The Goldilocks Dilemma

An old posting about why intermittency is not a big deal came to my attention today.  I re-read some of what had been said, especially when I had just sent out a paper on the topic yesterday.

I believe that the value of electric “energy” is often overstated.  The author of the old posting, Chris Varrone, inadvertently acknowledges this when he wrote

However, the energy in wind is worth 100% of the energy in nuclear (or anything else) in the spot market; wind energy in the day-ahead market may be worth a little less, but this can be “firmed” using energy trading desks or by using other assets in the operator’s fleet.

If the day to day differential can be handled by firming with other assets, then the value of the electricity is not just energy.  It is not worth debating what to call this other value, but a substantial part of the value in the spot market is something other than energy.

As to The Goldilocks Dilemma, the paper I sent out yesterday, I began by asking

Is the price paid to dispatchable generation too high, too low, or just right for intermittent generation?

I then answer

Though intermittent generators often argue that they should receive the same price as dispatchable generation and some utilities argue that they should pay less to intermittent generators, sometimes intermittent generators should face a higher price than dispatchable generators, such as when intermittent generation is part of the market during instances of extreme shortage.

The entire paper is available on my web site, the companion to this blog site.  Look for the hot link to the library near the bottom of the first page.  A hot link for the article is near the bottom of library index in the section called drafts.

The Electric Transmission Grid and Economics

Tuesday, 2013 October 8, I went to the MIT Club of Washington Seminar Series dinner with Anjan Bose of Washington State University talking about Intelligent Control of the Grid.  Anjan began with giving two reasons for the transmission grid but then seemed to ignore the predicate in explaining what the government has been doing in regard to the grid.

The first slide identified two reasons for the electric transmission system.  The first was to move electricity from low cost areas (such as hydro-electric dams) to higher cost areas.  This is an obvious reference to economics.  The second was to improve reliability.  Anjan did not get into the discussion of how that is an economics issue, but it is.  Reliability is greatly improved by increasing the number of shafts connected to the grid.  We can produce the same amount of electricity with five 100 MW generator or one 500 MW generator.  The five units provide greater reliability but also higher costs.  The higher costs are associated  with various economies of scale, including higher installed cost per MW, less efficient conversion of the fuel into electricity, and the need for five sets of round the clock staffs.  A transmission system allows dozens of 500 MW units to be connected at geographically dispersed locations, achieving the reliability of many shafts and the lower cost of larger generators.

But, the presentation had little to do with the economics of the power grid, and the investigations into those economics.  I noticed that much of the discussion during the question and answer period did talk about the cost of operating the grid, so people were indeed interested in money.

Anjan said that the financial people used different models than did the engineers who operate the system.  I have long said that we need to price the flows of electricity in accord with the physics of the system, by pricing the unscheduled flows.  The engineers and operators may plan to operate the system in a prescribed way, but the flows of electricity follow the laws of physics, not necessarily the same was the way some people have planned.

Anjan said that deregulation[1] has caused a dramatic decline in new transmission lines, especially between regions such as into and out of Florida.  My feeling is that new transmission lines would be added more willingly if the owners of the new transmission lines would be paid for the flows that occur on the transmission lines.  For instance, twenty years ago a new high voltage transmission line in New Mexico began to carry much of the energy that had been flowing over the lower voltage transmission lines of another group of utilities.  The group of utilities called the service being provided “vampire wheeling” and refused to make any payment to the owner of the new transmission line.  The new line provided value in the reduced electrical line losses and perhaps allowed a greater movement of low cost power in New Mexico, but that value was not allowed to be monetized and charged.

I note that a pricing mechanism for the unscheduled flows of electricity would have provided a different mechanism to handle the 2011 blackout in Southern California, which began with a switching operating in Arizona.  Engineers swarmed to the area to find data to assess the root causes but were initially blocked by San Diego Gas & Electric’s attorneys who feared that any data could be used by FERC to levy fines pursuant to the 2005 electricity act.  I remember a discussion at the IEEE Energy Policy Committee on that proposed aspect of the bill.  The IEEE EPC voted to suggest creating mandatory reliability standards.  I was the sole dissenting vote, arguing that the better way was to set prices for the unscheduled flows of electricity.  Thus, SDG&E and the Arizona utilities would have been punished by the market instead of risking a FERC imposed fine.

