imports
Pipelines from gas-producing western Canada connect to the Northwest with a capacity
of about 4.6 Bef/day (billion cubic feet per day), to the Central part of the US with a
capacity of 4.2 Bef/day, to the US Midwest with 4.3 Bef/day capacity, and into New
England with about 3.5 Bef/day capacity (Tobin, 2005). The US Southwest has 15
interconnections with Mexico, with an aggregate export capacity of 3.6 Bef/day (Gaul
and Alic 2005).
In 2004, net imports to the US were 3.4 Tcf, which is a 4.3% increase over the previous
year, but below the 2001 volume of 3.6 Tcf. Net imports from Canada in 2004 were 3.2
Tcf. That same year, the US exported 0.4 Tcf to Mexico. Net LNG imports were about
0.6 Tcf (Gaul and Alic 2005).
Given both the small size of the current and potential volume of imports at existing LNG
facilities, we can consider North America essentially a closed system.
Impact of the 2005 Hurricane Season
Hurricanes Katrina and Rita caused about 800 Bef of natural gas to be shut in (as of June
19, 2006), which is about 22% of annual Gulf production, or almost 4% of annual US
natural gas consumption. Natural gas prices skyrocketed to over $15.00 per million Btu
(MMBtu) in the aftermath of the hurricanes, and residential consumers paid record prices
for natural gas heating in the winter. However, the market was successful in allocating
the reduced production among consumers, and no shortages developed.
The hurricane season was followed by a warmer than average winter. High gas prices
combined with a mild winter resulted in below-normal consumption, which in turn
resulted in the level of working gas in underground storage in the summer of 2006
exceeding the range of working gas in storage over the past five years, as shown in
Figure 4.6 below.
Figure 2: Working Gas in Underground Storage (red line) Compared with 5-Year Range
(grey area — showing minimum and maximum storage volumes at the same time of year for
2001 to 2005)
3,500 3,500
3,000 3,000
% 2,500 v4 2,500
2
3
2 2000 2,000
&
= 1,500 f 4,500
&
© 1000 4,000
500 500
0 a
3s 3 3 8 8 gS 8 8 8
¢ 2 3 = ¢ 2 3 5 ¢
3 8 4 = 3 8 é = 3
Source: EIA, at http://tonto.eia.doe.gov/oog/info/ngs/ngs.html
However, perhaps we should plan for a more severe test of nature than the 2005 hurricane
season followed by a mild winter brought. If we could determine whether the system
could continue to function in the face of, say, a couple of 2005 hurricane seasons back-to-
back, with a couple of harsh winters following them, then we would be more confident in
judging whether a strategic reserve would be needed or not. It is with this test in mind
that a model was created, which will be described below.
Model and Scenario Description
In order to better understand the dynamics involved and the effect a strategic natural gas
reserve would have on the system, an aggregate model of natural gas storage, supply, and
demand was created. The model is used to examine the effects a series of natural
disasters would have on the natural gas infrastructure and the benefits of a strategic
natural gas reserve. The natural gas system is modeled at an aggregate national level and
neglects regional effects, such as natural gas transfers between regions.
The model is an aggregate model of the natural gas system at the national level. As
shown in Figure 4.7, production from all regions feeds into a single storage area from
which national sector aggregated demand is removed. The model was balanced with data
by sector (residential, commercial, industrial, and power generation) from the Energy
Information Administration. The data used is a four (2001 to 2004) or five year average
(2000 to 2004), depending on availability. Data from 2005 onward is not used due to the
influence of hurricanes Katrina and Rita.
Figure 3: Model Overview
Aggregate
Production
Storage Bz >is
Strategic aa Consumption
Reserve Production a wv
“5 Augmentation Pri
rice
For the purposes of testing a strategic storage reserve, the model assumes that the reserve
contains 750 Bef of natural gas, is opened one month after the natural disaster and stays
open for five months. During this time period, the reserve is used at a rate that matches
the current amount of daily production being lost due to the disaster. The storage reserve
is not refilled during the course of the model run.
