Is there any way of getting Tesla or supercharge.info to identify V4 sites for the benefit of non-Tesla users with different charge port locations ?Fleet South is open. Just been informed from HQ.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Wiki UK and Ireland Supercharger Site News
- Thread starter WannabeOwner
- Start date
It will compared to a V2 site, considerably so.
On V3/4 the infrastructure can effectively shift 250kw to any stall up to the the sites available capacity. V2 you are stuck with 150kw pairs.
If you have 8 V2 stalls you’ll need 0.6mw grid connection and everyone gets up to 75kw when the site is full regardless of what each car is actually pulling.
On V3, if you had the same 0.6mw grid limit everyone could get up to 250kw. Say you had 8 cars on the charger, 4 of them were pulling 50kw, well that means the other 4 could pull 100kw. Or if there was only 2 cars on the station, they could both get up to 250kw.
Now, V3 has more than 75kw per stall but from what I understand it’s nothing like 250kw and it’s the clever load balancing which keep everyone charging quickly.
Sean.
Active Member
The latest cabinets I've seen at V4 sites have a maximum AC input of 385 kVA. There is of course the DC power sharing bus between cabinets but it's still going to be an absolute max average power stall of 96 kW. In reality prob no more than 90 kW per stall max average output with headroom and losses to be taken into account.
Recent V3 cabinets have 385kVA@480V input (for 4 stalls) compared to V2 with 160kVA@480V so V3 about 20% more input per stall. Possibly V3 are slightly more efficient, but there's only room for that to be a couple of percentage points of difference.
More significant is the sharing, where V2 shares only in units of 25% of the cabinet capacity, while V3 has fine grain sharing, and V2 is only sharing with the adjacent stall while V3 is at least sharing in groups of 4 and potentially pooling the entire site capacity if the inter-cabinet links are used.
The granularity of V2 means that whenever a car is given all that it can take, some of the input capacity is 'wasted' (in the sense that it cannot be used even if another car would like to share it). If both cars connected to a V2 are fully satisfied, then the V2 is wasting on average half a capacity unit per stall, so on average a whole capacity unit (25%) per cabinet - or in other words, average capacity usage would only be 75% of nominal.
That is a pessimistic estimate, because the case where the cars are being throttled (given less than they want) avoids this wastage due to granularity, but in this case the cars are charging more slowly than they are capable of, and any whole capacity units elsewhere on the site are being wasted. If for a rough estimate we assume those factors cancel out, then it would be reasonable to suggest that the maximum practically achievable throughput at a V2 site allowing for sharing inefficiency is 75% of the nominal rating. With the V3 site having 20% greater capacity per stall and not suffering from sharing inefficiency, that gives a comparison of 120% vs 75%, so the V3 site having more than 1.5 times the usable throughput of a V2 site with the same number of stalls.
That is considering the case of a site fully occupied and queueing, where the total power coming in to the site is the limiting factor. There's an additional advantage of the V3 site under fluctuating demand. When the site is not fully occupied, the V3 can charge some cars faster than they would on the V2 - both the case of cars that can charge faster than 145kW, and the residual bad sharing when drivers fail to pick the optimum spare space on V2. This means that the V3 site has less demand needing to be served during the busy (full and queueing) period, because some cars have filled up and departed before the start of the busy period while on V2 they would still be there.
Combining those two effects - and only at a 'finger in the air' level of accuracy given the huge assumptions - you might estimate that a V3 site has about double the potential throughput at a reasonable level of service (wait time) compared to a V2 site with the same number of stalls.
Finally, there's the effect of larger sites. Queuing theory dictates that a site of double the size can serve more than double the demand at a given quality of service. That's not so such a big effect with the size of sites we are talking about nowadays - going from 1 stall to 2 stalls is a huge factor (maybe 20), going from 2 stalls to 4 stalls is still quite big (maybe 6 times), but by the time you are talking of an expansion from 8 to 16 it's getting close to linear. Still something to be borne in mind when talking about site expansions.
More significant is the sharing, where V2 shares only in units of 25% of the cabinet capacity, while V3 has fine grain sharing, and V2 is only sharing with the adjacent stall while V3 is at least sharing in groups of 4 and potentially pooling the entire site capacity if the inter-cabinet links are used.
