Megacharger costs… (Tesla Semi Part Two)

Elon Musk announced the Tesla Semi months ago, now. Besides the low cost, one of the things people are most incredulous about are the Megacharger costs. Tesla announced just 7 cents per kWh, flat price, to charge at a Megacharger. And that’d be done with solar power, potentially even unhooked from the grid. How is this feasible?

First we must look at existing Tesla Superchargers. Superchargers are capable of up to 145kWh charge rates (internally). Some versions were made by simply ganging up multiple home charger units together to get the required power (like 12 individual 12kW units). The high power was produced by clustering mass-produced smaller units that also are provided with each car. Just a single connector goes to each car, however.

Superchargers use a very large amount of power. A bunch of supercharger stalls being used at once at a Supercharger location can draw Megawatts. Simply installing a megawatt connection to the electric grid is expensive. So Tesla has started installing Powerpacks (the larger versions of the Powerwall designed for utility and commercial installations… around 210kWh apiece… themselves composed of 16 individual battery packs similar to those used for the Model S/X/3 and each with a DC-DC converter, with a DC-AC inverter of between 50 and 500kW of output, depending on configuration). This allows Tesla to add additional supercharging stalls to existing supercharger locations without upgrading the utility connection, and for new locations allows them to avoid utility peak charges which could actually end up being larger than the actual energy usage charges. As a side benefit, it also means that Superchargers have backup power in case of power outages:

…and in principle, Tesla could also optimize when the Powerpacks are charged to minimize time-of-day charges. That gives Tesla access to sub-7-cents-per-kWh electricity prices already. Industrial electricity prices are around 5 to 8 cents per kWh on average in the US (exception is Northeast, with about 9-12 cents per kWh), with off-peak electricity being about 2 or 4 cents per kWh less.

So Tesla probably ALREADY pays less than 7 cents per kWh for electricity on average. But they also have a significant amount of capital cost in the form of the chargers themselves and the batteries. The Batteries go for about $400/kWh retail, but Tesla’s internal price may be more like $150-250/kWh, especially without the inverter. That’s about 3 cents per kWh if they last for 20-25 years (which isn’t actually too unreasonable, given careful charging and discharging). 4 cents is more reasonable. But long-term, they hope to get cells below $100/kWh, and packs at, say, $120/kWh. (Raw material costs are about $35-45/kWh.) So <2 cents per kWh for the packs themselves is feasible, especially if they can get them to last a while. And ultimately, these packs can be assembled in an automated fashion, like their Model 3 packs. In fact, they could actually use the same battery line. (The biggest argument against automation is, like reuse, that it's not worth it at the volumes Tesla is considering, but if the same line is producing fairly standardized packs for multiple uses, that can dramatically improve the automation business case.)

But if you already are using battery packs for peak power reduction and maybe even time of day shifting, then the idea must occur to you: why not get rid of the utility entirely?

Solar cells are currently as low as 16 cents per Watt on the spot market (average 17.5 cents), without federal subsidies: That means 16 cents of cells produces 16 cents of electricity (at 7 cents per kWh) in a SINGLE year in a place like the American Southwest that has a capacity factor of about 26% or better, paying back the cost of the cells. Modules are more, obviously, but still cheap at 27 cents per Watt. If Tesla can automate installation, they may be able to install them and string them together for less than 40 or 50 cents per Watt. That doesn’t pay for a connection to the grid or the inverter. Because Tesla doesn’t need those. In fact, the batteries already contain a DC-DC converter. Careful selection of voltages could allow a nearly direct connection of the solar panels to the batteries, perhaps with a small and cheap “power optimizer” (i.e. DC-DC converter) to improve solar array efficiency. These things only cost a few cents per Watt at utility scale, so let’s call it an even 50 cents per Watt. But Tesla can avoid the grid connection at both the solar array side AND the Supercharger side. And can avoid the cost of the inverters and the inefficiencies/losses from converting to and from AC. That previous 210kWh Powerpack thus has more like 225kWh. And over 25 years, therefore, a solar array that costs just 50 cents per Watt to install means electricity at less than 1 cent per kWh (0.9 cents) in a place with 26 percent capacity factor. However, there are sometimes cloudy days, where solar arrays will be less effective. To counter-act that, we make the solar array about twice the size. Cost per kWh doesn’t quite double, however, as not all the balance-of-system costs double. So let’s say 1.5 cents per kWh. Batteries also need to be about double, so cost of the battery is about 4 cents per kWh total, maybe less. But total cost of electricity is thus just 5.5 cents per kWh, leaving room financing costs. Doable if financing can be kept at low costs (and the solar arrays can actually last a LOT longer, like 50-100 years… and by doubling up the batteries as we did, they can also last a lot longer). And remember, solar costs will continue to decrease, long-term (tariffs notwithstanding). 3 cents per kWh raw solar+battery costs and 5.5 cents per kWh assuming roughly doubly up number of both li-ion and photovoltaic cells.