The pricing mechanism assumes that price adjusts as a factor of demand, production, and
storage volumes. Price is used to adjust demand to keep the system in balance. A price
increase causes a decrease in demand and a price decrease causes demand to increase.
The elasticities of demand used are 0.1 for the commercial and residential sectors
(Gresham 2002), 0.15 for the power generation sector, and 0.2 for the industrial sector.
The algorithm compares current demand and storage volumes with historic values for
those numbers and current production with maximum production capacity and adjusts
price to bring the system back into equilibrium. For example, if current volumes in
storage are lower than what they have historically been for the current time of year, a
reduction in demand is necessary to keep the system in balance, so the price increases. If
volumes are higher than historic values, then a demand increase is necessary to keep the
system in balance, and the price decreases.
Production and storage are weighted as having a larger effect on the price than demand,
with price adjusting in proportion to the ratio between current and maximum production
and storage and as the square root of the ratio of current and historic demand. Price is
adjusted with a 30 day delay to represent the time it takes to observe the information and
make changes to the price.
Baseline Model Run
The baseline run of the model has no disruptions and does not include the effects of price
on consumer behavior. As shown in Figure 4.8, the storage levels for the nominal run of
the model exhibit similar dynamics to the EIA 5 year storage range shown in Figure 5.1.
Also shown in Figure 4.8 is the baseline run of the model with the pricing mechanism
turned on. Storage levels are the same for both model runs, as are the consumption levels
shown in Figure 4.9. The baseline run of the model with the effects of price on consumer
behavior active is considered the ‘nominal’ run of the model, since it illustrates the
dynamics seen in the actual system.
Figure 4: Storage levels for nominal model runs
Storage
4,000
3,000
2,000
1,000
0
0 6 12 18 24 30 36 42 48 54 60
Simulation Time (month)
Storage : baseline Bef
Storage : baseline + price Bef
Figure 4.9: Consumption levels for nominal model runs
Consumption
100
85
70
35
40
0 6 12 18 24 30 36 42 48 54 ~~ 60
Simulation Time (month)
Consumption : baseline BeffDay
Consumption : baseline + price BeffDay
Scenario One -- 2005 Hurricane Season
The initial testing scenario of the model is a disruption equivalent to the combined effects
of hurricanes Katrina and Rita followed by a warmer than average winter. The hurricane
effect on production is shown in Figure 4.10. The hurricane disruptions are implemented
in the model by a linear drop in Gulf Coast production of 6 Bef per day to 0 Bef per day
(return to nominal) over the course of 9 months starting in September. The zero time of
the model is January | of the year, thus month 8 represents the first day of September and
month 12 is the start of January of the next year. Base production levels are assumed to
be constant unless disrupted.
The warmer than average winter is implemented by decreasing residential and
commercial sector demands for natural gas by ten percent for the months of December,
January, and February as shown in Figure 4.11.
Figure 5: Production with the Katrina/Rita hurricane disruption
Aggregate Production
70
62.5
55
47.5
40
0 6 12 18 24 30 36 42 48 54 60
Simulation Time (month)
Aggregate Production : katrina BeffDay
Figure 6: Consumption for Katrina/Rita warmer winter
Desired Consumption
100
85
70
35
40
0 6 12 18 24 30 36 42 48 54 60
Simulation Time (month)
Desired Consumption : katrina winter BeffDay
Desired Consumption : baseline BeffDay
Scenario Two — Two 2005 Hurricane Seasons and Two Cold Winters
The more severe scenario that is created for use in the model is an 820 Bef production
disruption beginning in September of the first year of the model run, followed by a colder
than average winter for the months of December, January, and February. In the second
year of the model run, there is another 820 Bcf production disruption, followed by
another colder than average winter.
The colder winters are implemented by raising residential and commercial sector
demands for natural gas 10 percent for the months of December, January and February as
shown in Figure 4.13.
