The term “vehicle-to-grid” is mentioned twice in passing in the report detailing the impact of electric cars on our grid (see previous post). I wondered whether this vehicle-to-grid was the solution to their problem. After all, their calculation was done by averaging consumption, which is not really what will happen in the real world. But when they assume some top-down system of regulating demand, then I could somehow understand their reasoning.
I didn’t find any reference mentioning “vehicle-to-grid” in the report, but I wanted to know where the CREG got these assumptions from. I found that, to my surprise, the CREG earlier wrote a report on the impact of electric cars on a vehicle-to-grid system (pdf, Dutch ahead). The report is not new, it was published in 2010 with the data from 2007 and 2008. The subject of their research is the impact of the introduction of electric cars on the electricity spot market price.
The result of the 2010 report was similar to the 2016 report. They also researched the impact of 1 million electric cars and found that only 2.5% extra electricity needs to be produced on average (compared to 4% in the 2016 report) and that base load could easily absorb that extra electricity demand. The general conclusion of the 2010 report is that charging cars during off-peak hours will lower the spot prices. This because part of the capacity of the car battery could be used to trade on the energy market, buying electricity from the grid when it is cheap (during off-peak hours) and selling it at a high price when it is expensive (during peak hours).
It gets interesting when they explain their assumptions (on page 15 – 16):
- The electricity demand in each 15-minutes time frame should equal the production in the same time frame (seems rather wide to me)
- There are 1 million electric cars and at least 85 % of these are always connected to the electricity grid
- Average mileage of a car: 15,123 km/year or 41.4 km/day
- Capacity of the battery: 15 kWh (very low by today standards, but probably reasonable in 2010)
- Consumption: 0.15 kWh/km
- Capacity of charging- and discharging: 3 kW
- The efficiency of charging and discharging is 90 % (therefor the efficiency of the charging and discharging, a so-called round-trip, is 81 %)
- Charging- and discharging characteristics are linear
- A fully charged electric car with a usage battery capacity of 15 kWh can drive 100 km.
- If an electric car consumes 0.15 kWh/km, then this means a consumption of 41.4 * 0.15 = 6.21 kWh per day
- With a charging efficiency of 90% there is an average consumption of 6.21 / 0.9 = 6.9 kWh/day
- The average car driver wants at least a 30% safety margin for the autonomy of the car. Therefor a reserve of on average 6,9 * 1.3 = 8,97 kWh is needed
- The rest of the energy in the battery (15 – 8,97 = 6,03 kWh), can be used for energy arbitrage purposes
- For this 6.03 kWh there is 6.03 / 0.9 = 6.7 kWh necessary (with a charging efficiency of 90%)
- It is possible to sell 6,03 kWh * 0.9 = 5,43 kWh (with a discharging efficiency of 90%)
Just as in previous post, these calculations are based on several averages. This is not necessarily valid in the real world. There are a few difficulties that I can see (there might be more).
First, to have 850,000 cars hooked on the grid, at least 850,000 charging points should be available. Not only for example near super markets, but in all places where one could leave one’s car. These should also be smart charging points and those 850,000 cars should be able to charge from the grid as well as discharge to the grid (not all electric cars can go two ways).
Second, a problem (that is acknowledged in the report): the car driver is a weak link. The drivers need to be able and willing to hook up their car to the grid when they don’t drive it, this in order make it possible to put the surplus capacity of the car battery to use. If they, for whatever reason, fail to do that in large numbers, then that might be a problem for the operation of that vehicle-to-grid system.
It is not possible to know what the driver wants to do. Maybe he wants to make a long trip, but then maybe not. Will the driver find enough juice in his battery to do what he wants to do and when he wants to do it?
Third, another problem is the aging of the battery, which is acknowledged in the report. In the vehicle-to-grid system, the battery will not only be used for driving, but also for arbitrage. There are only a limited number of charging cycles that the battery can go through, so using that battery for arbitrage might age the battery faster than when the car is only used for driving.
In the report it is calculated that the lifespan of the battery will decrease almost by half, which seems a lot to me (but the authors admit that there was not much data on this at the time). This has some consequences. What will happen when a battery that could be used for say 10 years, will need to be replaced within 5 years because that car was part of the vehicle-to-grid system? If this is really the case, then the big issue will be: who will pay for the replacement of the faster aging battery before its intended lifespan?
The calculation counts on there being 850,000 cars hooked on the grid at any time, but if the battery degradation is high, then car owners who take care of their stuff might be reluctant to hook their cars on the grid when possible via those smart charging points.
Fourth, the most interesting is what is not mentioned. The 2016 report doesn’t mention the impact of intermittent power sources like solar and wind on the grid. It is assumed that cars are charging during off-peak hours and in their calculation the authors assumed this to be at night. However, in their conclusion they change this into “charging in due time”, suggesting that this could be a different period.
This means that their conclusion is based on a system with a stable base load without the influence of intermittent power sources. This is not the reality when we will reach those 1 million electric cars. The strategy of the Minister of Energy is to substantially increase the share of those intermittent power sources, therefor adding an additional level of complexity that is not taken into account in the calculation.
Then there is not only the (predictable) fluctuations of electricity where the calculation is based on, but also the (unpredictable) intermittent power. Does that conclusion then still holds? What will happen when there is a prolonged period of not much wind and traditionally less sunshine in winter, yet backup needs to be available to meet peak demand and the higher demand during the off-peak hours?
The 2010 report also doesn’t calculate the influence of intermittent power sources, but mentions in its conclusion that it is easier to integrate these sources in the system by using the vehicle-to-grid system. Which may or may not be true, but they didn’t included that into their model. They started from a stable base load and constructed a conclusion from there.
Last, but not least, energy is least available when it is needed most. We need the most energy in winter, at weekdays during rush hour at the morning peak and even more at the evening peak. At that time many cars will be in traffic, driving from or to work, therefor not connected to the grid. In a way, that is a good thing, they will not be able to draw energy from the grid, but at the same time it is also not possible to depend on the energy stored in their batteries to help balancing the grid load…
Also, more energy need to be left available in the battery for the driver in winter at our latitude. The lights and the heater will be on, drawing extra energy from the battery, meaning less capacity of the battery will be available for arbitrage and balancing the grid when it is needed more.
This may well change their conclusion, especially because the conclusion was build on the premise that the consumption is constant throughout the entire year.
Just as in previous post, I again have a problem with the averaging in a dynamic, intermittent system. Purely mathematically speaking, their conclusion is perfectly correct. But then, can we just average consumption of electric cars, knowing that this is not the case in reality and expect that the result is somehow relevant in the real world?
Trust, yet verify 