Coal, Nuclear

Falling Fuel Inventory Levels Represent Growing Operational Risk

Issue 4 and Volume 106.

By: Stan Kaplan,
PA Consulting Group

While controversy rages in the generation business over the multitude of deregulation issues pounding on management’s front door, the back door has been left open for the financial and reliability wolves. The back door is the backstop of power plant fuel stockpiles – a stunningly mundane topic, until the fuel inventory is needed to keep the lights on. Many asset owners have stripped the inventory cupboard bare, exposing themselves, and the public, to significant cost and operating risks. The miscalculation is due to a combination of ill-considered cost cutting efforts and a reliance on historical supply patterns that may have little relevance to the future.

The Incredible Shrinking Stockpile

Generators maintain fuel stockpiles to buffer routine surges and shortfalls in fuel deliveries, and to provide for emergency fuel if normal supplies are interrupted. Coal and some oil-fired plants maintain fuel reserves for both reasons. Gas-fired plants keep oil on-hand for emergencies.

Click here to enlarge image

In the 1950s and 1960s, periods of stable fuel supply and prices, the average coal plant and oil/gas plant fuel inventory declined (Figure 1). At coal plants the average inventory dropped from 98 days of full load burn at the beginning of the 1950s to 49 days by the end of the 1960s, a 50 percent decline. The drop in oil inventories at oil and gas plants was similar, from 14 days to 8 days of full load burn.

Fuel markets in the 1970s were far from stable. Generation owners faced erratic coal supplies caused by several factors, including regulatory and labor relations issues and a surge in demand. Natural gas, suffering from the consequences of federal price controls, was often scarce and boiler fuel supplies were frequently interrupted to meet higher priority demand.

In response to these volatile and uncertain markets, plant owners increased inventories. By the end of the 1970s and the first half of the 1980s, utilities had inventories equivalent to almost 60 days of full load burn for coal and almost two weeks of full load burn for oil. But by the mid-1980s, utility management views of stockpile requirements changed dramatically. Fuel supplies stabilized and memories of the 1970s faded, particularly at the regulatory commissions. Emboldened by multiple nuclear plant and coal contract fiascos, the commissions aggressively hacked away at utility costs, particularly costs tied to inventory levels.

The working capital tied-up in fuel inventories became a prime target, not because of the amount of money involved, which was trivial, but because it was easy to understand. Unlike complex accounting or technical issues, anyone could grasp the concept of days of burn, and make simple comparisons of stockpile size with supply interruption history.

Utilities effectively made their oil and gas inventories sitting ducks by having no rationale for the size of the inventory other than rules of thumb or simplistic comparisons to peer utilities (the lemming approach to management). The arguments for cutting inventories were no more sophisticated, but given the inclinations of the commissions, the cost-cutters won more often than not. By the latter half of the 1980s, coal inventories shrank to under 50 days of burn and oil inventories to eight days.

In the 1990s, with the quickening pace of deregulation and the sale of generating plants to non-utility generators, cost cutting became priority one. Once again, because inventories represented a cost category easy to grasp, and because fuel supplies continued to be stable, inventories were targeted. Further, by converting fuel inventory into kWh, the owners of the plants could convert money tied up in working capital into cash.

By the late 1990s, the average coal stockpile fell to about 35 days of burn. But the really radical change was on the gas side, where oil inventories (for gas and oil plants) plunged to four days of burn. Developers opted to pursue simple-cycle and combined-cycle plants with no fuel oil inventory, and without even the technical capability to burn oil. In some locales this occurred because environmental limits made it expensive and difficult, or impractical, to permit operation on oil. But even where oil could be permitted, oil-burning capability was not installed. The following logic prevailed: gas would always be available during the summer, and in the winter, as long as the buyer was willing to contract for firm supply and transportation, it could count on reliable supplies.

Faith-Based Planning

The philosophy that smaller is better for fuel inventories needs to be reconsidered. On a common-sense basis, the radical cutback – or in the case of gas, actual elimination – of fuel inventories seems difficult to justify. Pipelines freeze, break and sometimes even explode. Railroads become congested and coal mines suffer force majeure mechanical failures or coal quality problems. Consider that the cutbacks and elimination of fuel inventories has taken place at the same time that utility management is often willing to accept generation reserve margins much lower than in the past. The prevailing philosophy is that the old standard, a reserve margin of 20 percent of generating capacity, was excessive. Margins of 15 percent or less are increasingly acceptable to planners and executives.

Logically, if the reserve margin shrinks, it becomes more important to keep units available. Profitability, not to mention the public interest, is at stake. Yet the coal piles shrink and the oil inventories disappear.

It is true that through the 1990s gas supply has been reliable, but coal has had a very bumpy ride. During the past ten years there have been at least four major disruptions in coal transportation – gridlock following the Union Pacific (UP) merger with the Chicago North Western railroad, the UP “meltdown” following its infamous merger with Southern Pacific, the Midwestern floods of 1993-94, and the disorder following the Conrail breakup.