So that’s how Tesla can offer 7 cents per kWh for the Tesla semi while disconnecting from the grid and using solar power.

This entry was posted in Uncategorized. Bookmark the permalink.

7 Responses to Megacharger costs… (Tesla Semi Part Two)

  1. Paul451 says:

    “but still cheap at 27 cents per kWh.
    less than 40 or 50 cents per kWh.”

    Did you bork the units there?

  2. Chris Stelter says:

    Yup. Fixed.

  3. ech says:

    What about maintenance costs? Someone has to come and clean the solar panels from time to time, if nothing else.

  4. Chris Stelter says:

    Technically, you don’t have to. You could just let them get dusty and eat the capacity reduction. But it probably does make sense to budget some maintenance.

    Data from utility-scale solar arrays says that maintenance is approximately half a cent per kWh for such large arrays like a Megacharger location would need. Not a huge deal.

  5. Paul D. says:

    Some amazingly low bids are being received in Colorado now for wind and solar power.

    ‘In January, the utility released the results of a solicitation that returned a median price bid of $21 per megawatt-hour for wind-plus-storage projects and a median bid of $36 per megawatt-hour for solar-plus-storage.

    Bids highlighted in Xcel’s new electric resource filing are even lower. The proposal includes “unprecedented low pricing” across a range of generation technologies, with wind at levelized pricing between $11-$18 per megawatt-hour, solar between $23-$27 per megawatt-hour, and solar-plus-storage between $30-$32 per megawatt-hour, the document states.’

  6. antrik says:

    I’m sorry to say that these calculations are based on a lot of arbitrary guesses and questionable assumptions.

    First of all, the panel price makes up only a fraction of a solar installation’s cost. The actual going rate for utility-scale solar installations in the U.S. is around $1/W. Since this is a very competitive market, there is no reason to assume that Tesla can do it significantly cheaper. With solar prices constantly falling, the cost is expected to be *somewhat* lower though when Tesla will begin installing the Megachargers at scale in 2020 or so…

    Since the Megachargers need to be built where demand is, rather than picking the best solar locations, capacity factor is likely to be below 25% on average.

    As far as I’m aware, for a typical utility-scale solar installation, with a 20 years power purchase agreement, the final price per kWh, accounting for operating costs and financing, comes in at about twice the amortized investment cost alone. The situation for the Megachargers is a bit different though, since Tesla doesn’t get a fixed selling price for 20 years. Having to compete with new installations coming online at slightly above half the cost within 10 years, effective depreciation is higher than with a 20 years fixed-price PPA. On the other hand, this is partially offset by lower financing costs when looking at a shorter payback period…

    Of course we shouldn’t forget the significant investment tax credits available for both solar and storage, and presumably still available in 2020.

    All in all, my estimation is at about 3.5 cents/kWh for the generation part alone; which seems to fit in more or less with actual numbers I believe I have seen thrown around for medium-scale solar installations.

    (Note that the sometimes much lower numbers coming up in auctions are for very large projects, which are to be installed many years from now — so they are not really comparable…)

    However, if we are looking at an off-grid project with an assumed 2x overbuild, that means the generation part alone would come out at about 7 cents/kWh…

    The numbers you cite for the storage part seem even more arbitrary. Real costs are harder to estimate here though, with the market still being small and very much in flux. I believe $400/kWh sounds about right for current projects; but again, there is no reason to believe Tesla can calculate with significantly lower costs internally. On the other hand, storage prices are falling much faster than solar — so around $250/kWh in 2020 actually sounds plausible.

