Tesla Semi

[abasile]

I'll just say that it's cheaper to ship excess energy across the grid to whereever it's wanted or wherever it can be stored... than it is to ship it in the form of a truck driving around. We probably will have a few "pre-charged" trucks for "fast swapping", but it won't be large.

In general, I agree. This is what the grid is supposed to do for us!

There may be cases where truck power is cheaper, especially with high demand rates: pop up Superchargers.
What's the surprise at Semi unveiling?

Yes. Lack of flexibility on demand fees, on the part of utility companies, may sometimes make it worthwhile to bypass the grid and instead pay for a mobile battery. It shouldn't be this way, but change takes time.

Longer term, there will always be cases where mobile power is needed on a short term basis. I am eager to see batteries displace generators.

 

[jhm]

I'll just say that it's cheaper to ship excess energy across the grid to whereever it's wanted or wherever it can be stored... than it is to ship it in the form of a truck driving around. We probably will have a few "pre-charged" trucks for "fast swapping", but it won't be large.

I've already assumed the at least some of the wind and solar will be via PPAs. So I already have in mind that the same PPA can supply any Megacharger WITHIN the same transmission grid. But trucks cross transmission grid boundaries the same way they cross state boundaries. So there are unique opportunities to power up where it is actually cheaper to do so.

There is this idea that a carbon neutral future will force grid planners to create massive transmission backbones so that solar power in Australia, for example, can be transmitted to Japan. I am skeptical. I think there are many better ways to get the same effects for lower costs. For example, rather than transmitting power from Australia to Japan, it may prove more economical to move factories from Japan to Australia. Simply shifting where load happens is an alternative to building ever larger transmission grids.

The concept is simple as this. If you had a truck and could charge at 10c/kWh here or 5c/kWh a few miles down the road along your route, you'd be inclined to charge where it is cheaper. That's the microeconomic view. In the macro view, when trucking comes to represent 5% or 10% of demand for electricity, the response of trucking at the margins of grids will capture these natural arbitrage opportunities and help balance out price differences without need of transmission infrastructure. We're not talking about moving power, just charging where it is cheaper.

 

[cpa]

Oddly enough, the genesis of this discussion was in an announcement that Tesla had secured trucking for Q4 this year. That is, Tesla has been addressing the many challenges of scaling up at three or more times where they were last year. So part of securing trucking capacity for this quarter has been to actually buy a few trucking companies outright. Presumably, these companies are already capable within the trucking industry; otherwise, I don't know why Tesla would want to buy them. Thus, the developmental path is to start with a capable trucking company and then to introduce Tesla Semis. This will give Tesla valuable field experience in integrating electric semis into an actual trucking company. I believe this plan is also inclusive of client relationships these trucking may have had with other companies. Thus, Tesla get experience hauling its own freight and freight from other clients. We don't know just how far Tesla will take it beyond this, but even at small scale this is valuable experiences a few years before Tesla is selling Semis at scale. So I believe this developmental path addresses most of your concerns.

I think it all depends on how Elon Musk, er, Tesla, decides to operate this business segment. Will Tesla be willing to decentralize this division and let those who really know the business operate independently? Will Tesla bring this division into its current structure and control it from Palo Alto and Fremont?

Tesla routinely wins Consumer Reports' customer satisfaction survey by a wide margin

My belief is that these surveys (and I can be wrong) have to do with the finished product after the customer has driven the vehicle for a while. I do not know (and I am sure that you will correct me!) that these surveys are not broad enough. Do they drill down into the customer service (or lack thereof) of acquiring the car? Do they address issues that are not directly related to the performance of the vehicle?

I am sure Tesla Semi will have teething pains just like any completely new product, but saving trucking companies oodles of money with a fantastic product should help take the sting away.

I think that is the rub. Individual customers of Tesla rarely rely upon their vehicles to make a living the way transportation companies do. The transportation companies have customers who will depend upon them to deliver the goods timely. Delayed deliveries frequently results in contractual adjustments to the trucking company. My clients that transported fresh produce from California to the east coast could see up to a 20% reduction in their contract price for late deliveries.

ya.. so what?