Figure 7: Disruption scenario production
Aggregate Production
70
62.5
47.5
40
0 6 12 18 24 30 36 42 48 54 ~~ 60
Simulation Time (month)
Aggregate Production : baseline disruption Bef/Day
Figure 8: Consumer demand with two colder than average winters
Desired Consumption
100
85
70
35
40
0 6 12 18 24 30 36 42 48 54 ~~ 60
Simulation Time (month)
Desired Consumption : baseline BeffDay
Desired Consumption : baseline disruption BeffDay
Simulation Results
The model is first used to simulate the combined effects of Katrina and Rita and a
warmer than average winter (10 percent reduction in demand) for comparison to current
data for storage levels which are higher than normal, as shown in Figure 4.14. As shown
in the results in Figure 4.14, storage levels increase to higher than nominal after the
hurricane season and then begin to return to normal.
Figure 9: Katrina/Rita and warmer winter historical example
Storage
4,000
3,000
0 6 12 18 24 30 36 42 48 54 ~ 60
Simulation Time (month)
Storage : nominal Bef
Storage : katrina Bef
Figures 4.15 and 4.17 show the comparison runs of the model for the nominal model run
without disruption, the disruption run, and for the disruption with the strategic reserve
enabled. The results of the pricing mechanism on the disrupted scenario is the run labeled
“nominal disruption” shown in Figures 5.7 and 5.8. Price increases when nominal
production and storage levels decrease and as demand increases. This causes shortages
of natural gas to drive the price up, and drive down consumption according to the
elasticity of the different consuming sectors.
Higher prices drive down consumption levels after the hurricanes, resulting in lower
demand and an increased amount in storage. Since demand is curtailed there is no
shortfall in available natural gas. By the onset of the second hurricane there is excess
natural gas in storage, which drives the price down and consumption up to cause the
system to begin to return to its nominal levels.
Therefore, in the face of both the test given by the 2005 hurricane season, as well as the
simulated test of two 2005 hurricane seasons and two cold winters, the US natural gas
system seems to be resilient.
Effect of a Strategic Reserve
A strategic natural gas reserve is introduced for the “nominal disruption + reserve” run
shown in Figures 4.15 and 4.17. The strategic reserve contains 750 Bef of natural gas
and is opened one month after the hurricanes, and allowed to remain open for five
months. During that time period, natural gas is moved from the strategic reserve to
normal storage to match the current production that is still shut-in due to the hurricane.
The use of the strategic storage reserve is shown in Figure 5.11.
Figure 10: Natural gas daily storage levels.
Storage
0
0 6 12 18 24 30 36 42 48 54 60
Simulation Time (month)
Storage : nominal Bef
Storage : nominal disruption Bef
Storage : nominal disruption + reserve Bef
The strategic reserve is fed directly into storage at a rate to make up for the currently lost
production due to the hurricane. By the end of the first five months of usage of the
reserve, about 200 Bef remain. This reserve is quickly used up once it is again opened
for the second hurricane season and lasts for a little over a month. After this point
storage levels start to fluctuate similarly to the run without a reserve.
Figure 11: Use of the strategic storage reserve during disruption
Strategic Reserve
800
600
400
\
0 6 12 18 24 30 36 42 48 54 60
Simulation Time (month)
Strategic Reserve : nominal Bef
Strategic Reserve : nominal disruption Bef
Strategic Reserve : nominal disruption + reserve Bef
Figure 12: Natural gas daily consumption levels.
Consumption
100
75
50
25
0 6 12 18 24 30 36 42 48 54 ~~ 60
Simulation Time (month)
Consumption : nominal BeffDay
Consumption : nominal disruption Bef/Day
Consumption : nominal disruption + reserve —————————_————._ Bef/Day
Storage levels rise higher after the hurricanes in the run without the strategic reserve as a
result of higher price volatility. Without the strategic reserve, production levels are
below normal as is the amount in storage, this drives price upward and curtails demand so
that more natural gas ends up in storage. With the strategic reserve active, production
levels are lowered, but storage levels remain much closer to nominal causing a smaller
price increase. The lower price increase results in less curtailment of demand, so less
excess natural gas in storage.
