Except under the lowest demand scenarios, natural gas inventories are on track to exit March below comparable year-ago and five-year-average levels, making shifts in weather all-important at a time of diminished gas-to-coal switching potential in the power stack, according to analysts.

A mild Christmas could assuage storage fears after hefty deficits helped prices break higher along the winter strip through the early part of the heating season. But significant upside risk remains heading into the new year.

Indeed, “the fate of the gas storage market is teetering on how cold or warm this winter will be, with the current deficit making the market especially susceptible to the upside,” RBN Energy LLC analyst Sheetal Nasta said.

As it stands, anything more than a milder winter could see the recent volatility spill over into the 2019 injection season. In a recent blog post, Nasta pegged end-March inventories near 1,130 Bcf, about 200 Bcf below the year-ago period and about 500 Bcf lower than the five-year average.

To reach the five-year average 1,630 Bcf by end-March, the market would need to limit withdrawals to just 11 Bcf/d for the December-March period, less than the 12 Bcf/d withdrawn during the same stretch in the mild 2015/16 winter, according to Nasta, who modeled several scenarios that could play out this winter, isolating for shifts in demand from the weather-sensitive residential/commercial, power burn and industrial sectors.

Starting from a baseline of 2,991 Bcf reported by the Energy Information Administration (EIA) for the week ended Nov. 30, this “implies the market would have to withdraw no more than 1,638 Bcf total (the equivalent of 13.7 Bcf/d) in the 120 days between Dec. 1, 2018, and March 30, 2019, in order to get to last year’s benchmark,” Nasta said. “Over the past five years, the lowest daily average withdrawal over that period was 12 Bcf/d in 2015-16, when the market experienced a relatively mild winter.

“But daily average withdrawals have been in the 15-20 Bcf/d range the past two winters. So, for this year’s inventories to withdraw no more than 11-14 Bcf/d would require the supply-demand balance to be considerably looser than the past two years,” she said. All of this means that “unless weather-related consumption comes in on the low end of recent years during what are typically the coldest months,” deficits will linger past March. “And an ongoing deficit past withdrawal season could extend price volatility well into 2019.”

Gas-to-Coal Cushion Threadbare

Meanwhile, after recent retirements, gas-to-coal switching may not be able to provide much of a cushion should a sustained cold stretch send natural gas prices racing higher come January or February.

After a certain point, higher natural gas prices won’t have much of an impact on the relationship between gas and coal in the power stack, with the impact approaching zero once Henry Hub prices approach $4.50-5.00, according to Genscape Inc. analyst Eric Fell. He said economic dispatch could swing as much as 8 Bcf of demand, but most of that occurs at prices below $3.

“The dynamics of gas versus coal generation in the power stack are quite complex,” Fell wrote recently during a discussion on Enelyst, an energy chat service hosted by The Desk. “Several factors drive changes in gas and coal generation from day-to-day with renewable/nuclear output, power demand, weather, power grid constraints, power plant outages and regional gas prices all playing a role in determining the amount of gas and coal dispatched on a given day.

“Because prices are only one of many factors that drive dispatch, quantifying gas versus coal generation as a function of price can look very noisy from day-to-day or week-to-week. However, by taking a step back and looking at many data points over time, one can observe some useful trends.

“At the national level, we see that coal generation has shifted lower permanently (and conversely gas has shifted higher),” Fell said. “However, the relative impact of prices on gas versus coal gen has remained comparable over time. The punchline is that gas prices have a non-linear impact on demand, with a very steep demand impact below $2.50 and a minimal impact above $5.00.”

Fell said regional dynamics can also introduce substantial variation into the relative economics of coal versus gas.

As an example, in ERCOT (aka, the Electric Reliability Council of Texas), “Genscape proprietary power plant monitor data from 2016-17 shows that coal generation ramps down hard when Houston Ship Channel prices fall below roughly $2.25, while above $2.25 the sensitivity is much smaller,” Fell said, referring to Daily GPI spot prices. “We also see ERCOT coal generation shifting lower in 2018 after the retirement of three large coal plants took more than 4 GW offline in the first quarter.”

Morningstar Commodities Research analyst Matthew Hong recently looked at gas versus coal economics this winter in the PJM Interconnection, where “a decade of retirements in coal generation and mine shut-ins in Central Appalachia have shrunk the safety net coal plants used to provide.”