Every company that is pushing the envelope on a new technology or process will have to go through the same growing pains. You are comparing against Dinosaurs that have done the same thing over the last 100 years and have ironed out their supply chain

"that a 10-year-old company ..."

They have been mass manufacturing only the last 5 years, and even then I would say real mass manufacturing started only the last 6 months. "10-year-old company" is only true from the registration perspective, but what counts is 2013 is when they really started cranking out their first car, and the one in real volume just 6 months ago.

Yes, there have been, and there will continue to be growing pains. I mentioned nothing about the supply chain, so that seems to be a straw man position. You seem to focus solely on the mass manufacturing process. I am more concerned with events since we purchased our S in 2014. Sixty-minute waits on telephone hold. Dead links on our personal web page. No way to contact a service center directly to make an appointment. Supercharger issues that are not resolved after more than a month. Failure to deliver our Model 3 that received no resolution despite many emails to three different people. Phone calls went unanswered and messages were not returned. Emails to Tesla, including the "escalation" feature that are not returned or even acknowledged.

None of these issues has anything to do with being in business like the dinosaurs for one hundred years. These are just common sense areas of customer relations that Tesla seems to give short shrift to. It seems to me that Tesla is trying to save as much money on overhead as they can, so perhaps they are significantly understaffed and under trained. But eventually they will have to start focusing on these things. Not all consumers will be as tolerant and forgiving as those of us who believe in the company and its products. They will spend their money elsewhere.

 

[neroden]

Yes, there have been, and there will continue to be growing pains. I mentioned nothing about the supply chain, so that seems to be a straw man position. You seem to focus solely on the mass manufacturing process. I am more concerned with events since we purchased our S in 2014. Sixty-minute waits on telephone hold. Dead links on our personal web page. No way to contact a service center directly to make an appointment. Supercharger issues that are not resolved after more than a month. Failure to deliver our Model 3 that received no resolution despite many emails to three different people. Phone calls went unanswered and messages were not returned. Emails to Tesla, including the "escalation" feature that are not returned or even acknowledged.

None of these issues has anything to do with being in business like the dinosaurs for one hundred years. These are just common sense areas of customer relations that Tesla seems to give short shrift to. It seems to me that Tesla is trying to save as much money on overhead as they can, so perhaps they are significantly understaffed and under trained. But eventually they will have to start focusing on these things. Not all consumers will be as tolerant and forgiving as those of us who believe in the company and its products. They will spend their money elsewhere.

This. This is why Tesla is dead when EVs start competing with each other. (Which is not yet.) Unless they fix this.

 

[abasile]

The concept is simple as this. If you had a truck and could charge at 10c/kWh here or 5c/kWh a few miles down the road along your route, you'd be inclined to charge where it is cheaper. That's the microeconomic view. In the macro view, when trucking comes to represent 5% or 10% of demand for electricity, the response of trucking at the margins of grids will capture these natural arbitrage opportunities and help balance out price differences without need of transmission infrastructure. We're not talking about moving power, just charging where it is cheaper.

I suspect that Tesla has long been considering energy costs when deciding where to place Superchargers.

For example, I'm personally familiar with Supercharger sites that are very, very close to the CA/NV state line. I don't think it's any coincidence that all three are in Nevada (State Line by Lake Tahoe, Topaz Lake on US-395, and Primm along I-15). EDIT: There's a fourth SC under construction in Incline Village, NV, also very, very close to the state line by Lake Tahoe.

More generally, I don't see anyone moving factories from Japan to Australia in the near term to take advantage of cheap, abundant solar energy. But we might see this happen as factories become much more automated, with fewer dependencies on the local availability of human labor.

 

[jhm]

Math Quiz

Suppose a charger is complete fed from the grid. It pays $600/peakMW/day in demand charges and $60/MWh for energy. Suppose serves 20 trucks per day that charge at 1MW each for 1 hour. Which of the following is the best option?

