Free Destination Charging: 50x L2 80A Stations @ Caltech, Pasadena, CA

[FlasherZ]

For this deployment we are using single circuits to each station. The circuits do share a common large 4 inch conduit/gutter before it branches out to 1 inch conduits for each station. An added bonus of flexible pilot signals is that if a station is too far away, we could just software the limit the station to the calculated capacity, like 68A instead of 80A.

For NEC 310.15(B)(3)(a), wouldn't it seem that 160A would be overkill?

In my opinion, yes, but unfortunately inspectors have to go by the code. The reasoning is the potential heat build-up trapped in the conduit from many circuits carried inside one raceway. It was originally intended to deal with the practice of bringing a bunch of smaller branch circuits through a single huge conduit, but the way it's written makes it apply here too.

(3) Adjustment Factors.
(a) More Than Three Current-Carrying Conductors. Where the number of current-carrying conductors in a raceway or cable exceeds three, or where single conductors or multiconductor cables are installed without maintaining spacing for a continuous length longer than 600 mm (24 in.) and are not installed in raceways, the allowable ampacity of each conductor shall be reduced as shown in Table 310.15(B)(3)(a). Each current-carrying conductor of a paralleled set of conductors shall be counted as a current-carrying conductor.

There are some exceptions: cable trays have a special section dealing with that 392.80), raceways less than 24 in. are exempted as are raceways shorter than 10 ft managing a transition for underground conductors entering/leaving a trench. Armored cables (AC/MC) have special provisions, and XHHW-2 insulated conductors (thicker XLPE insulation) are exempted as well. Not sure if you're using XHHW-2 wire; if not, none of these exceptions seem to apply here.

This is why in many commercial installations you'll see a bunch of parallel conduits running along the wall instead of one giant conduit. You can carry 3 circuits in a smaller conduit (6 current carrying conductors) at 80% of the wire's rating vs. having to reduce everything to 50% for 5-10 circuits.

 

[MITE46]

How about parallel conduits connecting junction boxes? I would imagine the calculation is for the conduit span? As in the junction boxes are not counted?

BTW, thanks for all of the info, it's very helpful!

 

[FlasherZ]

How about parallel conduits connecting junction boxes? I would imagine the calculation is for the conduit span? As in the junction boxes are not counted?

BTW, thanks for all of the info, it's very helpful!

The calculation is for the conduits between the junction boxes. Where you have junction boxes, they're governed by a different article in the code (314), with its own calculations for minimum size, fill maximums, etc.

 

[MITE46]

It's been about two weeks since the soft opening and quite a few people have stopped by...There are the regulars and also visitors...but any how, been dealing with WiFi issues for the last week, enough so that we have re-designed the log collection mechanism. Now everything is stored in a local sqlite and then replicated up to the cloud, this way after a network disruption, which happens a lot, we recover the missing data...More on that after some testing. We're able to push the updates over the air to the stations, much like the Tesla updates...

Here are some more details on the touchscreen interface...Our splash screen...more of a screen saver animation etc.

Once the screen is tapped, we have a quick survey that allows you to pick your car, specify desired miles, and stay duration...The point is to gather a much information such that the scheduling algorithm can figure out the best plan to meet everyone's requirements. We are taking into account the fact that some people might lie, or just are unsure about the accuracy of their choices...

The reason why we have drivers pick their car is so we can approximate Wh/mile. Of course elevation and weather will change that number, but again it is just an estimate...

We should integrate the Tesla reservations page so people could directly order their car on the touch screen...or buy any EV on that page lol.

So on Monday's we see the heaviest usage...I guess that is because most drivers run almost empty on the weekend and then limp in on Monday...Here's a tale of two Tesla MS...

Figure it would be a good time to test cranking up the amps...adjusted to 60A halfway. There is a slight dip, I'll have to see what happened in the logs...

Some more cool stuff coming soon!

 

[GSP]

Very nice work! Thanks for sharing it here at TMC.

I noticed that EVs that are no longer in production, but still operating on our streets, are not selectable. If someone arrives in their Coda, or Th!nk, or Fisker Karma, or Ford/Azure Transit Connect EV, etc. Or an ICE car they converted to electric, they do not have the option to select "other." Adding such an option should improve the accuracy of your data, since they will not be forced to select an incorrect car.

GSP

 

[FlasherZ]

Love it - what a great project!

 

[Ulmo]

This is what we need! Not just one or two stations, but a large enough number so that you don't have to worry about finding a spot.

I especially like how they saved money on the installation by making it more convenient: so many stalls were put in that they tell you you can park there all day, and then they install lower cost electronics and take care of the allotments to turn lower cost * more time (because more quantity and less moving) into a high enough state of charge for their parkers.