Hong estimated that based on recent prices for eastern coal, PJM generators could expect to pay a delivered price of about $3.81/MMBtu equivalent, below the Texas Eastern M-3, Delivery day-ahead price during the second half of November and first half of December.

This comes after gains in winter CSX coal prices (January-March 2019) over the past year, climbing from $58.15/short ton during the summer to closer to $75/short ton in December, according to Hong.

“On paper, the cheaper coal price should be encouraging gas to coal switching,” the analyst said. “Yet we believe rising coal prices this year are more closely tied to increased exports rather than to any uptick in switching demand from domestic power generation. In part this is a consequence of the last decade of coal retirements that have decimated domestic markets, with the result being several producers exiting the space. That has left little capacity for domestic mines to rally production to service an unexpected reversal in coal generation economics.”

Noting broader downtrends in both demand for domestic coal and power plant inventories, Hong said recent PJM monthly average generation data showed coal generation declining year/year during November and early December as natural gas filled the void, even amid the stronger natural gas prices.

“The data offers no evidence of gas to coal switching even as coal economics have become competitive,” the analyst said. “We believe that this reality reflects generators’ uncertainty around short- and long-term forecasts that have yet to converge on a consistent theme for the rest of winter. That uncertainty challenges the need for generators to order additional coal” as the independent system operator “moves into the peak winter months unless unforeseen operational or weather events require them to act.

“This reluctance to restock even when economics are good suggests coal will play a shrinking role in meeting system load. Added to tepid demand from the declining coal generation fleet is the reluctance of producers to crank up output in the absence of long-term prospects. The high capital cost to develop additional mines forces producers to weigh opportunity costs associated with increased production.”

Playing Out The Scenarios

Looking at balances through the remainder of winter — assuming natural gas production, imports and exports remain flat to recent levels — RBN’s model estimates a 15.5 Bcf/d withdrawal pace from Nov. 30 through the week ended March 29. That scenario accounts for recent 15-day forecasts and assumes 10-year average temperatures from late December until end-March.

“That pace is enough to reduce the deficit versus last year or the five-year average, but not eliminate it,” Nasta said. “In fact, it would leave storage at the end of March 2019 more than 200 Bcf shy of March 2018 and about 500 Bcf short of the five-year average for that time of year, though higher than the 5-year low set in 2014.”

A high-demand scenario — based on the five-year high for domestic consumption — would see the market exit the heating season about 18 Bcf/d undersupplied, resulting in end-March inventories of 835 Bcf, according to Nasta. A scenario based on last year’s domestic consumption would result in total demand of 106 Bcf/d, about 15 Bcf/d undersupplied.

But illustrating the broad range of outcomes the market has had to consider, a low-demand scenario (assuming 93 Bcf/d of total demand) would leave the market just 2.5 Bcf/d short for a total 300 Bcf pull for the 120-day period from December through March.

What’s more, a three-year average demand scenario (including the milder 2015-16 and 2016-17 winters) would also see market erase storage deficits by end-March, coming in just 9 Bcf/d undersupplied based on total demand of 100 Bcf/d.

RBN’s model calculates production at around 86 Bcf/d, 8.0 Bcf/d higher year/year for the period. “Assuming no major disruptions to production, that is somewhat conservative, considering there’s at least another 800-900 MMcf/d of takeaway capacity due online from the Marcellus/Utica shale producing region before 2018 is over,” Nasta said.

On the demand side, RBN’s projections assume an average 4.5 Bcf/d of liquefied natural gas (LNG) exports for the balance of winter, reflecting incremental feedgas volumes for commissioning activities at Cheniere Energy Inc.’s Sabine Pass Train 5 and Corpus Christi Train 1.

“There is the potential for more upside to that as Sempra’s Cameron LNG reportedly has begun commissioning activities for its first 0.7-Bcf/d train as well, though we have yet to see feedgas flows to the facility,” according to Nasta. “Between these three trains, total LNG export capacity could reach 4.9 Bcf/d by the end of winter.”

Exports to Mexico will average 4.8 Bcf/d for the balance of winter, according to RBN’s analysis. The firm is modeling pipeline receipts of regasified LNG imports at 0.3 Bcf/d for the balance of winter (in line with the past three winters) and net imports from Canada at an average 4.5 Bcf/d, down from recent winters due to increased competition from the Marcellus/Utica and reduced deliverability across the border at Sumas, WA.