Why don't Tesla Powerwalls have online UPS features?

[a_scanner_brightly]

I'm evaluating home solar power setups (which I'd install myself, of course).

What I'm leaning towards is installing PV cells on my roof, batteries (maybe a Tesla Powerwall) and a grid-tie.  The local utility offers a net metering contract and will even pay wholesale rates for surplus power.  So far so good.

The PV cells would charge the batteries and the combination of the two should meet household demand.  If there's a surplus of power from the PV cells, it sends it to the utility to spin the meter backwards and maybe even make some pennies.  If there's a deficit, it'll pull power from the grid.

Nothing unusual here.  The Powerwall advertises that it does just this.

Here's where the Powerwall falls short and makes me mad: if the energy grid goes down, your household will experience a short blackout before it switches over to the batteries.  Outrageous!  Why isn't it seamless? 

I'm not asking for magic.  Indeed, in a server datacenter, an online UPS will isolate the servers from "dirty" street power: they run entirely off of the batteries and won't be affected by voltage spikes or drops. 

Practically speaking, is a few second blackout that the Tesla Powerwall gives you a big deal in the end?  No, but falling short on full isolation suggests some kind of Bullshit is afoot, because there are a lot of common parts between the two systems; surely Tesla thought about this but decided not to provide full online UPS style isolation features.

Anyone know why?  Although 99% of people won't care, it strikes me as an obvious deficiency.  It feels wrong.  It'll keep me up at night.  Is there some regulation that prohibits this?

[paddedhat]

I've been involved with wiring new data centers.  I'm of two minds here. My first thought is, you are thinking in apples and oranges here. A data center is a huge undertaking of incoming utility power, (often dual feeds from different parts of the grid, or even two grids) a massive battery bank that only needs to function for seconds, and diesel generators that take over as soon as they are running, stabilized at the correct speed and frequency. So, it essentially has little to do with your situation, and cost mega-dollars per KW to install. The other thought is that, you are correct,  a straight relationship between grid-battery-PV could be designed with zero lag as it bounces back and forth, and should be possible to be controlled digitally, with little mechanical switching or drama.

So, at that point I would think it comes down to safety. Utilities are obsessed with accidential customer backfeeds, and rightly so. There have been many lineman killed in the past when moron homeowners backfeed utility lines during a power outage. As a result,  a lineman who has been working for 20 hrs. straight, in shit conditions, and is trying to get a patch of overhead lines back together, when he makes a fatal error. He looks left, and sees where the primary wires are severed by a tree. He looks right, and sees the same thing. He gets up in the bucket to repair a section that is obviously dead, and is electrocuted as he grabs the lines. What he didn't know is that some fool has a $99 harbor freight generator backfeeding that section of "dead" line,  by illegally wiring his genny into his dryer outlet, without hitting his own service disconnect.

This leads me to my best guess on the issue. I would assume that utilities will not allow equipment that does not fully de-energize equipment on the utility side before transferring to battery back-up. So essentially a safety oriented design involving "if then" logic. Incoming utility power to the house is lost. If there is a mechanical disconnection from the grid, by opening a switch and confirming such, then the battery can tie to the customer side of the system and begin operating. This standard would prevent a situation where a system is drawing the batteries down to feed the home, while unknowingly still connected to the grid, and backfeeding the closest transformer, creating thousands of volts of energy on supposedly dead lines.

All of this BTW, is just my best guess. I'm sure there are a few members her, more knowledgeable on the topic that might have more to add.

[sokoloff]

Power efficiency under normal operations is the reason.

Truly online UPS (with one exception) have low efficiency, wasting a lot of power in the conversion processes. There is a (relatively) new design for online UPS that one vendor calls "Delta Conversion" which eliminates some of the losses, but at low load is still only 90% efficient.

Standby and line interactive UPS have higher efficiency scores when the utility feed is up (the normal condition), but still can carry the load fairly uninterrupted for most electronic devices. These are not considered "online" from a power protection industry perspective.

Online UPS have over 10% losses in normal operation (they can get more efficient as the power demand goes up, but at low load, the efficiency is low). Most consumers who are shopping for PV systems with Powerwall backup are probably not looking to increase their electricity consumption 15% or so to provide whole-house online power protection. Point of use standby UPS, laptops, etc. can carry those loads that need uninterrupted power.

For datacenters, I care about power efficiency to some extent, but I carry about uninterrupted, dual-feed power more.

[paddedhat]

so, you're saying that, absent efficiency issues, it's possible to integrate an off the shelf, rationally priced UPS that would incorporate a Powerwall, or full load capable battery storage, into a grid tie system, and meet utility requirements? As I wrote my response, I was vaguely remembering this being an issue. A quick search after your post turned up several references to utilities banning the practice of using a grid tie system as a stand alone "island" PV/battery power supply during an outage, much less seamlessly doing it with a UPS. They claimed it was due to safety issues. That said, apparently the most aggressive actor in this argument is FPL. Florida's incredibly corrupt utility that has effectively made FL. a total backwater in the solar game, a place they should be leading the nation.