A) Charge all 20 trucks in the same hour directly from the grid.
B) Install a 19MW/19MWh stationary battery and charge all 20 trucks in the same hour from battery and grid.
C) Charge no more than 5 trucks per hour directly from the grid.
D) Install 4MW/16MWh battery and charge no more than 5 trucks per hour from battery and grid.
E) Charge no more than 1 truck per hour directly from grid.

You might suppose that the cost of an installed battery is $100k per MW plus $200k per MWh. Suppose also 6% financing on a 20-year lease which amounts to a daily lease of $232 per $1M principle. Or supply your preferred assumption and why.

So let us know what option you think is best and why.

 

[GleanerC]

Suppose a charger is complete fed from the grid. It pays $600/peakMW/day in demand charges and $60/MWh for energy. Suppose serves 20 trucks per day that charge at 1MW each for 1 hour. Which of the following is the best option?

I'm trying to see how the demand charge can stay the same under these different scenarios. Demand charges, I am familiar with, are usually accessed on peak demand flow over a short period and usually only during certain times of the day,week and/or year. Our local example, demand charges are determined during the summer months, weekdays, noon to 8 pm. The highest flow during this time, over a 15 minute period, determines your monthly/daily demand charge for the year. (This is only for commercial not residential.) If one where to charge all 20 trucks or 5 trucks at the same time I would think the demand charges would be much, much higher than in options B, D and E.

I'm thinking the demand charges would be the same between options B, D and E. The charging facility could be pulling peak 1MW per hour to charge the stationary battery. In scenario B the battery would need to be full before the trucks began to charge. Scenario D gives the operator more options. If the battery is full before charging begins all twenty trucks could be charged over 4 consecutive hours while still only pulling 1MW from the grid. Again, this would allow a much lower demand charge by keeping peak flow from the grid to only 1MW. (these are very basic estimates that don't include charging losses and being able to us the full capacity of the listed battery, neither would actually be the case in the real world)

Now, I'll admit I didn't do any of the math comparing the costs and financing of the batteries as I'm thinking the fixed $600/peak/day in demand charges wouldn't be correct between the different scenarios listed. So, what did I miss or am I working under incorrect assumptions (or just our local example) of how demand charges work?

Thanks!

 

[landis]

So let us know what option you think is best and why.

Flexibility isn't free. If one charge cable is adequate then of course E is the best option.
Otherwise D is the most cost effective option (if 5 charge cables are adequate)
Even at one third of provided demand parameter ($200), D still undercuts C.
At the provided demand parameter, B also undercuts C.

Storage cost approximately same as energy cost, and less than the equivalent demand cost.
If my calculations for A-E are correct: $13,200; $3,122; $4,200; $2,635; $1,800

I'm thinking the demand charges would be the same between options B, D and E

I'm thinking that was the point of the scenario.

If one where to charge all 20 trucks or 5 trucks at the same time I would think the demand charges would be much, much higher than in options B, D and E.

Yes, so even at $200 demand rate, the storage is 'free' given the number (5 or 20) of required charge cables.

 

[jhm]

I'm trying to see how the demand charge can stay the same under these different scenarios. Demand charges, I am familiar with, are usually accessed on peak demand flow over a short period and usually only during certain times of the day,week and/or year. Our local example, demand charges are determined during the summer months, weekdays, noon to 8 pm. The highest flow during this time, over a 15 minute period, determines your monthly/daily demand charge for the year. (This is only for commercial not residential.) If one where to charge all 20 trucks or 5 trucks at the same time I would think the demand charges would be much, much higher than in options B, D and E.

I'm thinking the demand charges would be the same between options B, D and E. The charging facility could be pulling peak 1MW per hour to charge the stationary battery. In scenario B the battery would need to be full before the trucks began to charge. Scenario D gives the operator more options. If the battery is full before charging begins all twenty trucks could be charged over 4 consecutive hours while still only pulling 1MW from the grid. Again, this would allow a much lower demand charge by keeping peak flow from the grid to only 1MW. (these are very basic estimates that don't include charging losses and being able to us the full capacity of the listed battery, neither would actually be the case in the real world)

Now, I'll admit I didn't do any of the math comparing the costs and financing of the batteries as I'm thinking the fixed $600/peak/day in demand charges wouldn't be correct between the different scenarios listed. So, what did I miss or am I working under incorrect assumptions (or just our local example) of how demand charges work?