 

[Ulmo]

These Kali folk are quite progressive. Our local college has a petroleum engineering department and don't want to scare of the paying habibis with Panameras
and M5s, so many pleas for even a level 2 charger have fallen on deaf ears.

This is all about pandering to the wrong crowd. This is why we have oil wars, terrorism. No disrespect meant to you at all, Kandiru. I'm speaking of the college's long term planning decisions. They could coordinate a time to downsize the petroleum engineering department at the same time refineries start going offline, which isn't today but is happening long-term. For now, someone needs to know how to keep those refineries open. I actually think a college like this stuck in the past should put in their admissions literature "no EV's at this campus!" so no children sign up for college there and when they get there find out they are outcast for being modern.

 

[LetsGoFast]

Fantastic project, very well documented!

 

[Ulmo]

470/432*208 or 226 Volts. ... 504/432*208 or 243 Volts

Is that 470*432/208, rather than 470/432*208, and 504*432/208 instead of 504/432*208?

The real issue with using the other taps is that your L-N voltage can become dangerous if anyone wanted to use it. 226V L-L on the wye ends up becoming 130V L-N, which can be a bit too hot for some appliances. As long as no one ever intends to connect 120V loads to that transformer, then there's no issue with using the higher boost taps.

So.... leave a note? I see the student figured this out readily by saying they're going to get signage.

@FlasherZ: I'm surprised there is nothing in the code that requires +-10% of 208v. Good point on the L-N...looks like we will have to invest into some permanent signage on the panel...

Every time I read the code, I have noticed that it is very lenient on application; they know you're going to use it for something. They are most interested in safety. I find this tendency in the electrical code a good thing. I see now that FlasherZ has answered already, and threw in his professional superior knowledge

Cleaned up some of the graphs today...here's a sample station, I think a Tesla then a Volt or something...

Nice! But I personally think there is such a thing as too clean. I liked your original graph's faint dotted lines to know where the graph points line up, so I can track what each curve means, rather than just looking at it like it's a pretty view.

In the US, non-armored cables must not be exposed where they are "subject to physical damage". In residences, this is typically interpreted as anything below 2.5 meters or so. Commercial installations require even more protection.

One question for you, though - do you have a single circuit to every EVSE or are you sharing a small number of large circuits with multiple EVSE's? I saw that you said you were using #4 for 80A circuits (up to 64A charging current), but I didn't see one conduit per station there.

NEC 310.15(B)(3)(a) table shows the derating figure for number of current-carrying conductors. 4-6 is 80%, 7-9 is 70%, 10-20 is 50%. Each EVSE uses 2 current-carrying conductors - so if a single conduit supplied 5-10 stations, each one on an individual circuit, you'd need a wire capable of carrying 160A. That would be a 1/0 conductor for THHN2 (90 degC). How are you working with that limitation (it's a fairly big one in a garage, as you'd imagine, and I get asked that question a lot).

I had the same question. My answer to myself was that perhaps in the next few garages they do they'll put in more subpanels and work on centralizing distribution as much as possible. But in reality, I think this question will come up again and again, and there needs to be some way to make it affordable for the whole country in all of its parking lots, and a lot more engineering detail put into it.

For instance, an engineer could calculate the various possible temperature situations and electrical consequences. What if it were a 115ºF day, and every single stall got a new model something or other that draws maximum power at the same time parked there? Calculate the safety factors, and come up with engineered specifications, and various optimizations to the installation (both better quality and lower cost, sometimes both, sometimes one and not the other, for various pieces). The students ought to use this as a learning experience to learn how professional construction engineers do this all day long at many places, but that's kind of underneath the level of many CalTech students (but they'll need to know who they're working with and respect their work).

One thing I'd be tempted to think of would be to put a T in a few parts of the conduit with assistance from HVAC engineers, and pump cool air through there, and put thermostats in the conduit to both control the HVAC AND to lower the use of electricity when it gets too hot (either HVAC failure or runaway conditions that an installed HVAC wasn't designed to overcome, or more likely, both at the same time). But this opens up spread of fire potential and static electricity problems (zap!). Hmmmmmmm .... ambient cooling via the chaseway is very efficient and relatively well known and safe. It's pretty hard to beat that. An engineer could suggest heat sinks on the conduit at key locations :/

