Sunday, April 6, 2014

Revisiting Texas's Electrical Power Predicament—Part 3

I hadn't intended there to be a part 3—but I would like to share three interesting pieces of news that nicely fit in with some themes of this series, and, I have a bit of free time on my hands this weekend to write about it.

The Bankruptcy and Restructuring/Breakup of Energy Future Holdings

Energy Future Holdings, Texas's largest power provider, came into existence in 2007 with a $45-48 billion leveraged buyout of TXU Corporation, as brokered by the likes of KKR, TPG and Goldman Sacks.  Overvaluation of TXU's assets, especially in the face of the subsequent economic depression of 2008 and continued low energy prices (thanks to the fracking boom), has made it impossible for EFH to repay the increasing debt burden associated with the LBO, and now, bankruptcy looks inevitable (see, Greed doomed the TXU buyout).  The recent departure of KKR CEO, Marc Lipschultz, for the Board of Directors of EFH probably signals a likely Chapter 11 restructuring of EFH's assets.

EFH is the parent of three subsidiaries: Oncor, which handles electricity transmission to 3 million customers, TXU energy, the states' largest electricity retailer, and Luminant, which generates about 18 percent of Texas' electricity, at least in the summer months.  Both Oncor and TXU energy look economically viable, and therefore, are attractive subsidiaries to buy. 

As I wrote in Part 2, Luminant owns five of those old coal-fired electricity generating plants, some of which have been getting mothballed during the winter month, and, which are too unprofitable, at least a today's energy prices,  to retrofit to comply with the Clean Air Act requirements.  Those five plants have a name-plate capacity of about 8000 MW which correspond to about 11-12 percent of Texas power generation capacity at any given time during the summer months.

There is speculation (In Energy Future Holdings shakeup, size matters) that a restructuring would go easier if the coal-fired plants were to be split off from Luminant's nuclear and natural gas, and, further speculation, that as part of a restructuring, those coal fired plants will not just get mothballed, but torn down.  I wonder who would want to buy Luminant if the maintenance or retrofitting these five plants was part of the deal (even a tear down of these plants would be very expensive, I think).

If the closure of these five plants were to happen, however, then ERCOT's summer power reserve margins would get blown away.  For instance, a loss of 8000 MW of power generating capacity would drop ERCOT's previously expected reserve margin from 13.6%, for the summer of 2014, to only 1-2%.  Under this scenario, I would expect that that there would be multiple rolling blackouts, and a spike in electricity prices, this summer.

Does the fate of Texas' coal plants and potential permanent loss of 8000MW of power have ERCOT worried or sounding the alarm?   Why no, quite the opposite.

ERCOT'S Revised Load and Reserve Margin Forecast
In a dramatic turn-around from its previous forecast of May 2013, Capacity-Demand-Reverse Report for 2013, in January 2014, ERCOT announced peak demand margins going forwards that implies that the possibility of rolling blackouts is remote.    There is no electrical power predicament after all.  

What happened?

Instead of an annual growth rate in peak electricity of 2-3 percent/year, as predicted by ERCOT last May 2013, ERCOT's new forecast predicts 1.3±0.1 percent/year demand growth for the next decade.  This results in a dramatic increase Texas's expected reserve margin for the next several years. 

What is the reason for this abrupt about-face?  ERCOT explains:

In the past, ERCOT has used weather and economic indicators, such as non-farm employment, to forecast future electric demand. However, the relationship between economic growth and electric demand has changed in recent years. While peak demand growth has slowed to about 1 percent annually, the economic forecasts and non-farm employment statistics used in recent load forecasts have resulted in growth forecast estimates of 2 to 3 percent in the two- to three-year outlook.

This recent trend implies a less direct correlation between these economic indicators and electric demand than in the past. To address this decoupling, ERCOT staff has developed a new load forecasting model that uses forecasted growth rates in customer accounts, or premises, to project future growth trends in each region served by the ERCOT grid.