These events had utilities literally scrapping the bottom layers of stockpiled coal out of the dirt, and backing down coal generation in lieu of gas or purchased power. This should have been an expensive lesson, but it seems to have been widely ignored.

The Tale is in the Tail

The risks of a small inventory, and additional insight into why utilities have opted to cut or eliminate inventory, can be discerned from a sample quantitative analysis. The following case is for a hypothetical 500 MW combined-cycle generating station located in the southeastern U.S. The plant runs year round, with a summer peak and a secondary winter peak. The unit is assumed to rely exclusively on firm gas supply and transportation.

The key variables in this study are probabilistic values in a Monte Carlo analysis. The probability of a gas supply interruption is 5 percent – that is, the odds are for one interruption every 20 years – with a duration of one to eight days, and a curtailment percentage that can range between 25 percent and 100 percent. Monthly power prices are also treated as statistical distributions, based on historical experience in SERC. The model permits no more than one interruption annually, and power prices are never allowed to exceed $2,000/MWh.

In the Monte Carlo analysis the model is run for 5000 iterations, with values selected randomly in each iteration from the probability distributions for each variable. If an outage occurs during an iteration, the model calculates the cost of replacing the lost generation with purchased power vs. the cost of burning oil.

Table 1 shows the probability distribution for a typical run. In 98 percent of the iterations the result is zero – that is, there has been no interruption in gas service or, if an interruption occurred, purchased power was available off the grid and was cheaper than burning oil. The mean benefit of oil-burning capability is only $62,000. Since the annualized cost of adding oil-burning capability is about $500,000 ($3 million dollar investment recovered over 10 years at a 10 percent cost of money), the savings appear to equal only a small fraction of the cost.

Click here to enlarge image

However, Table 1 also illustrates that the distribution of results has a long tail of low probability, high expense results. The probabilities are very small – less than one percent of occurrence for all the events shown combined – but the consequences can be enormous if a long interruption coincides with a period when power prices are high. The maximum savings from oil-burning capability is $63 million. Put differently, the operator of this plant has a small but measurable chance of taking a $63 million hit if it does not have oil-backup on the combined cycle.

The Tyranny of History

But are the probabilities really this small? The choice of a mere 5 percent probability of an interruption reflects the reliability of the gas markets since the 1970s. But the implicit assumption here is that the past tells us a lot about the future. In fact, it may not.

In 1987, the stock market plunged by 23 percent on a single day. A retrospective analysis of the stock market’s historical volatility prior to “Black Monday” (October 19, 1987), concluded that “had the market been open every day since the creation of the universe, the odds would still have been against its falling that much on any single day.”1 Similarly, an oil buyer contemplating long-term prospects in 1969 could have looked back over decades of historical experience, and found no statistical basis to plan for the price shock of 1973 (Figure 2).

Click here to enlarge image

Stuff does happen, and one of the places it happens is in the highly dynamic energy markets. The energy markets are constantly changing in respect to demand and supply, end-user and producer technology, and political and regulatory factors, just to name a few considerations. This dynamism makes the predictive value of historical experience questionable at best.

Given this uncertainty, the point of the combined-cycle analysis presented above is not that the probability of a high cost event is minute and can therefore be ignored. The point is that the lack of oil storage could be disastrously expensive and that if the cost of amelioration is modest, just a few million dollars for oil-burning capability, the investment should be made given the uncertainties about what the future holds.

The Roulette Wheel is Spinning

Which brings us to the events of the past year. The coal market has seen an extraordinary run-up in prices (the price of Powder River Basin coal almost tripled for a time) accompanied by supply shortages. Gas supply has remained reliable, but the price of gas has ridden an extraordinary roller coaster, up to $9.00/MMBtu and back down to $2.00/MMBtu in a matter of months. This experience with gas prices surely must say something about the overall stability of the gas market, in respect to supply as well as price.

Had any industry analyst predicted these price and supply developments 18 months ago, he would have put his job at risk. It is also significant that there is no historical analog for these events in the post-war history of the coal or gas businesses.

And finally, terrorism must be considered. If even one nuclear containment is breached, dozens of nuclear stations could be shut down for months or years to enhance security. In such a case the demand for coal and gas generation will skyrocket, supplies will become tight, transport systems will be over-stretched – and the contribution of fuel inventories to the reliability of the electric grid critical.

For 15 years the generation business has been skimping on the security of adequate fuel inventories. At gas-fired stations with no backup fuel, it is as though there is no guard at the gate. The industry needs to re-evaluate its fuel planning, before its leaders end up as star attractions at Congressional hearings about why the lights went out.


  1. Lowenstein, Roger, When Genius Failed: The Rise and Fall of Long-Term Capital Management, p. 72, citing Jens Jackwerth and Mark Rubenstein, “Recovering Probability Distributions from Option Prices,” The Journal of Finance, Vol. 51, No. 5 (December 1996), 1612.


Stan Kaplan is a Managing Consultant in the Washington, D.C. office of PA Consulting Group. He has 24 years’ experience in the generation business, as a consultant, regulator and utility fuel manager. He can be reached at 202-442-2553 or [email protected]