Of Apples and Oranges
In October of last year, Energy Central posted an article written by Michael O’Boyle for Forbes, titled “The U.S. Southeast: A Hotspot For Uneconomic Fossil Power, Already Costs Consumers Millions, Risks Billions In Stranded Assets” [q.v.]. Its opening paragraph sets the scene:
Several major utilities in the U.S. Southeast have recently announced goals to reach low- or zero -carbon status. But you won't exactly get whiplash trying to keep up with a wave of immediate fossil fuel shutdowns following these commitments.
After noting that the plans in question stretch out over three decades and involve only a slow replacement of coal-fired plants with natural gas, O’Boyle continues:
The problem for consumers and the climate is that 95 percent of coal plants in the region are uneconomic compared to clean energy today, costing tens of millions of dollars annually. Of the six major power providers in the Southeast—Southern Company, Duke Energy, PPL, Dominion, Entergy, and TVA—five could replace every single megawatt of their coal-fired capacity now with cheaper renewable energy.
O’Boyle’s impression that existing coal-fired plants could be directly replaced by cheaper renewable energy highlights a major point of confusion in the clean energy debate. It stems indirectly from an overly broad application of the term LCOE (levelized cost of electricity). Comparisons between the LCOE for intermittent renewables and the LCOE for fossil fueled plants, as usually computed, suggest just what O’Boyle has concluded. But the comparisons are apples to oranges. Intermittent renewables cannot directly replace the coal-fired plants that O’Boyle is writing about.
As-Available vs. On-Demand electricity
We really need to start distinguishing between two different "flavors'' of LCOE. One is the levelized cost of as-available electricity (AAE). I'll refer to that as LCAAE. It's the cost for the flavor of electricity that intermittent renewables deliver. The other is the levelized cost of on-demand electricity (ODE) I’ll dub that LCODE. It's for the flavor of electricity that utility ratepayers expect and that the grid is required to deliver. Both LCAAE and LCODE relate to the cost of a kilowatt-hour (kWh) of electricity, but they’re not the same.
We think of electricity as a commodity, and see it referred to as such. But habits of thought can trip us up. When we think about the cost of a commodity, we’re nearly always considering a physical “thing”. It can be purchased and held, and then later used or resold. In other words, it’s a storable commodity. But electricity is not that. It’s ephemeral. It must be generated at the time it’s used. Even what we think of as electrical energy storage is actually conversion of electricity into another form of energy — typically chemical or gravitational potential energy — that is storable. Later, the stored potential energy can be converted back to electricity. The paired conversions achieve the desired effect of electrical energy storage, but always at a cost.
The difference between LCAAE and LCODE is that the latter reflects all costs needed to make electricity reliably available on demand. The former does not. It reflects only the cost to generate electricity when the sun is shining or the wind is blowing. Any load that’s present at those times can use it, and take advantage of its low cost. But any load that requires service at other times is effectively told “sorry, that’s not our department”.
Applications that want to take advantage of cheap AAE must be able to use power intermittently as it happens to be available. These are discretionary loads. They may require a certain total amount of energy over a given period of time, but there is latitude as to just when that power is drawn. Pumping water between reservoirs, or from wells to storage tanks, is the classic example. Pumps and motors are cheap for the power they consume. Being cheap reduces the economic penalty from running them only at times when cheap power is available.
If we’re serious about moving toward 100% renewables, we need to encourage more such applications. The larger the fraction of total load that can be run from AAE, the less costly it will be to provide ODE for the remainder. If the fraction becomes large enough, it will be possible to retire some of our fossil-fueled plants outright rather than keeping them around on standby as capacity reserve. But we’re not there yet.
About discretionary loads
Unlike simple water pumping, most applications require at least some degree of modification to accommodate intermittency. Often, the modification will involve integrating some form of storage. It could be simply buffer storage for the intermediate products of a production line or process that will be running intermittently. Or it may be thermal storage for applications that involve heating and cooling.The details will depend on the particular application being considered.