Thanks!

You bring up a good point. There are lists of different kinds of rate plans out there, including ones that are pricerier at different times of day or year. It would be good try out some different kinds of plans to see if batteries are more or less valuable in those contexts. Also to see if adding solar or wind is more valuable. For example a plan that makes summer power more expensive might make solar really attractive.

In the simple rate plan I suggested, the whole exercise is how to avoid demand charges. The demand charges are high enough that it is really easy for a battery assist to save a serious amount of money. But is is also important to spread demand out which reduce the size of battery needed to defeat demand charges.

Thanks.

 

[mongo]

You bring up a good point. There are lists of different kinds of rate plans out there, including ones that are pricerier at different times of day or year. It would be good try out some different kinds of plans to see if batteries are more or less valuable in those contexts. Also to see if adding solar or wind is more valuable. For example a plan that makes summer power more expensive might make solar really attractive.

In the simple rate plan I suggested, the whole exercise is how to avoid demand charges. The demand charges are high enough that it is really easy for a battery assist to save a serious amount of money. But is is also important to spread demand out which reduce the size of battery needed to defeat demand charges.

Thanks.

A larger battery install could also be a joint venture with the power utility to provide some modicum of load leveling. In return, the utility creates a use&storage rate plan.

If the internal topology of the mega pack is similar to the powerpack*, then it itself is the mega charger.

*In the power pack, each battery sub-module has its own DC-DC converter for charge/ discharge control. These are run in parallel to the inverter, but that internal HV DC bus could just as easily connect to the semi's packs. Mega has 11 inverter sub modules, so that could allow two semis to charge at a time (based on the 8 terminal charge port, which makes me think the semi is 4 independent pack/drive units)

 

[jhm]

Flexibility isn't free. If one charge cable is adequate then of course E is the best option.
Otherwise D is the most cost effective option (if 5 charge cables are adequate)
Even at one third of provided demand parameter ($200), D still undercuts C.
At the provided demand parameter, B also undercuts C.

Storage cost approximately same as energy cost, and less than the equivalent demand cost.
If my calculations for A-E are correct: $13,200; $3,122; $4,200; $2,635; $1,800



I'm thinking that was the point of the scenario.



Yes, so even at $200 demand rate, the storage is 'free' given the number (5 or 20) of required charge cables.

You got the same results I did.

Good point if the demand charge were just $200/MWpeak/day. Here the battery still saves money between case A and B. But it is just about break even when there is moderate spread of demand over the day, C vs D.

Option E seems most unrealistic to me. Even if you owned your own trucking company, it would be operationally very hard to charge only one truck per hour for 20 hours a day.

We can also think about the value of batteries developmentally. Suppose that within the span of 12 months demand at this source grows from 20 MWh per day to 30 MWh per day. It will be even harder to avoid a peak hour with more than 5 trucks in one hour. And the value of the battery goes up. Or put another way, the optimal size of the battery increases with daily demand.

 

[mongo]

You got the same results I did.

Good point if the demand charge were just $200/MWpeak/day. Here the battery still saves money between case A and B. But it is just about break even when there is moderate spread of demand over the day, C vs D.

Option E seems most unrealistic to me. Even if you owned your own trucking company, it would be operationally very hard to charge only one truck per hour for 20 hours a day.

We can also think about the value of batteries developmentally. Suppose that within the span of 12 months demand at this source grows from 20 MWh per day to 30 MWh per day. It will be even harder to avoid a peak hour with more than 5 trucks in one hour. And the value of the battery goes up. Or put another way, the optimal size of the battery increases with daily demand.

