Solarcity is providing grid services, not the home owner. Solarcity gets paid for the services, then has an arrangement with participating homeowner to split proceeds.
In theory, Solarcity could contract anyone with a capable home system, so they could conceivably have sunrun, vivint, local installer installed homes under grid services resource contracts. This means they are not limited to solatcity only customers. This may prove that solarcity's aggregation software platform is extremely value IP. Since Solarcity has 1/3 of the market, their network might provide the greatest return for participating home owners, therefore quickly raising huge barriers to entry for other aggregation services competitors like sonnen, sunpower and the rest...
Aggregation as demand response looks very compelling as a peaker resource. From a prior post in August:
PJM spent $2bln on a peaker plant that has been used 13 times since 2009, which averages out to $25,000 per Mwh for demand response product. SCE spent $200mln in 2013, at $18,000 per Mwh for demand response.
what would a estimated cost/Mwh for Solarcity demand response in comparison? Might it be significantly cheaper to the grid then above numbers?
This is good stuff. You can really see how the utility world wants to compensate these peakers for MW capacity rather than MWh production as they are an economic abomination on a MWh basis.
So 300 Powerwalls should provide 1 MW capacity and 2.1 MWh max per event. (I'm actually not too impressed by arguments that batteries sustain fewer hours per peak event given that peaker plants are only utilize about 4 to 5 % or maybe 60 minutes per day. Their ability to sustain way underutilized and can easily complement batteries in a sustained event.) 300 Powerwalls would cost about $1.2 M to install. Note this is $1.2/W which is competitive to new gas peakers which are in range of $1/W.
So let's suppose 1% utilitization. That is 87.6 hours per year, roughly 1 to 2 hours per week. Over 10 years time the at 1% utilitization the cost would be $1.2M/MW ÷ 876h = $1370 MWh, excluding the cost to charge. So by PJM standards, this is super cheap. But I think the utilization can be pushed out to 5%. At this levels the cost reaches $274/MWh. 5% utilitization is stI'll just 437.5 hours per year, 1.2 hours per day. Under day cycling, there are 2.1 hours available each day. So 5% utilitization consumes about 57% cycle depth. Max ulilization would be 2.1 hours per day or 8.75%, which would imply $157/MWh, excluding the cost of charging. This should be seen a theoretical lower bound. 5% utilitization is probably a more robust assumption and gets us to $274/MWh.
Fuel cost, the cost of charging the battery, is even more interesting. Firstly, since we are talking about grid services and costs to the grid, we should consider power at wholesale prices. So the most economical wholesale price is the daily min, which can sometimes even be negative. But even when positive, the daily min can be so low that generators are operating at a loss. Thus, providing load at these prices to secure more profitable price support is an economic benefit to the grid. Suppose coal is at $3/MMBtu and a plant must burn 10.33 Btu/Wh. That's a fuel cost of $31/MWh. But because coal plants cannot ramp down and up quickly, they are willing to operate at below the cost of fuel for many hours a day. They may even need to pay the grid to take load, which is how the spot price can go negative. So batteries can soak up surplus baseload for prices below the cost of fossil fuels, below say $30/MWh. Moreover, if there is surplus wind or solar, this can push wholesale prices down to $0/MWh in absence of baseload plants (coal,.gas, nuclear) pushing the price negative. Indeed, the presence of nuclear power seems to guarantee that when demand is fully satified by wind and solar, spot prices will be negative. So longer-term as renewable penetration increases to a certain level, the reliable cost of fueling batteries may be under $1/MWh.
But what about those transmission and distribution costs? The energy losses in transmission are proprtional to the square of load. Thus, if you can avoid transmission during times of congestion, peak demand, this has a big impact on energy losses. About 10% of generated power is lost in transmission, but most of this is at times of peak demand. So here is where out 300 aggregated Powerwalls really shine. They are distributed. When discharges the power is consumed onsite or distributed within the substation. Thus, they avoid congestion when discharging. Charging happens local or at times of very low load. Not only do they reduce congestion, but they reduce the capacity needed to handle congestion. Thus, distributed batteries can reduce both the capex and opex costs of transmission for the grid. This implies negative marginal cost of transmission and distribution for distributed batteries. Note that centralized batteries used as a grid scale peaking facility do not generally offer this advantage. They have to push power through transmission at times of greatest congestion. The placement of the batteries has a big impact on the transmission losses. T&D costs for residential ratepayers is about $70/MWh. Let's say $50 is avoided on discharge at peak times and $25 is avoided when charging at times of minimal load. So altogether about $70/MWh is avoided in T&D.
Let's put this together. The cost of 1 MW is $1.2M. At 5% utilization, opex cost is $274/MWh, fuel cost $1 to $26 per MWh, and T&D an avoided cost of about $75. Thus, the net cost to the grid paying $1.2 M for this is $200 to $225 per MWh. Coincidentally, $1200/kW divided by 120 months is $10/kW/month. This is really not a bad price for standby power and substantially less than the $190/kW/year that California paid for standby power. If gas peakers really do need $190/kW/year to be profitable, they are going to have a tough time competing with aggregated home batteries at $120/kW/year.