[1] I prefer to use the more accurate term restructuring, since the entire industry is still regulated, even though generation is often subject to “light handed regulation” by FERC, which approves concepts instead of specific prices.

Storage/Pricing — Chicken/Egg

On Tuesday, 2012 November 27, I attended the Heritage Foundation’s discussion of Jonathan Lesser’s 2012 October paper “Let Wind Compete: End the Production Tax Credit.” The only philosophical statement on which there seemed to be agreement was that improved storage systems could improve the market for wind.

But who would own the storage systems necessary to make wind even more viable? Unless the ownership is in common with the wind systems, how would these storage systems be compensated?

  • And, can we expect entrepreneurs to build these storage systems and then expect FERC to set an appropriate price? Beacon Power produced a flywheel storage system but couldn’t get FERC approval of a tariff before it ran out of operating cash and is now bankrupt.
  • Or should FERC put into place a pricing mechanism that could compensate storage systems when they arrived on the scene? I look at this as the Field of Dreams mantra of “If you build it (a competitive market appropriate for storage systems), they (storage systems) will come.”

Truly, a chicken and egg issue.

Wind has been accused of having two failings. Wind often provides a lot of power at night, when electricity is not highly needed.  Wind provides less power on the hot mid-summer afternoon, when electricity is needed the most. This is an intra-day issue for storage to handle. Wind power also follows the wind speed. A wind gust can push power production up to great heights. A wind lull can suddenly drop power production. Storage could be useful for handling this intra-hour issue.

Not all storage can handle both the intra-day and the intra-hour issues well. For example, the storage part of the Heritage Foundation discussion mentioned only pumped storage hydro as a representative storage technology to help wind. Pumped storage hydro has been used for decades to transfer power from the nighttime and weekends to the midweek daytime periods. That is, pumped storage is known as a way to handle the intra-day issue. I like pumped storage. My first job after getting a Masters from MIT’s Sloan School was with American Electric Power which owned a pumped storage plant. This perhaps accounts for some of my bias of liking pumped storage hydro.  (Actually I like to have a variety of generation options available, not just pumped storage.) Pumped storage hydro is excellent for intra-day transfers of power.

I have never seen anyone use pumped storage hydro for intra-hour transfers of power, or even propose it for such purposes. The absence of a historical use of pumped storage to provide intra-hour storage doesn’t mean that pumped storage could not be used for that purpose. After all, many people tout pumped storage for its ability to respond in seconds to changes in the need for electricity.

Pumped storage is often touted as being about 75% efficient. For every 100 MWH used for pumping, 75 MWH can be subsequently generated. We can model the effect of shorter duty cycles by beginning with the assumption that 0.5 hours in the pumping mode is ineffective. Under this modeling assumption, for 13 hours of pumping, there is the equivalent storage of 12.5 hours. With the 75% efficiency assumption, the system can generate for 9.375 hours, for a revised efficiency of 72% (9.375/13). Reduce the pumping time to 5 hours will reduce the generating time to 3.375 hours and the revised efficiency to 67%. Reduce the pumping time to 1 hour will reduce the generating time to 0.375 hours and the efficiency to 37%. This is not a very good efficiency ratio but we normally don’t think of running pumped storage on an intra-hour basis. I don’t know that pumped storage can run with just one hour of pumping, just that trying to do so will be costly, indeed very costly.

The intra-hour situation has been handled by batteries, flywheels, magnetic storage devices, and theft of service. Theft of service is a harsh term. When an electric utility faces the intra-hour problem associated with rapid changes between wind gusts and wind lulls, the physics of the electric system results in inadvertent interchange, electricity moving into and out of the utility.  With the inadvertent interchange going both ways, which utility is providing a service to the other utility?