[sokoloff]

I don't think "rationally priced" is going to be part of the specs.

Most people are going to be OK with a low standby loss, momentary loss of power system. That's what Powerwall delivers, so anything that comes along to disrupt that is going to have an even harder time without the first mover, Elon Musk hype machine, Tesla tie-in, etc.

The incorporation of rationally priced, OTS UPS that I contemplate above is, "Do everything the Powerwall and grid-tie solar system does today. Then buy a consumer or SMB point-of-use UPS and put it under your desk for loads that can't tolerate 10-30s of power drop." That is going to be cost effective as it's mass-market. (Most of these are standby, not online, in design, so are efficient in the normal case.)

In terms of local generation/storage, we have datacenters in multiple different jurisdictions, each with their own requirements on permitting, etc, but we haven't run into anyplace that flat out told us, "You can't have local power generation because of islanding concerns." Of course, we are also not in the business of trying to backfeed locally generated power from our diesel generators; we're just looking to provide 5 nines of electric power feed to our racks, chillers, etc.

[paddedhat]

I assumed that given the cost, the fact that most residential customers will place little value on a huge premium for avoid a 20 second drop, and utility concerns (both real and fabricated to slow the roll of PV generation) would make the idea unworkable.

[a_scanner_brightly]

Thank you for the high-quality replies so far!

[sokoloff]

I assumed that given the cost, the fact that most residential customers will place little value on a huge premium for avoid a 20 second drop, and utility concerns (both real and fabricated to slow the roll of PV generation) would make the idea unworkable.

Totally agree. If I didn't get that across in my writing, that's on me.

[a_scanner_brightly]

@sokoloff: if you could factor out the efficiency issues, do you know if the utility's safety concerns that paddedhat speculated about would be legitimate blockers on rolling out a Powerwall-like-device with online UPS features?

I used to manage datacenter ops in the UK and I don't remember anything about isolating ourselves from the utility coming up in the design discussions either.

At this point I'm simply curious what the ratio of real vs fabricated is on the safety concerns.

[sokoloff]

Anything is possible if you have enough money to throw at the problem.

The topic of not backfeeding didn't come up, because datacenters "automatically" isolate themselves from the utility (preventing backfeeding of the utility with the [typically] diesel generator) with a automatic transfer switch (electromechanical) or a static transfer switch (semiconductor-based) in a typical install. In our typical single-feed install, we have commercial power coming in, feeding a generator transfer switch (electromechanical), feeding a double-conversion online UPS, feeding the PDUs. At low loads, we can lose 20% of power to power conditioning equipment losses.

The PV case with battery backup is an interesting one. The user wants the PV to backfeed the grid generally, wants to be able to take power from the grid generally, and in the special case of the grid power being unavailable or unstable, the utility wants to ensure no backfeeding in that circumstance and the user wants to be able to draw down the batteries. The Powerwall skips the last feature only.

Based on DC costs, adding a transfer switch to an install is likely a ~$50/Ampere cost and the semiconductor version a high-single digit multiple of that. For a home with 200A service, that's a $10K or $75K figure. Obviously, there are some economies of scale likely and it might be possible to get those figures down with volume (rather than being a one-off, "How much ya got?!" piece of critical datacenter infrastructure), but this is not consumer-priced equipment.

The "don't backfeed the grid" that paddedhat raises is a legitimate concern IMO and utilities are right to demand that backfeed capable systems cannot backfeed the grid when it's down (or out of spec). I have two UPS at home, but only because I used to run IT Ops and felt like playing around with a home lab. My end-user equipment nowadays is all laptops and I'm willing to take the every year or so 15 second outage in New England on most of my house. The batteries I have aren't going to carry my load for an hour anyway. They carry the racks for about 45 minutes for the light rack and 25-30 minutes for the heavier rack. Outages up here are either <1 minute or multi-hour; seem to be very few 5 minute outages.

The bottom line: I don't think consumers care enough about power availability to pop an extra $5K out for a system that does everything the competitor does, plus can carry the load without interruption in the event of a grid failure. Consumers who would buy a Powerwall today and worry (or even think) about the power blip problem have another option which is a $200 UPS for a few critical loads and 9V batteries in their alarm clocks to prevent having to go reset all their clocks.

[paddedhat]

"critical load" management is the key. In a residential setting, most of the loads (fridge, water heater, lighting, etc....) are not impacted in the least if you drop out for twenty seconds.  As Sokoloff notes, the small remaining critical loads can easily be addressed with one, or more, low cost stand alone UPS supplies. Therefore, My guess is that, unless it's a very low cost option in the future, and of zero concern to utilities (both of which are unlikely) it's probably not going to happen. Remember, in the states at least, utilities absolutely despise the entire idea of user owned power generation, in any form. They spend tens of millions, or more, every year trying to roll it back as much as possible, and delay any potential expansion. If Tesla, or anybody, marketed an uninterruptable  storage source on a grid-tie, the power co. lobby would immediately use it as a weapon to argue that it endangered their workers in an outage. It may be BS, but a good story and a fat campaign contribution are all it takes to fabricate laws that suit your needs, in most state legislatures at this point.