Speaking of more batteries:

400 miles of charge in 30 minutes assuming an 800 kWh 500 mile pack is 640 kWh at an 1.280 MW rate. Mega pack is rated at 2,673 kWh, applying the usual 4:1 energy to power ratio, that gets us right around the 770 kW rating of the inverter. So one semi may really need two Megas to charge from, unless the pack can handle a 2:1 ratio.

Looking at it another way, a semi with trailer is 70-80 ft long. Mega is just under 24', so a unit of 4 mega and one transformer (the setup being proposed for the PG&E site) allows two semis with trailer to charge simultaneously, one on each side. This setup minimizes cable length, especially if the semis face opposite directions allowing the back to back Megas to pair up. This is the natural orientation if there is only one charge port.

 

[jhm]

Speaking of more batteries:

400 miles of charge in 30 minutes assuming an 800 kWh 500 mile pack is 640 kWh at an 1.280 MW rate. Mega pack is rated at 2,673 kWh, applying the usual 4:1 energy to power ratio, that gets us right around the 770 kW rating of the inverter. So one semi may really need two Megas to charge from, unless the pack can handle a 2:1 ratio.

Looking at it another way, a semi with trailer is 70-80 ft long. Mega is just under 24', so a unit of 4 mega and one transformer (the setup being proposed for the PG&E site) allows two semis with trailer to charge simultaneously, one on each side. This setup minimizes cable length, especially if the semis face opposite directions allowing the back to back Megas to pair up. This is the natural orientation if there is only one charge port.

Cool. It is nice to see that the Megapacks are at a scale that Megachargers will need. The 4:1 ratio is good to keep in mind. I think this is well suited for Megachargers. Particularly, if you think about spreading your peak out over 4 hours (as I did in options C and D in the math quiz), then you need a 4:1 ratio.

In terms of power rating, paring 770kW of battery assist to a max charging load of 1280kW charging load seems about right. Specifically the difference of 510kW would be supplied direct from the grid.

Roughly 5 Megas is what you want to support 5MW of charging load (about four trucks at 1.25MW per truck) as suggested in option D. Note that I used "trucks" in units of 1MW to keep the arithmetic simple. Really this just translates into to so many MW of charging load, rather than actual trucks.

Another thing that is key about the design of the Megas is the installation convenience. So I could envision that a Megacharger station is designed with plots of land where the Megas will eventually go. But there is little need to install all of them at once. Rather as demand at a station grows, so will its need for Megas. So Tesla is in a position of setting up new stations and monitoring demand. As they look across all stations they can do an analysis that determines the marginal value of adding one more Mega to a station. This gives Tesla a way to rank order which stations will get an upgrade each month. In our math quiz example, 1 well positioned Mega can offset $462 in demand charges per week, or $168,630 per year. Assuming financing at 6% over 20 useful life, this has a present value of $2.0M.

Do we know the likely price or cost of the Mega? It's nice to know that at least in areas where businesses are faced with $18/kWpeak/month ($600/MWpeak/day) peak charges, this beast will be worth about $2M.

 

[mongo]

Cool. It is nice to see that the Megapacks are at a scale that Megachargers will need. The 4:1 ratio is good to keep in mind. I think this is well suited for Megachargers. Particularly, if you think about spreading your peak out over 4 hours (as I did in options C and D in the math quiz), then you need a 4:1 ratio.

In terms of power rating, paring 770kW of battery assist to a max charging load of 1280kW charging load seems about right. Specifically the difference of 510kW would be supplied direct from the grid.

Roughly 5 Megas is what you want to support 5MW of charging load (about four trucks at 1.25MW per truck) as suggested in option D. Note that I used "trucks" in units of 1MW to keep the arithmetic simple. Really this just translates into to so many MW of charging load, rather than actual trucks.

Another thing that is key about the design of the Megas is the installation convenience. So I could envision that a Megacharger station is designed with plots of land where the Megas will eventually go. But there is little need to install all of them at once. Rather as demand at a station grows, so will its need for Megas. So Tesla is in a position of setting up new stations and monitoring demand. As they look across all stations they can do an analysis that determines the marginal value of adding one more Mega to a station. This gives Tesla a way to rank order which stations will get an upgrade each month. In our math quiz example, 1 well positioned Mega can offset $462 in demand charges per week, or $168,630 per year. Assuming financing at 6% over 20 useful life, this has a present value of $2.0M.