    Life times for Li-Ion batteries however aren’t anywhere close to 20-25 years. AFAIK Tesla promises 10 years of daily cycling for their Powerpacks — and daily cycling is what we have to assume in this application. Considering the precipitous price falls for storage projects, calculating over a longer time period doesn’t make much sense anyway…

    The storage part doesn’t have to cover all the energy used though, as the energy generated during daytime can be used directly. Still, during winter season, I’d say about 75% of the energy must be stored.

    On the other hand, I don’t see a need for the storage part to be overbuilt. For a completely off-grid system, we essentially have to assume any number of bad days in succession; which means we have to overbuild the generation part to cover even a bad day, while the storage part never needs to cover more than any single day.

    All in all, my estimation for the storage part alone, also comes in somewhere around 7 cents/kWh — on top of the 7 cents/kWh for the generation part.

    To add a different angle, a fairly large Tesla/SolarCity solar+storage micro-grid on Kaua’i was built in 2016 at 14 cents/kWh. Another, even larger project built by another developer at around the same time boasted 10.5 cents/kWh.

    Logistics costs for these island projects are surely higher than for mainland installations; and solar+storage costs should come down in general quite a bit by 2020. On the other hand, climate conditions for mainland projects are worse. The size of the Megacharger installations should be smaller too, i.e. costs are likely higher. What’s more, the projects I mentioned are not entirely autonomous AIUI, since they can still get back-up from other existing generation — i.e. they don’t need much overbuild. And last but not least, the size of the storage part for a Megacharger needs to be even larger than for a generic micro-grid, since it likely doesn’t benefit from a significant portion of demand taking place during daytime…

    Looking at it either way, 7 cents/kWh in 2020 for off-grid projects just doesn’t seem realistic.

    The good news though is that the Megachargers do not actually need to be completely off-grid. They can pull extra energy from the grid during winter time, and export excess energy during summer — thus avoiding the need for overbuild. Also, the storage part only needs to do balancing / peak shaving, along with limited load shifting mostly for evening hours; while at night they can buy cheap spare power from wind and hydro instead.

    This totally changes the economics, and makes 7 cent/kWh in 2020 entirely plausible I believe.

  7. Chris Stelter says:

    “Life times for Li-Ion batteries however aren’t anywhere close to 20-25 years. AFAIK Tesla promises 10 years of daily cycling for their Powerpacks — and daily cycling is what we have to assume in this application. ”
    Guarantee of 10 year lifetime, but plenty of residual capacity well after that, particularly if the batteries are over-sized like I mentioned. I do think >20 years is a legitimate estimate for majority capacity, given results in this paper:

    “On the other hand, storage prices are falling much faster than solar — so around $250/kWh in 2020 actually sounds plausible.”
    I don’t see why Tesla cannot do MUCH better than $250/kWh, especially if sharing some of the balance of system costs with the solar installation. Heck, the $180,000 Tesla Semi probably needs 900kWh (or more), giving you a $200/kWh price INCLUDING THE WHOLE TRUCK. Tesla Semi basically depends on near-$100/kWh pricing for battery modules to begin with, so it’s perfectly logical to extend this to grid storage (which can be made cheaper and/or same price and longer lasting due to lower C-rate).

    As far as solar, state of the art solar installations are down to below 2 cents per kWh now (adjusted for inflation). 7 cents is much too high for even an over-sized array. Remember you don’t need to oversize the balance-of-system costs so much, just the cells and the modules which are INCREDIBLY cheap, having dropped from 17 cents and 27 cents–respectively–at the time of my post to now down to as low as 12 cents and 24 cents per watt on the spot market. And the balance of system costs are amenable to innovation in automation even more than the baseline

    As far as prices at 7 cents per kWh in 2020, I suspect Tesla to play the same games as solar power purchase agreements (i.e. the early pricing might be lower than spot market prices, but average price over the long term is profitable), so I think using solar power purchase agreements as a baseline for comparison is valid.

    And I think you’re under-estimating the need for storage to enable the very high charge rates that these Semis need. Even some existing supercharger stations use stationary storage for this reason.

Leave a Reply

Your email address will not be published. Required fields are marked *