If the wind gust occurs first, the power is stored on a neighboring utility system. If the lull occurs first, the utility is borrowing electricity and then gives it back. There is no systematic payment mechanism associated with this storage or borrowing of electricity. It is a free service as I described over two decades ago in “Tie Riding Freeloaders–The True Impediment to Transmission Access,” Public Utilities Fortnightly, 1989 December 21.

Most of the currently operating pumped storage systems were put into place by vertically integrated utilities. AEP often looked at its coal fired generating system as providing cheap, efficient capacity, allowing AEP to make large sales to its neighboring utilities. But the pumped storage system also helped AEP with its minimum load issues. The large AEP generating units were very efficient. The investments made to achieve these efficiencies hampered the ability of generators to cycle down at night, during minimum load conditions. Pumped storage systems helped AEP with that situation. Now many pumped storage systems operate in advanced markets operated by ISOs/RTOs, where their value can be assessed based on their interaction with the advanced market.

The thought process of testing how a pumped storage system would operate on an intra-hour basis also provides some information about profitability issues. For 13 hours of pumping and 9.375 hours of generation requires the off-peak price to be less than 72% of the on-peak price to achieve breakeven revenues, that is revenue from the sales to be equal or exceed the payments for pumping energy. The off-peak price has to be even less for the pumped storage system to have book income, that is the ability to cover its investment and other operating costs. The shorter the operating period, the smaller the break-even off-peak price relative to the on-peak price. A competitive market for storage systems needs to have very low “off-peak” price relative to its “on-peak” prices.  In this context, off-peak price and on-peak prices could be better described as storage prices versus discharge prices.

The advanced markets have hourly pricing periods that are consistent with the dispatch periods of pumped storage.  But for rapid response storage, hourly energy prices do not provide any incentives for the storage system to operate on an intra-hourly basis.  Indeed, if storage systems are to operate on an intra-minute period, then prices need to be differentiated on an intra-minute basis, not just on an intra-hour basis.  Area Control Error (ACE) is an intra-minute utility metric that can be used to set an intra-minute price for storage systems that are expected to be operated on an intra-minute basis.  India has developed a very simplified pricing vector that uses ACE to set the price for Unscheduled Interchange on an intra-dispatch period basis.

In India, the regional system operators set hourly schedules for the utilities and for non-utility owned generators.  Though the schedules are hourly, the utilities and non-utility owned generators are nominally required to achieve an energy balance every 15 minutes.  Each 15 minute energy imbalance is cashed out using a pricing vector that indexes the price for all imbalances against system frequency.  In India, system frequency is the equivalent of ACE.

There are ongoing discussions in India about modifying the pricing vector to reflect the hourly settlement price, to expand the pricing vector for more extreme values of ACE, to geographically differentiate the price, etc.  Though there are discussions about revamping the pricing vector, the pricing vector concept has greatly improved the competitive system against which the utilities and non-utility owned generators have be operating.  The pricing vector concept could be used to price intra-dispatch period storage to provide the competitive market from which the storage systems could draw power and into which the storage systems could discharge power.

Utilities, including ISOs/RTOs, use ACE to determine dispatch signals for their generators.  ACE is calculated every three or four seconds using the frequency error on the network and the interchange being delivered inadvertently to other utilities on the network.  Generally, the convention is that a positive ACE means that the utility has a surplus, while a negative ACE means that the utility has a shortage.

  • A surplus means that the utility is giving away energy, not getting any money for the surplus energy.  Under the situation of a positive ACE, the utility will want its generators to reduce their generating levels and would want storage systems to store energy.  As demonstrated by the earlier thought experiment, the market price for unscheduled energy into the storage system would have to be low for the storage system to absorb the energy economically.  When the utility is giving the energy away and not getting any money for the giveaway then any price, even a low price, for the energy going into storage can be appropriate.
  • A shortage means that the utility is taking energy from its neighboring unities, without paying for the shortage.  This is one form of the theft of service I mentioned facetiously above.  Under the situation of a negative ACE, the utility will want its generators to increase their generating levels and would want storage systems to produce energy.  As demonstrated by the earlier thought experiment, the market price for unscheduled energy coming out of the storage system would have to be very high for the storage system to produce the energy economically.  When the utility is stealing energy, then any price, even a very high price, for the energy coming out of storage can be appropriate.

