W8MM

@gnochi Great write-up. I'm an EE whose wife has a Tesla Roadster 1.5 and a Model S.

Your distillation of all the inter-connected considerations was a pleasure to read.

ocgarza

[QUOTE=gnochi;14110260]@ocgarza:

1) There are many reasons I drive a Porsche instead of an EV. This is one of them.

Well said...

4) That's an interesting topic. 95% of the car could probably be done at an indy - wheel bearings, transmission rebuilds, cooling line replacement, various interior bits and bobs - but I just can't see a way for the core powertrain to be serviced outside of a dealership. Perhaps they'll encourage dealer-level service with a core credit on those elements, even after the warranty has expired?

Got it. With the EV electrical core it almost sounds like a repair job for an ee.


What about pure EVs involved in accidents? Any electrical danger to the driver and is there an emergency kill switch for first reponders?



With an EV, you really do care about that 4kW AC compressor, since it cuts your 60mph range by about 25%! As such, you'll only run those elements when you need to, with only as much power as they require, instead of accepting the additional parasitic losses.

This is another big issue here in the deep south Texas (only a few hours north of Mexico). It is April 18 but my car air conditioner has been on since mid February.


Thanks again, great reading...

gnochi

@W8MM

Thank you! Engineering at its finest

@ocgarza

You're very talented at asking leading questions. Thank you!

Service

An EE is a big help in the repair process - usually they have tools to tell roughly where in the system a failure occurred, but it makes a big difference whether something failed closed or open, or with a slightly off resistance value. Sometimes you need to make the system try to respond in a certain way to determine whether it's a closed failure at module 5 block 3 or an open failure at module 5 block 4, but there are enough possibilities that you can't account for all of them in software. At that point the question of whether it's worth it becomes a big discussion.

Usually, the engineering team behind whatever system failed will want to collect the relevant elements, so they can do these analyses and figure out what needs to be changed to prevent those failures in the future. That's particularly important now - the modern electric vehicle is a mess of new and relatively unproven technologies tossed in a blender with a bomb.

So, you have something that is likely going to be ridiculously expensive to diagnose and repair. In addition, the engineering team likely wants the defective system. As such, there's a strong push for the manufacturer to in some way encourage people to get the system replaced at the dealership, and a core deposit (or unlimited mileage warranty) is an excellent way to do that.

Accidents and Safety

Regarding accidents, there are two main dangers: the electrical hazard, and the fire hazard. Surprisingly, the electrical hazard is almost negligible.

Electrical Hazard

If the engineering team behind the powertrain has done their due diligence, the HV system will be floating relative to chassis ground with some degree of insulation resistance (min. 1kOhm per volt of HV system voltage, and a good target is min. 10x that for CYA reasons) and isolation (5x system voltage due to inductive loops). As such, this requires a minimum of two simultaneous isolation faults to generate an arc - taking an extreme example, a crash that damages the unswitched pack-internal highside and lowside busbar insulation and brings both into contact with the bare metal of the pack cover - or otherwise cause a potentially electrically hazardous situation to an occupant or first responder.

This is also a factor in why *every* piece of conductive material that isn't part of the HV electrical system is grounded to Chassis - the resistance point-to-point on the chassis is on the order of microOhms, and the resistance point-to-point in the human body is about 200kOhms. Even if there's a full-pack hard short to chassis, that 1) won't last very long before the cell-level fuses vaporize and b) won't electrocute anyone, even if they're touching two different sections of chassis, because effectively no current will pass through them instead of said chassis. Compare that to if an ungrounded conductive steering wheel shorted to unswitched pack highside and an ungrounded conductive brake pedal shorted to unswitched pack lowside simultaneously. At 400V pack voltage, that's 2mA DC, which will hurt, but shouldn't stop your heart. (The AC ripple at the inverter switching frequency and nominal motor current might.)

So, let's take a slightly different case, and say we have a single hard short to chassis. This means that a second short is required to make any current flow, and can be relatively easily detected. For CYA reasons, the car will probably stop working, tell contactors to open, and require a tow to a dealership to help prevent that second short from occurring while there are occupants.