Figure 13: Natural gas pricing
NNG: Perceived Price
40
30
20
10
0 6 12 18 24 30 36 42 48 54 ~ 60
Simulation Time (month)
"NNG: Perceived Price" : nominal
"NNG: Perceived Price" : nominal disruption
"NNG: Perceived Price" : nominal disruption + reserve
AAR
Once the system is disrupted, use of a strategic reserve causes the storage and
consumption levels to start to return to their normal levels faster than without the
strategic reserve. It should be noted that natural gas price, which here is expressed in
dollars per MMBtu, is not meant to be a forecast of actual natural gas prices. Regardless
of the precise value, in this model high natural gas prices depress demand, while low
prices encourage it.
Mitigating Negative Consequences: What are we Concerned About, and What can
be Done?
Economic Consequences versus Physical Shortages
When prices are allowed to increase following a supply disruption, such that demand is
decreased enough to be equal to the decreased supply, then no shortage occurs, but there
can be considerable price spikes and an increase in cost to consumers. This is in fact
what we observed in the wake of Hurricanes Katrina and Rita — US natural gas
production on an annual basis was reduced by about 3%, which caused a sharp increase
in price, but did not result in a shortage.
At the same time, it is possible that supply could be reduced so suddenly and drastically
that the entire system (or large parts of it) would be brought to a halt; in this case no
amount of price increase would be sufficient to prevent shortages. This was very nearly
the case with petroleum product supplies to the East Coast in the aftermath of Katrina,
when the Colonial petroleum product pipeline was shut down for 55 hours due to power
outages at several of their pumping stations. Most fuel depots have just three to five days
of supplies (Cummins and Gold 2006).
Clearly, running out of natural gas would be a much greater national security concern
than having to endure increased costs that successfully ration smaller volumes.
We believe that a difficult hurricane season that greatly disrupts natural gas production in
the Gulf is the worst natural disaster that we can reasonably expect. And apart from a
few local exceptions, we believe that such a disaster is much more likely to result in
increased economic costs (from potentially dramatic increases in the spot market price for
natural gas) than in a regional or national shortage.
Results of modeling, as described in Section 5, also support the contention that even after
a couple of damaging hurricane seasons and cold winters, if price is allowed to be set
freely then the market should be able to cope without having emergency reserves.
Nevertheless, it makes sense to explore what options there are for having additional
supplies of natural gas on hand in case of an emergency, how much it would cost, and
how long it would take to obtain such a reserve.
Parallels Between the Strategic Petroleum Reserve and a Potential Natural Gas Reserve
In order to assist our analysis of whether establishing a strategic natural gas reserve
makes sense, it would help to first review the rationale behind establishing the Strategic
Petroleum Reserve and the rules for accessing the stored petroleum.
The US Strategic Petroleum Reserve (SPR) is comprised of crude oil storage in five salt
dome caverns in Texas and Louisiana. The SPR oil can be withdrawn at a rate of 4.3
million barrels per day (m bbl/day) for the first 90 days of withdrawal, with declining
rates afterwards (Bamberger 2006).
The origins of the SPR stem from the 1973 Arab oil embargo on the US. As a result of
US support for Israel in the 1973 Yom Kippur war, Arab oil producers embargoed the
US, in order to create a clear supply shock and show that they had leverage over the US.
The embargo resulted in a dramatic increase in world crude oil prices, and the supply
reduction coupled with price controls in the US led to gas lines and shortages.
As a result of this experience, Congress authorized the SPR in the Energy Policy and
Conservation Act (EPCA, PL 94-163) in 1975 to help prevent a repeat of the situation.
Though it was understood that no amount of strategic storage could completely isolate
the US from the price of oil in a crisis, the logic was that such storage could help blunt
the magnitude of the market response. Moreover, strategic stocks would buy time for the
crisis to resolve itself, or time for the US to seek a resolution to the crisis before it
escalated. It was also hoped that the existence of strategic stocks would discourage oil
exporters from using oil embargoes as a weapon.