Do we know the likely price or cost of the Mega? It's nice to know that at least in areas where businesses are faced with $18/kWpeak/month ($600/MWpeak/day) peak charges, this beast will be worth about $2M.

Yeah, if the inverter has equal charging capacity as discharging, and the use of grid power at that rate is acceptable, they could charge one semi at full rate per mega: half pack, half inverter. Options include something similar to the current setup, one semi per two Megas from pure battery, or two semi at full rate from grid+battery. That may align well with usage, refill as needed during the day, charge fleet at night. Could even drop the rate and hook up 2 tractors per mega.

 

[neroden]

In the simple rate plan I suggested, the whole exercise is how to avoid demand charges. The demand charges are high enough that it is really easy for a battery assist to save a serious amount of money. But is is also important to spread demand out which reduce the size of battery needed to defeat demand charges.

Tesla has never, never made the decision to try to change the demand profile rather than building larger batteries. Tesla ALWAYS goes for the big battery choice. They're the only carmaker to do this until recently, they're the only truckmaker to do this, they dropped the "backup only Powerwall" specifically to make this choice, etc. etc.

They're going to build the batteries, and tell the truckers they can recharge whenever they want.

Case closed.

 

[neroden]

Another thing that is key about the design of the Megas is the installation convenience. So I could envision that a Megacharger station is designed with plots of land where the Megas will eventually go. But there is little need to install all of them at once. Rather as demand at a station grows, so will its need for Megas. So Tesla is in a position of setting up new stations and monitoring demand. As they look across all stations they can do an analysis that determines the marginal value of adding one more Mega to a station. This gives Tesla a way to rank order which stations will get an upgrade each month.

Bingo. That's what they're going to do. Start with a station designed for a low expected usage, add more batteries as usage goes higher. (And eventually they'll have to add some solar farms, to avoid upgrading the grid hookup.)

 

[cpa]

I suspect that Tesla has long been considering energy costs when deciding where to place Superchargers.

For example, I'm personally familiar with Supercharger sites that are very, very close to the CA/NV state line. I don't think it's any coincidence that all three are in Nevada (State Line by Lake Tahoe, Topaz Lake on US-395, and Primm along I-15). EDIT: There's a fourth SC under construction in Incline Village, NV, also very, very close to the state line by Lake Tahoe.

More generally, I don't see anyone moving factories from Japan to Australia in the near term to take advantage of cheap, abundant solar energy. But we might see this happen as factories become much more automated, with fewer dependencies on the local availability of human labor.

I believe that if it is a toss-up as to placement, then energy costs could be the deciding factor. However, I respectfully disagree with your assertion about the three (soon to be four) Nevada sites.

Stateline has pretty easy ingress and egress. South Lake Tahoe is generally a mess during ski season and during the summer months along US50. Maybe they could have placed it near the Y if there were adequate facilities. Maybe the casino offered free rent and even is paying for some or all of the electricity (which might tangentially validate your thesis.)

Topaz Lake is the only real settled area along US395 between Bridgeport and Gardnerville proper. There are a motel and casino. Walker, Coleville, and Topaz on the California side are uninteresting and essentially undeveloped. The gyrenes at the MWTC off SR108 don't want us plugging in! (LOL) Bridgeport might have been an option, but there is not much to do there other than tour the old Mono County Jail or eat at the handful of restaurants in town, and Bridgeport can be nasty cold in winter with not a lot to do.

Primm was a prime spot between Barstow and Vegas for through traffic. At the time Primm was opened, the only SC was in downtown Vegas. This was well off the interstate in a less-than-stellar part of town. Primm made a lot of sense to bypass Vegas and reach St. George. And you know as well as I that there is nothing on the California side of the line once you leave Baker. With the two sets of hills between Baker and the Nevada line, Baker would have been too distant for people to reach St. George comfortably, especially with 60s, 70s, and 75s.