We’ll look at some specific cases shortly below. The point to note here, however, is that in nearly all cases, the modifications to enable discretionary operation will involve at least some increase in capital equipment cost. At the same time, intermittent operation will reduce capital productivity. Hence there’s a tradeoff between the cost of energy for the application vs. the productivity of capital. That tradeoff has to be evaluated on a case-by-case basis.
Often, the evaluation will find that the tradeoff is unfavorable for discretionary operation, if only the lower cost of AAE is considered. But that shouldn’t be the end of the story. Discretionary loads, operated under a smart demand-response (DR) control system, benefit the electricity grid as a whole. By strategically increasing or reducing their power draw depending on the state of the grid, they can provide what are known as “ancillary services” to the grid for following non-discretionary loads and for regulation of voltage and frequency.
Ancillary services of this type help to reduce the need for curtailment of excess wind and solar generation. They also reduce the number of times that fossil fueled generators need to be started up or shut down over the course of a day. However, in many states, regulations in place don’t allow discretionary loads to be used to provide regulation services to the grid. Legislation may need to be passed to rationalize policies for regulation services. Policies must specify protocols and interfaces to enable discretionary loads to participate. The compensation that an application can derive through provision of such services may tip the tradeoffs in favor of discretionary operation.
Some Examples
A few examples may help to give a better feel for the tradeoffs and potential for discretionary loads to accommodate AAE on the grid. This is by no means a comprehensive list, but the examples described are representative of particular categories. The categories differ in terms of scale and in degree of latitude for discretionary operation.
Emergency load shedding. One of the earliest and most basic approaches to DR involved installation of special power switches on large air conditioning units in New York. The AC units represented the largest individual loads during summer power emergencies. Disabling selected units was a benign form of load shedding. It avoided resorting to rolling blackouts when electricity demand threatened to exceed supply. In return for agreeing to the switches, participating businesses were given a lower rate on their utility bills. The lower rate applied whether or not the cutoff capability had been invoked.
A notable feature of this application for DR was that it required no equipment modification beyond installation of the special power switch. It had little to do with load management except in the gross sense of providing a benign alternative to rolling blackouts. Nor did it have anything to do with leveraging the lower cost of AAE. It was a means of dealing with the capacity limitations of the ODE system that supplied the region. But it did deliver system value. It was more economical than adding additional generation capacity to the grid that would only be needed to handle rare power emergencies.
Commercial air conditioning remains a popular target for DR efforts. The trend is for more sophisticated controls. Instead of the stark cutoff of a load shedding switch, signals from the utility allow softer measures to be implemented -- temporarily raising thermostat settings or employing motor controllers to lower the power to the units without disabling them altogether.
Hot water supply. Certain types of hot water heaters constitute a form of discretionary load that goes way back. These are electric hot water heaters that operate overnight, drawing cheap off-peak power. The hot water tank is well insulated and large enough to hold a full day’s worth of hot water for the home or business in which it’s installed. Many homes built in the 1950’s had water heaters of this type. They were large, heavy, and relatively expensive. They fell out of favor once piping of natural gas to residential developments became common. But updated versions using heat pumps and smart controls could be attractive.
Refrigeration and AC with thermal storage. Adding thermal storage to a refrigeration or air conditioning system allows the heat pump to draw power when it’s convenient while cool or cold air can be delivered on demand. The thermal store is usually an insulated tank holding either cold water or a freezing slurry of water and ice.
The cost of cooling systems with integrated thermal storage can be substantially higher than for conventional systems. They can be difficult to retrofit into existing buildings. Except for off-grid situations in which AAE is the only option, or in a stressed grid when the AAE to ODE price difference is unusually large, they may be hard to justify on the basis of electricity cost savings alone. But for new commercial construction, they make a lot of sense.
Aluminum smelting. Production of primary aluminum ingots from ore is energy intensive. The required energy typically ranges from 12,500 to 14,000 kWh per ton of aluminum [q.v.]. Depending on market conditions, the cost of that electricity typically accounts for a third to a half of the price of aluminum.