Apart from fears of a possible petroleum shortage, the “economic dislocation”
(Bamberger 2006) caused by the Arab oil embargo was very real — and there was the
desire to prevent this from happening in the future. We interpret “economic dislocation”
to mean the power of petroleum exporting countries to affect a dramatic increase in crude
oil prices, and in so doing, affect a transfer of wealth from petroleum importing nations.
In thinking about a Natural Gas Strategic Reserve (NGSR), we should first point out the
areas of difference with the SPR.
First, North America does not rely on natural gas imports in the way that it relies on
crude oil imports. As North America consumed 27.7 Tcf of natural gas and imported 650
Bef of LNG in 2004, total imports comprised only about 2% of all natural gas consumed.
Therefore, at present, there is no need to have stocks to tide the US over in case of an
embargo, or to have stocks that could act to discourage an embargo, simply because an
embargo would have virtually no effect.
Second, with crude oil, large increases in oil prices amount to a tax on US residents by
petroleum exporting nations (Bernanke 2006). Increases in natural gas prices, on the
other hand, go largely to domestic producers. Since wealth largely stays within and
therefore is taxed within the US, natural gas price increases are of a different nature and
may be of less concern to policymakers.
Establishing a Natural Gas Strategic Reserve (NGSR)
As the 2005 hurricane season was unusual in its severity, and other natural disasters
would either have a much lower impact or are highly unlikely (such as an earthquake in
the New Madrid Seismic Zone), we believe that it is reasonable to compare the size of a
potential NGSR with the amount of natural gas production shut-in due to Hurricanes
Katrina and Rita.
We therefore propose that the volume of working gas in the NGSR should be large
enough to make up for lost production for the 2005 Hurricane Season, and that location
of the storage (as long as it can get gas on pipelines headed to the Midwest and
Northeast) is not so important. Since the lost production will be in the Gulf, it makes
sense to place storage in the same area. Given the proximity to major transmission lines,
the geology that lends itself to salt cavern development, and the number of spent oil and
gas fields, placing storage in the area would likely be cost-effective.
Moreover, we do not believe the nature of the storage to be especially important. What is
necessary is to be able to replace lost production by the end of the winter season, not to
immediately replace all lost production. In other words, if Gulf production is reduced by
9 Bef the first week after the storm, by 8 Bcf the second week, and then fully restored by
the third week, then it is only necessary to make up the lost 17 Bef by the end of winter —
it need not be done in the exact same two weeks that production was shut in.
Exactly how much natural gas production was lost as a result of the 2005 hurricane
season? From August 26, 2005 through June 19, 2006, about 800 Bef of natural gas was
shut in, which is equivalent to 22% of the yearly production of gas in the US Gulf. And
as of June 19, 2006, 9% of the normal daily production of 10 Bef remained shut-in (MMS
2006).
How much would it cost?
A typical two-cycle depleted reservoir storage field costs about $5 million to $6 million
per Bef of working gas storage, and salt cavern storage able to cycle six to twelve times a
year costs about $10 million to $12 million per Bef along the Gulf coast, and as much as
$25m per Bef in the West and the Northeast (FERC 2004).
If we assume that about 750 Bef of additional working capacity in storage is needed, and
that this additional storage will be concentrated along the Gulf coast as well as the
Midwest and Northeast, and that 80% will be in the form of depleted reservoir storage
and the remaining 20% in the form of salt cavern storage, then this yields a price for
storage construction of roughly $5b USD.
However, there is also the cost of purchasing base gas. If we assume that the depleted
reservoir storage facilities will require 50% base gas, and the salt cavern facilities will
require 25% base gas, then to have 750 Bef of working gas it will be necessary to have an
additional 650 Bef in base gas. If we take a natural gas spot price of about $6 per million
Btu (MMBtu) as the average price at which the storage gas could be purchased, then the
cost of the base gas is about $4 billion USD, and the cost of the working gas is about $4.6
billion USD.