Edit: I just thought of the first thing I would try when designing this without CAD: I'd calculate the effectiveness of using a bigger bus and smaller branches, i.e., single circuit going to all of the EVSE spots, and a circuit breaker coming off of the single circuit to each EVSE. So, like a water pipe. Thus, no derating. But that would have to be a big honking wire. Each one of those boxes where they are going under the concrete beams would be a good place to branch out a few breakers, and then whenever they down-step the wire size for the remaining bus (main circuit run), another breaker sized for that leg. That would require breaker rules, like the 3' clearance. STRIKE OUT (WRONG): So, I think this may require an update in the code to require more flexible installations; you don't need every EV user able to reach the circuit breaker from their car in a garage, and you can tow the car out to get the needed clearance to reach the breaker box. CORRECTION: they need disconnects at each charger, right? (So otoh) the disconnect can be part of the breaker, and that's going to be within reach of the car user anyway, so combine those two things there. Still, trying to save on wire cost for the T branch to each EVSE spot would be a few feet of fat wire or another breaker in the junction. Am I babbling? No, I'm not. Just T out same size wire for branches to the shutoffs, unless that's too far. But you want to save on wire size for further legs as you get further away, so you DO want breakers out of reach of the cars, AND you could also save on long branches from the main run this way too. So, not so much of a correction as a clarification.

Does the code allow for hard to reach breakers as long as the shutoffs are easy to reach?

 

[Cottonwood]

Let's do a little math on tolerances. If the input is 470 Volts and you use Tap 7, the output will be 470/432*208 or 226 Volts. Let's look at it another way, if the input rose to the highest this transformer expects or 504 Volts, then the output rises to 504/432*208 or 243 Volts, well within ranges normally seen by J1772 vehicles.

Is that 470*432/208, rather than 470/432*208, and 504*432/208 instead of 504/432*208?

I believe that I have the math correct for calculating the actual output Voltage on the nominal 208V secondaries.

 

[MITE46]

Another feature of the Caltech's Adaptive Charging Station is that it comes with a gravitational wave detector...

Just kidding...however, our servers are in the same room as LIGO on campus...anyway...it's an ultrasonic sensor just like the one that operates the PDC...

These sensors need a relatively flat surface to operate accurately, so I experimented with different heights...

Well it turns out that the easiest thing to do is just to mount it on the side so you can adjust the angle...We have them about 30 deg slanted downwards...This is also our updated faceplate...

Why would we have an ultrasonic sensor you ask? There are many reasons, here's a couple:

1. Improved accuracy of online station availability. You pull out your phone and see there is an available station nearby. You limp in with only a few miles left and find out that it is ICEd....The ultrasonic sensor can detect obstructions to the millimeter up to 30 feet and provide more accurate station availability online. For example, you can detect: blocked + charging, blocked + not charging, blocked + finished, vacant, etc.

2. Dynamic parking spot assignment. Face it, parking spots are more important than EV stations, ask any landlord, and they will tell you their view. At Caltech, the same is true, we would like to provide mass EV charging without designating/painting EV only parking spots. By using ultrasonic sensors we know exactly how many of the 54 stations are charging, available, and ICEd. However, we can always ensure that 10 EV spots are open by changing the signage on the LCD screen dynamically as spots fill up, etc.

 

[McRat]

Great work!!!





The nominal Voltage for J1772 L2 charging is 208-240 Volts, but we know that there is tolerance. I believe that the Tesla AC charger can take 100-250 Volts AC and versions can go up to 277 Volts nominal in Superchargers.

Let's do a little math on tolerances. If the input is 470 Volts and you use Tap 7, the output will be 470/432*208 or 226 Volts. Let's look at it another way, if the input rose to the highest this transformer expects or 504 Volts, then the output rises to 504/432*208 or 243 Volts, well within ranges normally seen by J1772 vehicles. I often see 250 Volts or very close at my Pagosa house when the Solar system is cranking out 20 kW to match the Tesla 20 kW load.

As long as there are no loads that actually expect 208 Volts, I say go for it. 9% faster charging is good for everyone! The only complication is that you will have to take the new output Voltage into account in your calculations for keeping the total kVA within limits on each transformer, but that should be easy.

I hope FlasherZ weighs in on this in terms of any possible code issues.


I can't wait to see results of a load test with active limiting to stay within your kVA limits. BTW, what are those limits?

Keep up the great work!

This is something I'm curious about, sort of a sidebar.

Why couldn't you just use single legs to neutral if you are losing voltage due to distance? You might be 265 or less.

My shop is "480vac 3ph", but actually sees about 493vac at the distribution point. My single legs read about 281 vac. I sort of wish I had voltage drop (not really), so I could run a L2 charger off a single leg instead of 208 (2 legs when stepped down to make 120 x 3).

I thought about experimenting with running a L2 on a "277" single leg, but that's actually over the 265vac specification. One thing I do notice, is that my 240vac 1ph at home charges faster than my 208vac 3ph at work.

 

[McRat]

It was a sorry oversight during the birth of modern EV charging that 480vac 3ph was never considered as a common source of power, since most commercial buildings have it. Just very minor changes to equipment would have allowed 277 single leg. Less wire costs, less step down losses, less run drop, light fixture compatible, and easier high kW ratings. Nearly all the lighting you see inside commercial building is 277 for a reason. It costs no more than 120v for the fixtures, the wiring is cheaper, and they cost less to operate.