The German firm Trimet, a large producer of aluminum, has modified one of its production lines to enable it to function as a “virtual battery” [q.v.]. In response to conditions on the power grid, the modified line can quickly ramp its power draw by up to 25% above or below its nominal value (i.e., from 75% to 125% of nominal). That range corresponds to 1.12 GW of flexible power available for balancing variable supply.
The minimum operating level of 75% leaves the plant still dependent on Germany’s coal–fired power plants when winds are not blowing. However, it greatly reduces the loss of plant productivity that complete dependence on AAE would entail. It’s a hybrid approach that could be adopted to reduce electricity costs for other energy intensive applications as well.
Grid storage. Grid-connected battery banks can be thought of as the combination of a discretionary load and a flexible power generator. The discretionary load -- the charging side of the system -- produces “fuel” (charged battery state) for flexible generation of ODE. The value of this application is that it’s able to supply real power to the grid when needed. It’s not limited to simply reducing the level of power that it draws.
That distinction is unimportant if the grid is well-supplied with baseload or dispatchable power. It becomes very important, however, when variable renewables dominate energy supply to the grid. Supply from wind and solar PV can occasionally go to near zero for days on end. Delaying operation of discretionary loads can temporarily reduce the shortfall at those times, but it can’t avoid the need for backup supply. Unfortunately, regular storage batteries can’t realistically supply backup for more than a few hours. Alternatives can and are being developed, but they remain unproven.
RO desalination. Water desalination by reverse osmosis (RO) is a large, energy intensive application. It is also capital intensive; the RO membrane assemblies are expensive, and require frequent maintenance and replacement. Being capital intensive normally weighs strongly against discretionary operation. One wants to utilize that expensive equipment as close to full-time as possible. However, for RO desalination, it’s possible to decouple the most energy-intensive part from the most capital intensive part.
The energy-intensive part is pumping the feed water to high pressure. The capital intensive part is the operation of the RO membrane assemblies to yield desalinated water. In some cases, it may be possible to build a storage reservoir for feedwater at a high elevation above the main plant itself. Feedwater can be pumped up to the storage reservoir using AAE; the reservoir and gravity then provides a sufficient head of pressure to drive the RO process at a steady rate, full time.
This approach is most easily implemented where coastal mountain ranges meet the sea. The elevated feed water reservoir requires at least 350 meters of head in order to drive the RO process. There are project synergies if the desalination facility is coupled with a pumped hydroelectric storage (PHS) facility. The two facilities can share environmentally engineered seawater intake and outlet portals and the tunneling work to connect the sea level facilities to the elevated storage reservoir.
The bottom line
What I've just written about the difference between as-available and on-demand electricity lends itself to misinterpretation. Pointing out that AAE from wind and solar resources cannot directly replace ODE from dispatchable generation could be taken as an attack on RE. I do not mean it that way.
CO2-driven global warming and rapid climate change are real and quite serious. The need to slash CO2 atmospheric emissions is urgent. With the very low LCAAE that new wind and solar resources are now able to deliver, rapid buildup of those appears to be the most expedient and economical way to cut carbon emissions. But it's conditional on solving the ODE problem in a cleaner and more efficient manner than keeping old coal and gas-fired plants online.
Discretionary loads of the types described above can help, but are not enough. More scalable solutions for the problem do exist, or can be developed. I’ll try to write more about those in the coming weeks. The trouble is, they are mostly in exploratory development, not yet economically competitive. They will require policy support similar to that extended to RE resources if they are ever to achieve commercial viability.
Unfortunately, that support will be hard to secure while the RE community as a whole remains in denial about the existence of any problem. The difference between AAE and ODE, and what it takes to convert the former to the latter, needs to be better and more widely appreciated. Until that happens, journalists like O’Boyle will continue to be perplexed by the slow pace toward 100% renewables. They will blame it on the influence of those dastardly villains in the fossil fuel industries, while remaining blind to their own culpability for advancing a flawed story of RE.