The granularity of V2 means that whenever a car is given all that it can take, some of the input capacity is 'wasted' (in the sense that it cannot be used even if another car would like to share it). If both cars connected to a V2 are fully satisfied, then the V2 is wasting on average half a capacity unit per stall, so on average a whole capacity unit (25%) per cabinet - or in other words, average capacity usage would only be 75% of nominal.
That is a pessimistic estimate, because the case where the cars are being throttled (given less than they want) avoids this wastage due to granularity, but in this case the cars are charging more slowly than they are capable of, and any whole capacity units elsewhere on the site are being wasted. If for a rough estimate we assume those factors cancel out, then it would be reasonable to suggest that the maximum practically achievable throughput at a V2 site allowing for sharing inefficiency is 75% of the nominal rating. With the V3 site having 20% greater capacity per stall and not suffering from sharing inefficiency, that gives a comparison of 120% vs 75%, so the V3 site having more than 1.5 times the usable throughput of a V2 site with the same number of stalls.
That is considering the case of a site fully occupied and queueing, where the total power coming in to the site is the limiting factor. There's an additional advantage of the V3 site under fluctuating demand. When the site is not fully occupied, the V3 can charge some cars faster than they would on the V2 - both the case of cars that can charge faster than 145kW, and the residual bad sharing when drivers fail to pick the optimum spare space on V2. This means that the V3 site has less demand needing to be served during the busy (full and queueing) period, because some cars have filled up and departed before the start of the busy period while on V2 they would still be there.
Combining those two effects - and only at a 'finger in the air' level of accuracy given the huge assumptions - you might estimate that a V3 site has about double the potential throughput at a reasonable level of service (wait time) compared to a V2 site with the same number of stalls.
Finally, there's the effect of larger sites. Queuing theory dictates that a site of double the size can serve more than double the demand at a given quality of service. That's not so such a big effect with the size of sites we are talking about nowadays - going from 1 stall to 2 stalls is a huge factor (maybe 20), going from 2 stalls to 4 stalls is still quite big (maybe 6 times), but by the time you are talking of an expansion from 8 to 16 it's getting close to linear. Still something to be borne in mind when talking about site expansions.
My feeling is V3/V4 sites are fast enough that most cars are well past peak charge rates before most drivers have finished their coffee even when all stalls are in use.
Is this view of how V2 shares accurate still? It certainly used to be the case that you would get different power based on how much the first charging car was drawing, but that changed a couple of years back. As soon as a car connects next to you on a v2 you go to about 64kW now, sharing the load, and it only increases if they disconnect.
cezdoc
Member
In the case of Supercharge.info this has been raised in its forum discussion; a similar debate was had when V3 started appearing. As I understand it, it's hard to do because it requires changes to the database structure. Previously, a filter for max power was added which allowed V3 to be differentiated from V2, but that won't work for the current V4 stalls which I assume are powered by V3 cabinets.
That said, they managed to add a checkbox for Tesla Only / Tesla + Other EV's so it evidently isn't impossible. It may be that all open-to-all sites are ultimately swapped to V4 stalls so the issue may go away (something similar happened when dual CCS/Type 2 cables started being added to V2 sites - eventually all sites were done so the need to flag it disappeared).
For Tesla it would be easy to update the app instructions but I suspect most folk would prefer to use contactless to initiate a charge (I know I would, based on bitter experience with other charging networks) so the set of people (a) using the Tesla app and (b) reading and acting on the charging etiquette instructions will, no doubt, be pretty small.
Or if the other driver hasn't preheated, I have been at a V2 and doing well and someone has parked next to me without my speed dropping.
Supercharger Info has incorrect POWER on plenty of the sites - they could do with sorting that out (e.g. synchronise against Tesla's data)
Poor database design / APP then, if they cannot easily accommodate database changes. Maybe I should speak to them ...
I had a look at V2, V3 and V4 sites on Tesla's "Supercharger finder". The old ones show, eg.., 150kW ... but both V3 and V4 appear to show 250kW, so I cannot see a way to distinguish them based on Tesla's public data. They may make available a more comprehensive list though?
I'm pretty sure that what I said (units of 25%) is still true, as it's a fundamental hardware feature. How those units are shared between the two stalls is a software matter and has changed over time. Taking away the first-car-priority potentially changes the probability of the various scenarios arising, but it's not clear how it changes the overall analysis.