As a related issue, it's fairly common for contactors to weld shut (due to sustained overcurrent that doesn't trip fuses), and cause the pack terminals to be live until the contactors are replaced. A service technician who isn't wearing safety gear and using uninsulated tools will feel a shock if they touch both terminals, and if they bridge the terminals they'll see an arc form. I've seen a pair of busbars with a wrench literally welded to both of them - as a demonstration unit for "what not to do and why". Arcs are extremely dangerous, and have a relatively high chance to permanently damage both vision and hearing within line-of-sight. As such, they need to be prevented when someone is actively working on the system, but are otherwise something to be avoided that probably isn't the end of the world (or the company).

The easiest way to do this, and stay safe if something does go seriously wrong?

• Wear your full PPE - HV-rated boots, nonflammable shirt and pants, category-appropriate rated oversuit, category-appropriate rated and tested leather-over-rubber gloves, earplugs, safety glasses, helmet, and category-appropriate rated face shield
• Use only category-appropriate rated insulated tools, and don't have uninsulated tools in the same work area at the same time
• Design the system so you can only physically access one live voltage on the HV system at a time
• Design the system so dropped screws/washers/nuts/tools can't bridge voltages - for that matter, don't use loose washers or nuts.

There are electrical kill switches and physical kill switches. By definition, physical kill switches allow physical access to the HV system. However, they do guarantee that, barring unusual circumstances, you can't have a full-pack-voltage short when they are removed. Not much of a guarantee.


Electrical kill switches work because all contactors used in HV systems are normally-open - they require low voltage DC current to close and allow HV power to exit the battery pack. As such, turn off the 12V to the system - which is never ever ever generated within the battery pack - and the contactors should spring open. (Unless they're welded, in which case it's a trip to the dealership and HV service in arc flash PPE.) In the event of an accident, the first thing the system controller should do is tell the contactors to open.



Brief case study: the Tesla Model S does not include a physical HV disconnect or kill switch. Instead, they route the LV lines to the battery pack in the trunk near the passenger A pillar and tell first responders to cut said lines. They also note, and this is very important, always assume all HV components are live at all times. Their logic is more or less as follows:

1. Electrical risk is actually pretty low
2. Electrical risk requires two simultaneous HV isolation faults
3. Physical disconnects allow access to live HV components
4. It is not necessary to disable high voltage to allow someone to be removed from the vehicle, thanks to the miracles of chassis ground
5. It is therefore riskier to include a HV physical disconnect that isn't guaranteed to help, than to exclude it in favor of a LV electrical system disconnect.
6. There is sufficient redundancy between the software system telling the contactors to open and the inverters to switch off, 12V power not being able to enter the battery pack if the lines in question are cut, separate high-side and low-side contactors (either of which can kill power to the pack-external system), and cell- and system-level fuses in the event of a pack-internal short.

Unfortunately, designing a system so that it is not possible to have any number of simultaneous failures is itself not possible. It's the world creating a better idiot, but with escalating car accidents and unlikely scenarios. 10^-8 failures per hour is considered safe for automotive use; I encourage you to look at ISO 26262 for further information. Note that this standard outright assumes that mechanical devices work if you successfully tell them to, so Tesla's system exceeds the standard.



Fire Hazard

You may have noticed me say things like "bomb" referring to HV battery packs, and I included a video of a thermal runaway in the thread opener. This is not an exaggeration. The more energy contained in something, the more exciting it is when that energy is released in an uncontrolled manner. Thankfully we're a few steps down from nuclear power plants.

There are a few ways a cell failure can present itself:

• Soft short: the cell slowly discharges itself and possibly throws the row out of balance, but this can be managed
• Hard short: the cell rapidly self-discharges internally, shorts out the row, generates a lot of heat, and hopefully doesn't enter thermal runaway before activating the internal CID and removing itself from the electrical circuit
• Thermal runaway: too much current goes through the cell before the CID or PTC can respond, the internal reaction is no longer slowed by the internal insulation, and you end up with a spout of fire out of the vents of the cell. This happens in some internal shorts at high state of charge, and in almost every mechanical failure at any state of charge

A well-designed battery pack can handle a single-cell thermal runaway without propagation to neighboring cells. This doesn't scale favorably with the number of involved cells, though, and a case in which multiple cells are mechanically damaged - by a trailer hitch or rebar through the bottom shield, or a high-speed pole impact crash - will almost always result in a full-pack fire that melts right through aluminum. I'm not going to follow this much further due to proprietary information, but here are a few things to keep in mind:

1. Due diligence is to ensure a first responder has a high likelihood of being able to remove an unconscious living occupant from an EV undergoing a full-pack fire.
2. It's possible for a body structure to absorb a lot of energy from a crash.
3. There exist structural materials other than aluminum.
4. People are, as a rule, more fragile than batteries.
5. A battery pack fire will almost certainly total a vehicle, and will usually burn it to the ground given enough time.
6. Compared to ICE fires, EV fires are much more rare per mile driven, burn at a lower temperature, occur at higher levels of structural damage, and have a higher minimum safety standard.