For an explanation of the Indian mechanism for pricing Unscheduled Interchange, I recommend “ABT – Availability Based Tarrif”,[1] a completion of postings on InPowerG, the Indian equivalent of IEEE’s PowerGlobe and “ABC of ABT: A Primer On Availability Tariff,”[2] written by Bhanu Bhushan, the developer of the Indian pricing vector concept.  For a discussion of advanced pricing vectors that could be used for pricing storage, see the papers on my web site,[3] especially those filed recently with FERC.

The advanced markets have prices for generators that respond to the dispatch programs in a rectangular manner. For instance, consider a 5 minute dispatch period.  The price does not differentiate between those generators that are ramping versus those that are constant or those that move up and down to counteract ACE excursions.  An intra-dispatch period price for generation excursions would reward those generators (and loads) that help with ACE excursions and charge those generators (and loads) that cause the ACE excursions.  A pricing plan that achieves such a concept would be worthwhile even before fast acting storage systems came on line.




Oil Storage

During the Arab oil embargo of 1973, some people speculated that the US had a strategic petroleum reserve in the form of gasoline sitting in the driveways of most suburban homes.  The speculation was that many people made a point to refill their gas tanks as soon as ¼ of the tank had been consumed.  At that rate, the average amount of gasoline in this mobile storage was 7/8 of the tank.  By some calculations that was the equivalent of a month’s usage of gasoline.  Whether the mobile storage was indeed the equivalent of a month’s usage of gasoline or was much less, the storage capability was quite large.


During the question and answer period after Adam Sieminski, Administrator of the US Energy Information Administration talked to the National Capital Area Chapter of the US Association for Energy Economics on 2012 October 19 on the EIA Winter Fuel Outlet, I asked Adam about the possibility of oil supply interruptions in the Northeast, which is the area most heavily dependent on residential heating oil.  I included in my question a reference to the gasoline shortage in California that had pushed gasoline prices there more the 50 cents a gallon above the national average.


Part of Adam’s response included a discussion of the high elasticity of demand for gasoline, the time it took for tankers to move gasoline from other parts of the country, or from overseas, and that historically such price spikes lasted about five or six days, much less than the fourteen days necessary to ship gasoline the requisite distance.  Later I wondered about my above musings, about the mobile inventory of gasoline.


Do people respond to gasoline price spikes by a partial depletion of their individual mobile inventory?  Does the average gas tank level drop from 7/8 to ¾ to ½, or even lower, by only having partial fill-ups?   After all, some newspaper articles included interviews of workers who changed their fueling practices in include partial fill-ups.


How could we estimate the size of the partial drawdown of this mobile strategic petroleum reserve?  Or even the size of the mobile strategic petroleum reserve before the drawdown?  Does EIA have sufficiently fine data to make these estimates?  And how would a drawdown of this mobile reserve effect the elasticity estimates that Adam identified?

Ramping–Wind Data from Kodiak, Alaska

A growing concern about renewable resources, such as wind and solar, is that they can ramp down and then back up in a few seconds.  The requirement that electric utilities balance their sources and uses of electricity on a real time basis means that the utility must incur a cost by contra-cyclically ramping up and then down other sources of electricity, whether the other source is generation, load control, or a storage unit.

Determining the cost of the countervailing generation is an accounting nightmare.  An alternative approach is to set a dynamic transfer price, where the dynamic pricing mechanism reflects the degree of imbalance on the network.  A large shortage should result in a high price.  A large surplus should result in a low price.  I first wrote publicly about a dynamic pricing mechanism in “Tie Riding Freeloaders–The True Impediment to Transmission Access,” Public Utilities Fortnightly, 1989 December 21, and recently wrote again about the mechanism in regard to “Pricing Unscheduled Ramping,” released 2011 September 15.  The latter is available on my web site,

Chugach Electric Association (CEA) is planning a 17.2 MW wind farm just outside Anchorage, Alaska.  CEA is interconnected with Anchorage Municipal Light & Power (MLP).  MLP is concerned about the ramping of the wind farm, since the ramping will jerk around the MLP system.  MLP obtained second by second wind generation data from Kodiak Electric Association (KEA) for the 4.5 MW wind farm on the KEA system for October 2010.  KEA operates asynchronously to CEA and MLP but it is one system in Alaska with a wind farm and thus with data about wind farm operations.