The old algorithm - 2nd car gets at least one unit, first car has priority over the remaining 3 units - makes it very likely that the first car will be wasting part of a unit, but fairly unlikely that the 2nd car will be wasting. Conversely, if the new algorithm gives 2 units each unless one car is using less than 1 unit then that cuts out wastage provided both cars can take more than 2 units (~70kW), but increases the wastage where cars are below 70kW.
So the new algorithm reduces wastage in the presence of mostly fast-charging cars (which is presumably why they made the change after Model 3 became predominant), but makes it worse in the case of slow charging cars (both third-party like Jason71's Fiat 500e, and aging Tesla Model S like mine where the charging curve has got worse and I spend quite a lot of time below 70kW).
That was always the case - in fact GRiLLA is suggesting that the modern behaviour makes that less likely than before (but still possible).
Yes, it's unlikely the hardware has changed, but the real user experience is that
v2 - Not sharing 120-150kW peak (actually I've never seen more than about 138 in real life). As soon as you are sharing it's 60-70kW max.
v3 - Never noticed any effect from sharing, car seems to be the bottleneck. I'm sure it's there, just is handled in a way you don't notice.
In neither model does the ability of the car that arrives make much difference. In a v2 it's the number of bays in use, in a v3 you largely get whatever you car can take.
As a supercharge.info editor and part of the dev team I just wanted to comment on this.
Synchronising the data is not easy as Tesla has multiple different sources of the supercharger data. Tesla supercharger map Web site, Tesla supercharger find us page, tesla supercharger API, in car supercharger data, tesla app supercharger data, etc. These can and often are all be different (shockingly bad from tesla as we all think as a world leading technology company) and some are more easy time to synchronise with than others. There is no simple 'synchronise' the data solution.
There are nearly 6000 locations now and data is sometimes wrong there. We rely on user corrections to inform us of these.
The database needs a full redesign to get things to show what we want. These things take time.Poor database design / APP then, if they cannot easily accommodate database changes. Maybe I should speak to them ...
The design was fine and fit for purpose when it was created 10 years ago but now we have many different stall types, split sites with different power, stall and connector types, expanded sites, open to all (which was hacked onto the original database design), and many more with different complexities for superchargers in different parts of the world. that have to all match up.
Take for example this very thread. It was fine when it was created to have all superchargers in the UK when there were only a handful but now there are hundreds this 6,000+ post megathread is not really fit for purpose where you can't easily find anything about a supercharger - compared to the design of the US supercharger threads where each having separate pages. Changing it here would require someone putting the work in.
The same for supercharge.info dev the work it takes time to redesign everything and someone has to be willing to volunteer.
If you want to step up to help you are more than welcome. Always open for new volunteers.
Last edited:
Electron Mike
Member
Drove past rownhams today and the gridserve units have been installed where I've highlighted green. I didn't see anything Tesla related but I was driving past at 70. But at least there is some sign of progress, even if not for the Tesla's
Attachments
Is it bad that I am more interested in the McDonalds drive through. Having to get out of the car at Rownhams for anything bar Costa is a first world problem
I was there last weekend. The concrete bases were in but the ground works were not fully completed (backfilling and tarmaccing required). I expect this would be now completed.
3 cabinets were on site but no sign of the 12 tombstones.
Edit: the transformer has been in for a few weeks. Tesla say Q1, they could be ready before then if things get connects up quickly.
Sean.
Active Member
cezdoc
Member
Regarding this thread, agreed it is huge but the search on TMC is very good. Usually I can quickly find information about a specific site using search-within-thread, so long as I can dream up enough alternative names for a particular location (e.g. Hamilton/Radstone).
I try to help keep the wikipost up-to-date, this details all sites opened this calendar year and all sites known to be under construction, with references to thread posts for cross-checking site information. There's also a link to summaries for 2021 and 2022 so all sites built/expanded in the last 3 years are covered by three posts.
Granted the thread itself goes back much further, and reading the annual summaries on a mobile device might be hard work, but I think it's still useable even in its current form. I won't promise to keep updating it for ever, but while I'm an owner with a vested interest there's enough motivation for me to keep visiting.
Thanks for your work on supercharge.info and everyone else involved - it's a great resource.
Similar threads