Incline is an interesting location. The California equivalent would be King's Beach. But that is awful close to Truckee-Broadway (18 miles?) Incline Village may have more seasonal residents who spend considerable amounts of time in summer and winter and would need to have a local place to charge, especially in winter when energy use by the car is quite high.

Onto Arizona:

You might be correct in your belief when we consider Quartzsite versus Blythe. The Quartzite location is behind a Taco Bell, and it is frequented by a number of sketchy types at times. Blythe is about 20 miles or so away, and offers up a little more for the traveler. Yuma wins out over Winterhaven, however. Winterhaven does not have much going for it.

 

[jhm]

Using diesel to charge EVs in the outback is greener than you think | The Driven

I'll just drop this here.

 

[jhm]

There is one more thing I wan to point out about the math quiz examples. Consider the average cost per MWh for the options (each with 20MWh of consumption).
A $660/MWh for 20MW peak with no battery
B $156/MWh for 20MW peak with 19MW/19MWh battery
C $210/MWh for 5MW peak with no battery
D $132/MWh for 5MW peak with 4MW/16MWh battery
E $90/MWh for 1MW peak with no battery

So here is the question, what is the marginal cost of delivering one more 1MWh of charge at 1MW? This depends whether it in on a peak hour or not.

On peak
A&B station is maxed out at 20MW (of charging) so marginal on-peak is not available.
C,D&E increases demand peak by 1MW ($600) plus energy ($60), combined $660/MWh marginal cost (just like scenario A average).

Off peak
A&B marginal $60/MWh, highly available 23 hours out of 24.
C&D marginal $60/MWh, available at least 20 hours out of 24.
E marginal $60/MWh, availability limited to just 4 hours out of 24.

So how would Tesla need to price on-demand charging. (Let's continue to ignore other cost of charging stations.) The price needs to be based on average power cost. Tesla may need to break even in scenario B. So a price like $160/MWh or 16c/kWh would be reasonable.

But after pricing on average cost there still is an opportunity to deliver charge at substantially lower cost to anyone who is willing to charge off-peak. If you offered that off peak price to the general public, then you'd have to raise the on-peak price too, just so that the average price covered the average cost. So you get to a pricing plan that is prohibitively high on-peak, which undermines the value of having 20MW charging capacity available. So you are really limited in how much you can resolve this with pricing to the public. If you can come up with a counterexample to this point, I'd like to see it.

So there still remains a wonderful opportunity to leverage low cost marginal charging that cannot be offered to the general public. If Tesla were to run its own fleet, it could simply avoid charging at a rate high enough to increase the demand peak. Generally this would only impact about 1 or 2 hours out of a 24 hour day in scenarios A thru D. Scenario E is already highly unlikely to be the case where most charging is on-demand.

So think of it this way. Tesla breaks even on selling charges to the public at about 16c/kWh, but it can charge its own fleet at 6c/kWh on a stand-by basis. That savings of $100/MWh is basically the potential profit to be earned if the public price is break-even. Under this pricing strategy, Tesla's profit margin on charging increases with the fraction of self-consumption. If the TT fleet only consumed 10% that would be profit of $10/MWh delivered. But at say 50%, it could be $50/MWh. This of course is in addition to whatever TT earns hauling freight. We're talk about what TT earns just for charging on a stand-by basis.

I hope it is now clear why I think stand-by charging is a substantial opportunity for Tesla. This is the case even when the only power source is the grid. It remains so when we bring in local solar and wind PPAs. I'm afraid in our conversation the complexity of integrating renewables was obscuring the value of stand-by charging more than it was illuminating it. Even in the simple case where we only charge from grid power, there is substantial opportunity to leverage marginally cheap power. We get even lower marginal cost opportunities when renewables enter the picture.

 

[neroden]

It's absolutely not clear to me why you think that "stand-by charging" is a substantial opportunity for Tesla.

I agree that Tesla will shift its internal trucking so that its own trucks charge off peak where feasible.