Other

AC since February? Ouch. Good thing you have ridiculously cheap gas.

Just out of curiosity, around what time of year does the temperature in Victoria first spike above 100F? I was in Austin for a few days last July, and while I expected it to be hot, I didn't expect it to still be 100F at 9:30pm.

<3mph

I'm going to have to read this a few times to get it, as the content is way out of my wheelhouse, but it's very interesting. Thank you! I really appreciate threads like this from experts like you.

Seneca035

subscribed... nice write up gnochi!

I don't think the Mission-E is a threat to Tesla with the Model 3 on the horizon. I'm very eager to see the production version of the Mission-E and Porsche's answer to long range charging infrastructure.

gnochi

@<3mph

Glad to provide! Let me know if you have any questions or if anything is unclear. (I might or might not be using this forum as a test audience for a more formal writeup... =D)

@Seneca035 Thank you!

Agreed; they are going for entirely different markets. I would be very surprised if anyone who was planning on getting either changed their mind and got the other.

Porsche announced at AABC last year that they were going to include the hardware necessary to charge their battery pack using 400V infrastructure; at the same time, they're strongly pushing 800V charging infrastructure for a lot of good reasons.

800V comes with problems, though. There's a phenomenon called "bearing currents" that causes premature wear on the bearings in electric motors; effectively, the motor shaft (which is theoretically floating and isolated from the windings) can never quite stabilize in potential because of the three phase power, and the small arcs that are generated cause pits in the grooves. This issue is exacerbated as phase voltage increases.

In addition, "common" HV components that can fit in a car tend to top out around 650V - IGBTs in the inverter and high power contactors in particular. Heaters and pumps tend to hover around 500V. You can get cables and connectors for 1kV. The problem is that anything in higher voltages is intended for locomotives and power distribution networks - and good luck fitting anything for those markets into a car.

This boils down to "creepage" and "clearance" - the higher the operating voltage, the more of a gap you need to leave between components that might be at different potentials. Just blocking line-of-sight doesn't work, as electricity can "creep" along the surface of an insulator. The following video is an extreme example, but the same concept applies at lower voltages. At an operating voltage of 400V, you need to protect against creepage and clearance at 2kV - about 1.5mm clearance and 3mm creepage - while at 800V you need to protect against roughly double that. (Note: these numbers change based on pollution degree, material specifics, local air pressure, etc.) This means that everything needs to be spaced out that much further, lest an arc form and lead to a short circuit.


All that said, Porsche has a solid engineering team. They have this figured out.

EDIT: Pressed enter too soon and forgot to include the good reasons for 800V, which basically boil down to current being hot, heavy, and unwieldy.

As I mentioned in an earlier post, heat is predominantly generated in electric systems through resistive losses - you may recall the formula is P = I^2 * R. In conductors, resistance is the length divided by cross-sectional area times the material resistivity (effectively constant). If you need a busbar of a given length - say, down the spine of a battery pack - cutting the current in half means that you generate one quarter as much heat at the same resistance. You can also use half as much material in cross section and double the resistance, cutting the heat generated in half, and also cutting the cooling capacity roughly in half, so everything evens out.

Copper is really heavy, so this exchange is almost certainly a good idea. (Note: you can also use other busbar materials; aluminum is more conductive than copper by weight, though not by volume. This is another one of those finicky optimization problems you keep running into in this industry.)

However, this doesn't apply in battery packs. If you're doubling the operating voltage and halving the required current, you still have the same number of cells. Instead of a 96s74p 18650 pack like the P90D, you'd end up with a 192s37p 18650 pack. At a given level of power, each cell is seeing the same amount of current, and therefore generating the same amount of heat, and therefore so is the vast majority of the pack. We can double-check this: at 0.049Ohm per bonded cell, the first example is 96*0.049Ohm/74=0.0635Ohm, and the second example is 192*0.049Ohm/37=0.254Ohm. In the 400V variant, we draw 500A to reach 200kW at 400V, with 15.9kW heat generation; in the 800V variant, we draw 250A at that same 200kW for a heat generation of that same 15.9kW.