During the 2,678,400 seconds that month, the KEA wind farm averaged 1,544 KW of generation.  The wind generators had auxiliary power needs, such that during 535,798 seconds (20.00% of the time), the power flow was negative, that is, the auxiliaries were using more power than the generators were producing, averaging 34 KW of net flow from KEA.  During another 6,852 seconds (0.26% of the time), the generation was zero.  For the 2,135,750 seconds when there was net power flow from the generators, the average net generation was 1,944 KW.  The median value is 988.3 KW, with half of the values being greater than or equal to 988.3 KW and half of the values being less than or equal to 988.3.

I used Excel to count the number of seconds during which the wind farm was within specified blocks.  The blocks were 100 KW wide.  The block containing the most seconds was for the range when the flow was negative, between -100 KW and 0 KW.  The next highest count was for the interval between 4,500 KW and 4,600 KW, roughly the capacity of the wind farm.

In “Pricing Unscheduled Ramping” I present graph of the Excel counts, including a presentation of the mean and median values.  The distribution has its maximum value for the 100 KW of negative value and for the interval between 4,500 KW and 4,600 KW.  This second highest count is roughly the capacity of the wind farm.  In “Pricing Unscheduled Ramping” I also present a cumulative distribution of the number of seconds during the month by the net generation during those seconds, including a presentation of the mean and median values.

Since I was concerned about the amount of ramping that the wind farm was imposing on the system, I then calculated the second to second change in power levels.  The maximum one‑second drop in power generation was 646.1 KW.  The maximum one second jump in power generation was 303.6 KW.  During 1,361,692 one second intervals (50.84% of the intervals), there was no change in the power level of the wind farm.  So, despite some large one second ramps that KEA experienced with its wind generation, most of the time (50.84% of the intervals) the wind farm was absolutely stable with no ramping at all.

Another measure of ramping is the summation of the ups and the downs.  Looking at just the instances when the wind farm ramped up, the amount of ramping was 8,351,700.90 KW.  Assuming a capacity of 4,500 KW, the wind farm during the month of October ramped the equivalent of its full load 1,856 times, or 2.5 times each hour.  Thus, on average, every 24 minutes the wind farms ramped the equivalent of going from zero to full load and back to zero.  Few fossil fired generators would be able to last very long if they had to react to a duty cycle of 2.5 times full load each hour.  Flywheels and batteries are likely to be the only devices that can react to the need for such a duty cycle.

In “Pricing Unscheduled Ramping” I present a cumulative distribution of the number of one second intervals during the month by the net generation ramp during those seconds.  As is apparent from the above discussion, the cumulative distribution had a large jump at a change of 0 KW.

FERC seems to be enamored with the way that Bonneville Power Authority (BPA) charges penalties for imbalances.  Under the BPA approach, the penalty price depends on the amount that the generator is out of balance, the greater the imbalance, the greater the unit charge for the penalty.  The pricing plan in “Pricing Unscheduled Ramping,” out of necessity, presents such a punitive pricing plan for ramping. 

I presented a non-punitive plan for pricing imbalances in “Reply Comments Of Mark B. Lively In Regard To Using Prices Instead Of Penalties For (1) Regulation And Frequency Response, (2) Energy Imbalance, (3) Generator Imbalance, And (4) Inadvertent Energy,” Preventing Undue Discrimination and Preference in Transmission Services, FERC Docket No. RM05-25-000 and RM05-17-000, 2006 September 20.

A non-punitive plan for pricing generation ramping (and generation imbalances) rewards those imbalances that are in synch with the ramping needs of the grid as a whole.  Thus, when the wind generators ramp up while the grid is ramping up, the wind generators would be rewarded for that ramp.   Conversely, when the wind generator is ramping down while the system is trying to ramp up to meet a ramp up in load, then the wind generator should be penalized.