The total cost, then, for constructing the storage facilities and filling them with gas should
be close to $14 billion USD.
How Long Would it Take to Build and Fill?
The amount of time it takes to construct a storage facility depends on specific site. Key
factors are the type of surface facilities needed, the proximity to pipeline infrastructure,
and permitting and environmental issues.
If we assume that the storage fields constructed would have the same average size as
existing fields, then this means that roughly 30 facilities would need to be constructed.
(This is so because there are currently about 400 facilities that can store about 8000 Bcf
of natural gas — and if we want to add about 1400 Bef of gas in storage, this would
roughly equal 30 facilities using the same ratio). It is likely that the first facility would be
finished in about two years, and that the last facility would be finished in about five
years.
Filling the completed storage facilities with natural gas is another matter. The North
American natural gas network is essentially a closed system at present, since only 2% of
the continent’s natural gas supply is imported. Moreover, North American natural gas
production has reached a plateau, and it is difficult to see how the level of production
could be increased. Since the amount of production and imports are essentially fixed, the
extra natural gas demanded by a new strategic stockpile would have to come from the
reduced consumption of others. In a market, this means that prices would have to
increase in order to allocate the same level of supply to a greater level of demand.
The very act of creating a strategic natural gas reserve, depending on how quickly the
reservoirs were to be filled, would therefore act to increase natural gas prices.
The same issues are at work in making deposits to the SPR. In April 2006, President
Bush suspended the replenishment of the SPR, which released oil in the aftermath of
Hurricane Katrina in September 2005, so as to minimize the impact of the additional
demand on the world oil price (McKinnon, Fialka et al. 2006).
In order to minimize the impact on natural gas prices, it would make sense to stock the
reserve with gas when natural gas volumes in storage are higher than average, and to
refrain from stocking it when volumes in storage are lower than average. If we assume
that about 350 Bef of extra demand per year could be accommodated without undue
pressure on gas prices, then this would mean it would take four years to completely fill
the NGSR.
Taking into account the time for construction, this would mean that it would take about
three years for the first storage facilities, and about six years for the last facilities, to be
built and filled with gas.
Evaluating Based on Tomorrow’s Infrastructure
When evaluating the need for an NSGR, we should evaluate not how it would interact
with today’s natural gas infrastructure, but with the infrastructure of five years from now
(as it will take this long to get the new facilities built and operational). While it is not
easy to predict how the system will look in five years, we can make some educated
guesses about the basics of the system. North American production should be about the
same as today, as new sources are being found at about the same rate old ones are
depleting. New LNG terminals will go into service, and exporting countries will develop
more LNG exporting capacity. At the same time, more natural gas-fired power plants
will be built. In short, demand should be higher, domestic supplies should be similar, and
the balance will be made up with imports.
In the distant future, when natural gas imports comprise a large fraction of North
America’s consumption, then an NGSR may make sense for the same reasons that the
SPR makes sense today. In five years’ time, it is unlikely that the fraction of imported
natural gas in North America will exceed 5%. By 2030, that fraction is projected to be
about 16% (EIA 2006). Given that the US has only about 3% of worldwide natural gas
reserves, the percent of imports will most likely continue to increase over time.
Conclusions
The 2005 hurricane season had a significant impact on the US’ energy infrastructure, of
which the natural gas system is a vital component. The recent hurricanes lead us to
question how vulnerable the sector is to other types of natural disasters, and whether
anything can or should be done to mitigate the effects of a future disaster.
Hurricanes Katrina and Rita caused about 800 Bef of natural gas to be shut in, which is
about 22% of annual Gulf production, or almost 4% of annual US natural gas
consumption. Natural gas prices skyrocketed to over $15.00 per million Btu (MMBtu) in
the aftermath of the hurricanes, and residential consumers paid record prices for natural
gas heating in the winter. However, the market was successful in allocating the reduced
production among consumers, and no shortages developed. As a result of both high gas
prices and a mild winter, natural gas volumes in storage stood at a record high in the
summer of 2006.