In short, ERCOT's new model assumes a decoupling between electricity demand growth and economic indicators, such as employment, and instead now looks at growth rates in customer accounts or premises to predict future electricity demand.  

In a another publication describing this new method of future load forecasting (based on a neural network model) ERCOT presented a figure which I think sums up their reason for adopting a new model (Figure 5):

As you can see, starting from the mid 1990s, there has been an increased decoupling between electricity demand and GDP.  ERCOT attributed this to things like increasing efficiency of electrical devices (think light bulbs), declining energy use per customer and improvement in electricity distribution.  But maybe there are other reasons as well.

Based on this new model, at the end of February 2014, ERCOT came out with its latest Capacity Demand and Reserves Report (CDR)—the report being released four months earlier than previous in years.  In particular, a new Summer Summary prediction for demand, capacity and margins, reproduced in Figure 6 below, reflects the new margin and peak demand estimates.

The result of the new model is a prediction of much higher reserve margins than previously forecast.  For instance, the reserve margin (red) is predicted to actually go up to 15.4% in 2015, and, it doesn't drop below 13% until after 2018.  In contrast, the May 2013 CDR (Figure 4) had predicted the 2018 margin to be 9.4%.  Notable, 13.75% has long been considered an acceptable target reserve margin by ERCOT.

With peak demand growth assumed to be 1.3 percent/year, the Feb 2014 CDR predict a peak power demand (blue) of 71806 MW by 2018, which is ~4400 MW lower than the peak power demand of 76186 MW by 2018, predicted in the May 2013 CDR.  

Moreover, this increased margin is in spite of the 2018 total power resources being lowered from 80760 MW, as predicted in the May 2013 CDR, to a predicted 79575 MW in the Feb 2014 CDR.

Brattle Group Report (BGR)
Commissioned by the Public Utility Commission of Texas (PUCT), this report published at the end of January 2014 attempts to estimate the "economically optimal reserve margin for ERCOT’s wholesale electric market."

The bottom line of the BGR is that, from an economic perspective, a reserve margin of 13.75% is "inefficient" because it is designed to meet a one-day-in-10-year load loss event (0.1 LOLE)—a standard to ensure loss of power, on average, only 1 day in ten years.  Instead, the BGR recommends a new reliability standard that would be more "economically optimal."  This new standard, termed, "expected unserved energy" (EUE) standard, weighs the "total system" cost of building more power generation plants to keep the reserve margin high, versus the costs related to a "scarcity-event."  The BGR finds that only a 9-11% reserve margin is economically optimal.  The report estimates that a 11.5% reserve margin, would be insufficient to meet peak loads only once every three years.  On average, a 1600 MW short fall would only need to be curtailed for 2.6 hours on average during such "load shedding" event.

Interestingly, this report was shortly followed by a memorandum to ERCOT from PUCT commissioner Anderson questioning the need for PUCT or ERCOT to take any quick action based on this report:

If the Commission were to adopt a mandatory reserve margin, what Brattle refers to as a “reliability based standard,” they recommend the adoption of the “normalized expected unserved energy” (EUE)6 standard because it considers the magnitude and duration of events among its factors.7 Adopting a EUE approach would constitute a total redesign of ERCOT’s reliability standard, which should not be done without careful evaluation by the Commission and the ERCOT stakeholders. Analyzing the EUE recommendation should be conducted as part of a truly broad and intensive study of the appropriateness of ERCOT’s reliability standard and resulting reserve margin.

The BGR's assertion that a 10 reserve margin is contrary to the North American Reliability Corporation’s estimate of an acceptable reserve margin of 15 percent—a margin that ERCOT has not been able to attain in the past several years.

Summary and Thoughts

I don't think that it a coincidence that the early publication of ERCOT's Capacity Demand and Reserves Report, and, the publication of Brattle Group Report are being released just as Texas's largest electricity provider, Energy Future Holdings, heads into bankruptcy. 

The new Capacity Demand and Reserves Report suggests much a lower rate of demand growth increase, and therefore Texas, can have "acceptable" reserve margins of 13.75%  for the next several years.  Brattle Group Report goes one further in suggesting that this 13.75% margin is way too high and could instead be 10-11% for "economic optimization." 