For a more complete discussion of the non-punitive pricing for unscheduled flows of electricity see “Tie Riding Freeloaders”, “Pricing Unscheduled Ramping”, or my reply comments in FERC Docket RM05-17-000.

Load Profiles and Unscheduled Flows of Electricity

In response to my earlier comments about a Fox News article equating decoupling with socialism, I received a question about time of use rate making and how my comments regarding unscheduled flows of electricity picked on small utilities. My response follows.

Most utilities group customers by their consumption patterns. You are grouped with McMansions and with hovels. Not because you have the same consumption, but the same pattern. Let’s say that a McMansion uses 10 times the energy that you use each month. Then, if you look at the consumption for the period defined as “on peak”, the McMansion will use about 10 times the amount of electricity that you use “on peak”. Similarly during any other well defined period. TOU pricing will not differentiate your average price from his average price, or at least not much.

Not much unless you or the utility invest in devices that control your consumption patterns. Thirty five years ago I worked for American Electric Power (AEP) in its New York City office. I helped investigate ceramic storage, a concept then in use in England. Turn on electricity to these “hot rocks” at night and then during the day just blow a fan across them to get heat into the room. Conversely, there is ice storage. Connect your A/C unit to a tank that looks like a water heater. At night dump refrigerant through the Ice Bear (a commercial brand I have heard of) and create ice. During the day, put the refrigerant through the Ice Bear instead of running your compressor. This creates huge swings in the multiplier between you and McMansion. During the night the multiplier might be down to 5 or less. During the day, the multiplier might be up to 20. The result is that you pay a lower average rate using TOU if you manage your load.

Dynamic pricing means to me a moment by moment change in the price of electricity. During the hottest day of the year, the price might soar by a factor of 100. Avoid 7 hours at that high rate and you have reduced your average price by 1% (730 hours on average in the month) or more, at least if I have done the math correctly. Many utilities have equipment that they attach to your AC compressor to allow them to turn it off during such periods. The standard pricing approach is to pay you $5/month. They save money by avoiding some of the energy at the 100 times multiple.

Pricing is considered by many to be a form of rationing. Economists say that pricing is the best way to ration a commodity. It gets people to give up lower valued uses when there is a scarcity. I like to think of it as a way to soak those who can afford it so I can put in my controllers that allow me to store energy thermally. In some jargons, my thermal storage can be considered to be a form of supply side, though most just look at it as demand side.

As to whether the small utility gets the free ride or the large utility, it depends on the circumstance. My employer AEP, one of the largest in the country, may have been getting a free ride by building fewer very large generators. That lowered their average cost but they got reliability by interconnecting with other utilities that had many, though smaller, generators. Even the transmission example is not clear cut either way, especially when the high voltage line fails and the low voltage line experiences huge electrical line losses.

Solar PV might be sufficient but needs storage and agnostic pricing

It has been said that the US Department of Energy is supposed to be technologically neutral.  Many people say that DOE isn’t technologically neutral, but let’s build on the concept anyway.

We should have imbalance prices that are technologically neutral.  If I pump more solar PV into the grid than I have under contract, I will want to get paid for that surplus.  That surplus will be absorbed by someone with a shortage, who should pay for absorbing that surplus.  Given that there are hundreds, or thousands, or tens of thousands, or even millions (think of the number of solar roof tops) of parties participating in this balancing act of surpluses and shortages, the real time balancing price needs to be uniform (except for geographic differentiation.)

We could price my solar PV surplus differently from a simultaneous coal surplus, or from a simultaneous solar PV shortage, or from every other surplus or shortage, but that would be an accounting nightmare.

One way to handle the shortages and surpluses is with storage.  The uniform simultaneous price concept would be a way for storage to be compensated for its actions, buying the surpluses when the price is low and selling the shortages when prices are high.  This pricing concept should also be technologically neutral with regard to the form of storage, whether the lead acid or lithium ion batteries discussed by others or thermal storage that would fully handle the surpluses and only indirectly handle the shortages by turning of the charging power.