 

[Cottonwood]

This is something I'm curious about, sort of a sidebar.

Why couldn't you just use single legs to neutral if you are losing voltage due to distance? You might be 265 or less.

My shop is "480vac 3ph", but actually sees about 493vac at the distribution point. My single legs read about 281 vac. I sort of wish I had voltage drop (not really), so I could run a L2 charger off a single leg instead of 208 (2 legs when stepped down to make 120 x 3).

I thought about experimenting with running a L2 on a "277" single leg, but that's actually over the 265vac specification. One thing I do notice, is that my 240vac 1ph at home charges faster than my 208vac 3ph at work.

I have thought about this situation. The Tesla HPWC install manual at Tesla HPWC Install Manual 80-12A has some info. On page 3, the manual states that the maximum Voltage of the HPWC is 250 Volts, while the max for the car is 265 Volts. On page 7, it does show that you can use the HPWC with single-ended power, L-N, as opposed to the normal double-ended power, L-L, used in the U.S. Of course, it is well known that the normal connection in Supercharger cabinets is 277V L-N, which is of course, 480V L-L.

240V is 15% more than 208V and it is tempting to go for that extra charge power. In the end, it is a fair amount of equipment/expense, and sometimes pushing of ratings for a fractional increase in charge speed.

I have come to the conclusion, that in most situations, going for that extra 15% is just not worth it. If you want to get a little faster charging at your shop, look into boost transformers, or perhaps get one of those multi-tap 480V-208V transformers like they have at CalTech. A 20 kW Boost transformer costs $1-3 thousand, including install. See Crowd Funding a SCH100 EV Charging Station in Grand Falls-Windsor, NL - Page 2 for some discussion.

In the situation at CalTech where upping the Voltage only involves selecting different input taps on the 480V-208V transformer, then I say go for it. The only risk is that the nominal 120V L-N connections will go up in Voltage. With correct labeling on this install that only uses the nominal 208V outputs, the advantage far outweighs the ever-so-slight risk.

 

[McRat]

I'm a NFG when it comes to EV charging. But I do understand a just a little about how commercial building are wired. One really tragic thing about current (har) technology for cheap L2 charging is you cannot use your parking lot lighting or outdoors light as a source circuit for L2 installations, which is a crying shame. There is no 120V so L1 is not an option, you have just 277 available. How simple and cheap it would be to put a L2 low powered (6.6kW) if you have light poles in parking lots. Just pull heavier wire and install on the light pole. But they never thought about how the real world is wired when the specs for onboard chargers were established. This would give all EVs at least 100 miles during a work shift with very low (sub $1000) total parts costs. To "ICE" a light pole in a parking lot requires 4 cars to ICE it instead of 1 or 2, and the charger is protected.

 

[chuck-h]

I love this.

In fact I am inspired to build a smaller cooperative-charging system for my condo using similar hardware.

@FlasherZ I am aiming to be squeaky clean on electrical code compliance. I can get a UL recognized version of OpenEVSE from Wattzilla & my electrical box internals will look a lot like the picture in MITE46's ultrasound posting above. Any thoughts about separation and isolation between line voltage and signal circuits in there? I'll need to run some external class 2 remote wiring into the Raspberry Pi as well.

Cheers
Chuck

 

[chuck-h]

@McRat RE: 240V charging from 277V line: would a $200 buck/boost do the job?
http://www.walmart.com/ip/ACME-ELECTRIC-T113073-Transformer-Buck-and-Boost-16-32V-1000VA/40830533

 

[hcsharp]

... Any thoughts about separation and isolation between line voltage and signal circuits in there? ...

I was wondering about this as well. Do you know what's causing the WiFi issues?

Great job on this project!

 

[FlasherZ]

I love this.

In fact I am inspired to build a smaller cooperative-charging system for my condo using similar hardware.

@FlasherZ I am aiming to be squeaky clean on electrical code compliance. I can get a UL recognized version of OpenEVSE from Wattzilla & my electrical box internals will look a lot like the picture in MITE46's ultrasound posting above. Any thoughts about separation and isolation between line voltage and signal circuits in there? I'll need to run some external class 2 remote wiring into the Raspberry Pi as well.

Separate conduits for communications vs. power wiring. This is for inducted current reasons as well as overall voltage safety.

800.133(A)(1)(d):
"d) Electric Light, Power, Class 1, Non-Power-Limited
Fire Alarm, and Medium-Power Netvvork-Powered Broadband
Communications Circuits in Raceways, Compartments,
and Boxes. Communications conductors shall not be placed in
any raceway, compartment, outlet box, junction box, or similar
fitting with conductors of electric light, power, Class 1, nonpower-
limited fire alarm, or medium-power network-powered
broadband communications circuits."