Modeling Scenarios: Hurricanes and a NGSR
As the 2005 hurricane season was the worst to date for Gulf oil and natural gas
production, we believe it reasonable to assume that the worst hurricane season we can
reasonably expect in the future will have an impact on production similar to the 2005
hurricane season. It is possible, however, to have a couple such seasons back-to-back,
followed not by mild but severe winters.
After building a model to simulate the US natural gas system in aggregate, we simulated
a scenario of a hurricane season similar to the 2005 season, followed by a colder than
normal winter, and then a repeat of the hurricane season and cold winter the following
year. We found, given our assumptions on the elasticity of demand, that the market is
successful in dealing with this scenario. In this scenario, the price increases following the
disruptions cause more conservation of gas than necessary to offset the disruptions;
leading to higher levels of storage than before the disruptions.
Adding a NGSR in the scenario helps to keep prices low, but results in lower overall
storage levels going into the next year — since prices stayed low and there was no reason
for consumers to conserve. In the second hurricane season, the NGSR is able to help
only a small amount, and then is exhausted, since it was not refilled between seasons.
The Role of a Natural Gas Strategic Reserve
We have calculated that creating strategic reserves of 750 Bef would require construction
of about 30 storage facilities, at the cost of about $14 billion (total construction cost
including base gas and working gas).
Even though having the reserves would likely dampen price increases following a
production disruption, the reserve would have to be refilled after it was used, which
would in turn put pressure on natural gas prices. Given that the North American natural
gas system is essentially a closed system, gas to refill the reserve currently could not
come from abroad, and could not come from excess domestic production (as excess
production capacity does not exist), but could only come from reduced consumption by
other consumers.
Moreover, the reasons for the establishment of the Strategic Petroleum Reserve (SPR)
should be taken into account. It was considered that having the reserve would discourage
other countries from using oil embargoes as an economic weapon, and that having a
reserve would allow the US the time to wait out or resolve the crisis in a position of
strength. Neither of these reasons applies to a potential NGSR, since as of 2005, North
America imports only about 2% of the total volume of natural gas consumed.
In short, if keeping natural gas prices low in the aftermath of large, sudden production
decreases is a key objective, then an NGSR can be considered. If the more stringent
criterion of avoiding natural gas shortages is applied, then the justification for an NGSR
is questionable.
References
Bamberger (2006). The Strategic Petroleum Reserve: History, Perspectives, and Issues,
Congressional Research Service.
Bernanke (2006). Remarks by Chairman Ben S. Bernanke Before the Economic Club of
Chicago, Chicago, Illinois. Wall Street Journal.
Cummins, C. and R. Gold (2006). In Katrina's Wake, US Oil Crossroads Remains
Vulnerable. Wall Street Journal.
EIA (2004). The Basics of Underground Natural Gas Storage.
EIA (2005). Annual Energy Outlook 2005. DOE.
EIA. (2005, September 6, 2005). "Natural Gas Gross Withdrawals and Production." from
http://tonto.eia.doe.gov/dnav/ng/ng prod sum _dcu NUS a.htm.
EIA (2006). Annual Energy Outlook 2006, EIA.
FERC (2004). Current State of and Issues Concerning Underground Natural Gas Storage,
FERC Staff Report.
FERC (2006). Major Storage Projects on the Horizon as of May 2006.
Gaul, D. and L. Alic (2005). US Natural Gas Imports and Exports: 2004, EIA, Office of
Oil and Gas, Natural Gas division.
Gresham, W. (2002). Natural Gas Consumption Trends and Price Elasticity. Presentation
to Southern Gas Association Gas Forecaster's Forum.
McKinnon, J. D., J. Fialka, et al. (2006). Bush Takes Steps to Expand Oil Supplies. The
Wall Street Journal. New York.
President Bush announced he would suspend deposits to the SPR to boost
consumer oil supplies.
MMS (2006). Press Release of June 21, 2006, Minerals Management Service.
Tobin, J. (2001). Natural Gas Transportation - Infrastructure Issues and Operational
Trends, EIA.