Let's just set aside the apparent paradox of ERCOT adopting a new method of predicting demand growth that specifically does not directly consider economic indicators, and, the BGR that specifically recommends lower reserve margins based on "economic optimization."  

To me these two reports may be setting the stage for a restructuring of Energy Future Holdings where those five coal-fired plants are rapidly phased out of ERCOT's power grid.  The ability to eliminate these uneconomical  coal-fired plants would make the purchase Luminant more attractive. 

Still, the loss of 8000 MW of potential power provided by the five plants reduces ERCOT's summer reserve margin to only 1-2 % and that would be a problem. How can we reconcile this?  

It seems like a no-brainer to say that, for at least the near future, at least some of these coal-fired plants have to be maintained and de-mothballed each summer in order to provide an acceptable margin of peak demand.  

But, perhaps now in light of the new CDR and BGR, it will seem more acceptable for some plants to be decommissioned immediately or next year. 

For example, consider the new CDR shown in Figure 6.  According to ERCOT, for 2014, summer resources equals 74805 MW and firm load equals 66179 MW, giving a 13.5% margin.   If one were to eliminate one or two coal-fired plants with a total capacity of about 2000 MW, then resources drops to 72805 MW, and therefore, the new margin becomes 10% which just happens to equal BGR's "economic optimum."  Likewise for 2015, if coal-fired plants with a total capacity of about 4000 MW are eliminated, then the margin again equals 10%.    

In my opinion, not running one to four of these coal fired plants during the summer months would obviously increase the risk of load shedding or rolling blackout events.  Texas only needs to have a summer as hot as the summer of 2011 to test this.  Perhaps from the Battle Group Report author's perspective, cozy in their offices in Cambridge, Massachusetts, the prospects of industrial and residential load shedding and rolling blackouts for several days in August fall within the scope of "economic optimization."   After all, this is estimated to only be for 2-3 hours per day and only occur every 3-4 years.  

I seriously doubt, however, that that the "total system"  economic repercussions of an electricity "scarcity-event" has been fully accounted for in the BGR.  For instance, will businesses still want to relocate to Texas when they hear news of several days of rolling blackouts or load shedding by  industries and residents in the summer months?  Also "economic optimization" does not necessarily mean optimal for human living.  People, especially the elderly, suffer or die in the summer in Texas when the air-conditioning goes out.   Is that considered as part of the "total system" cost?   Maybe, from an economic perspective, who cares, so long as the state government, PUCT or ERCOT doesn't have to pay for it directly. 

And, what about looking further out to more than a few years?  Look again at Figure 6, for 2020 and beyond.  Even with Luminant's coal-fired plants considered as part of the summer resources and assuming only 1.3%/year peak demand growth, the reserve margin was predicted to drop below BGR's "economic optimum" of 10 percent.  What then? 

With the present stricter enforcement of the Clean Air Act and other EPA rules, the prospects building of new power plants that would provide significant base load (i.e., coal, nuclear or even natural gas) in time to mitigate the reserve margin drop look very remote. The price of electricity would have to rise significantly before that would happen, I expect.  But doubling or tripling electricity prices would cause demand to go down.  And, there you have a downwards energy spiral, Texas-style.

Perhaps targeting peak demand growth to customer account growth does make more sense than using employment or other economic indicators.  After all, an unemployed Texan probably moves out of their own residence and moves in with someone else, moves out of state, or, out on to the street.  Consequently, their utility provider account gets closed. Of course, such repercussions of scenarios are really not ERCOT's problem.  Again the this reflects a difference between an "economic optimum" versus a "human optimum."

From a broader perspective, the increasing decoupling between GDP and electricity use (Figure 5), to me, just represents the increasing decoupling of GDP numbers from reality.  That is, the increased churning of digital money between parties all counts towards GDP, which increasingly, has little to do with the real physical world or  the life of the average human.