During their AABC presentation, Porsche showed a slide with all of the busbars and electronics required for a 400V pack on one side of a scale, and the same for an 800V pack on the other. They saved something like 35% of the non-cell weight of the pack, which is incredible.

Charging is also a particular benefit to higher voltage. As I mentioned in an earlier post, you're limited to 300A or so for a supercharging plug that's reasonable to handle and use. If you have a 400V pack, you're limited to 120kW charging; you can double that to 240kW if you can tolerate charging at 800V*. This does effectively cut the recharge time in half - assuming your cells can handle charging at full power. We're not there just yet, but we're working on it.

*Supercharging plugs are commonly rated to 1000V - it's a nice even number that nicely aligns with Class 0 equipment, and rated to 1000V means it is tested without failure to 5000V. 800V can use the same plug - it would just need to communicate that higher limit to the charging station.

W8MM

@gnochi

Now you've really done it! Who even knew there were such hobgoblins as "bearing currents" in AC motors driven by IGBTs?

Fortunately, there are web references to explain the origin of such phenomena: https://library.e.abb.com/public/8c2...e_No5_RevC.pdf

Thanks so much for your in-depth treatment of electric vehicle technology. Well done.

Lorenfb

Have you forgotten the vehicle's rolling resistance (a function of the vehicle's velocity).

Lorenfb

And where did this number come from? How about less than 10 - 15%.

Lorenfb

Where did this number come from? How about less than 10 - 15%.

gnochi

Rolling Resistance
I haven't forgotten it, I've ignored it. The power required to overcome rolling resistance changes linearly with speed (ignoring the minor normal force changes from aerodynamic lift and downforce) - and for a "typical" set of tires, it's about (vehicle weight)*(0.03)*(speed). It does have a measurable impact on motor-to-ground efficiency, but it's about half the impact of a single really well designed gear interface.

Compare that to the power required to overcome aerodynamic drag - which, as explored above, varies with the cube of speed. Below about 10mph, aero drag is almost not measurable. At about 40mph, aero drag power is about equal to the power required to run your AC at full blast. Above that, it dominates everything else by a pretty considerable margin.

AC Compressor Power Requirements
If it takes 12kW to go 60mph, and you have a 4kW AC system that you're now running, your load is now about 16kW. Taking a 96kWh pack for convenience, you can run 8 hours at 60mph = 480mi without the AC, or 6 hours at 60mph = 360mi with the AC, which is a 25% range hit.

If you're going at 80mph instead, yes, it'll be closer to 10-15%.

Lorenfb

1. Rolling resistance and drag losses are basically equal at 35 - 45 mph.
So rolling resistance is a factor, not to be "forgotten".

2. Another assumption; i.e. an EV AC system consumes 4kW.

gnochi

1. I re-ran the calcs, and that's fair, at lower speeds and even towards highway speeds, it is in fact significant. It's also closer to (1% vehicle weight*speed) than (3% vehicle weight*speed) if you're on anything resembling a halfway decent road. This does adjust values somewhat but the general idea of air resistance massively dominating holds.

The 12kW at 60mph is for several different sedans I've seen, tested, and worked with.

2. Every EV AC system I've seen - GM, BMW, Tesla, and now 2 EV startups I've been at - consumes about 4kW battery power for around 7kW of heat rejection at full blast. There's usually some extra capacity that's manufacturer dependent for refrigeration of the battery pack coolant loop; GM skips that and routes R-134a through the battery pack cold plate.

It sounds like you're also in the industry. It's entirely possible your numbers for, well, everything are different from the ones I've seen and played with. It sounds like that's likely the case.

Lorenfb

Using my Android app (LeafSpy - also a Tesla version) via the diagnostic port and monitoring the Leaf's
power consumption while turning the AC on/off, the power consumption for AC was about
1.2kW worst case. Given the larger interior of a Tesla, maybe the AC power consumption would
exceed about 2kW.

gnochi

HVAC power depends very much on trip length, desired temperature, and environment temperature - these systems are pretty well designed, and if the temperature delta isn't too extreme, or the interior temperature is already the desired temperature, it won't use much power at all. If you're in Flin Flon Manitoba in winter (admittedly, heater instead of AC) or Death Valley California in the middle of summer, it'll draw quite a bit more.