Considering getting a Powerwall

[arg]

Is this typically the case? What if charge battery overnight such that it uses North Sea Wind that would otherwise be curtailed. i.e. not displacing someone else to Gas (I presume?)

I can't find a good source of curtailment data, but if there's not much happening now, the growth of new wind means there will be very soon. Last night for example at 00:30-01:30 there was only about 2.8GW of gas (and 10-11GW wind; indeed as the demand went down it looks like the wind was following, though that could be coincidence).

However, there are problems - the E7 fixed time period is far too crude for effective use of the storage, but dynamic tariffs (like Octopus Agile) suffer from the hidden subsidy inherent in the standard tariffs (inc. E7). People with solar - the most likely customers for this sort of stuff - will see a disadvantage going to Agile tariffs, because they will have their major demand for grid-sourced energy during winter, which is the most expensive time for the grid to provide it. So when buying on a traditional tariff that quotes a fixed price and so averages the unit cost over long periods, these solar customers are paying less than their fair share: everyone is paying the same price per unit, but the solar customers are using more of the units that were expensive to generate, compared to the average customer. If you then create a dynamic tariff which, for the average customer, gives the same price (and so the same profit to the supplier) as the traditional tariff, that is immediately going to be more expensive for the solar customer.

 

[MrBadger]

indeed as the demand went down it looks like the wind was following, though that could be coincidence).

I think that was mostly still controlled by ccgt, although not exclusively, ie hydro.

I was looking at this last night in response to this question and wanted to put to clear graphs together, but I couldn't do clear so left it till later when I could be sure I was looking at like for like time periods. I was using the raw Elexon data from bmreports

However, what it did show, up to around 21:00, was that if you looked at the individual sources of energy, many were quite flat. However ccgt ( combined cycle gas turbine) was not and visually there appeared to be a fairly good correlation to demand. I say visually, at I wanted to put a more numerical set of data together before replying.

However, very briefly, to prevent this question going stale, I'll omit the numerical data and keep with the visual hence this post.

Electricity Data Summary | BMRS is a good one, system demand (left) and generation by fuel type (right).

This is this mornings data. 21:00 last night, whilst different, and huge quantity of wind, the outcome remained the same.
Visually the evidence is as follows. Note x axis scale differs between graphs, so demand is more compressed in x axis than generation - part of the reason why I was holding off a quick reply.

Removing the minor 'flat' fuel types and less flat interconnects (they make little difference to the shape of the graph and removal adds clarity). Removing ccgt, the typical marginal source, the first pair of graphs show a largely flat total curve, certainly not following the system demand.

However, adding back in ccgt, you see that the generation curve now tracks the system demand quite closely. A first glance conclusion is that ccgt remains the marginal source as it is the source that has significant control over the generation matching demand.

So based on this, even when wind is major proportion of generation, it is ccgt that is the primary energy source that is used to balance generation with demand, ie when someone flicks that switch. But other sources were also at work (they are better for fine tuning but not the primary method) like hydro (typically pumped storage), and some of the interconnects (which we have no control over their source of energy) - see below.



What is good though, is that average generation this weekend was significantly biased to renewable. But these events are currently, not a regular occurance. So the round trip losses are coming from more fossil fuelled sources.

As for solar time shifting, if this is the primary requirement for battery, the a Powerwall v2 being AC coupled is not the best choice. A DC coupled storage system is a better choice as solar goes direct into the battery with no AC/DC/AC losses, so that solar stored energy is as efficient as had it been used direct from solar conversion wise - both would typically use the same DC/AC inverter so the only losses incurred are in storage itself. Ironically, Powerwall v1 worked this way as it was a DC coupled system, but in a FiT based system, it had the disadvantage a very blurred line between battery storage and solar generation and the total generation meter - the meter measuring what you got paid on, which would have worked against FiT payments. So anyone wanting a battery storage system primarily for time shifting their solar, should also be considering a DC coupled system, ie not a Powerwall v2.

 

[arg]

However, adding back in ccgt, you see that the generation curve now tracks the system demand quite closely. A first glance conclusion is that ccgt remains the marginal source as it is the source that has significant control over the generation matching demand.

So based on this, even when wind is major proportion of generation, it is ccgt that is the primary energy source that is used to balance generation with demand, ie when someone flicks that switch. But other sources were also at work (they are better for fine tuning but not the primary method) like hydro (typically pumped storage), and some of the interconnects (which we have no control over their source of energy) - see below.

I certainly agree that (currently) CCGT is the major medium-timescale demand-response mechanism 24x7. However, it's not as simple as that, due to forecasting of demand. The CCGT have to be spinning in order to be available to throttle up and down to fill in - so once the forecasted demand is in, some proportion is going to be supplied from CCGT in order to provide that spinning reserve, even if that means deciding in advance to curtail wind generation.

If you then take a domestic battery and spontaneously charge/discharge it in a random fashion, it's going to be CCGT that's throttled back when you are discharging and throttled up when you are charging. However, if you use it in a predictable fashion then you become part of the forecast and potentially allow more wind to be allowed to run overnight if it was otherwise going to be curtailed.

So a possible situation consistent with Saturday night's data is that there was plenty of wind to handle the total demand (after subtracting nuclear and other fixed stuff), but 2.5GW of CCGT had to be run as the minimum needed to provide spinning reserve, with some wind curtailed to make space for it. I suspect that the true situation wasn't quite as extreme as that, but it's getting close and in the next year or two is likely to be occurring regularly.

Another aspect is CCGT efficiency. Although offer both high efficiency and rapidly controllable output, they don't do both at once: efficiency is maximum when running at continuous optimum output - even more so if the range of outputs is such that you are shutting down a complete generator overnight and having to warm it up again in the morning. Hence there's an efficiency gain from drawing CCGT-generated power at night to charge your battery and cancel out other CCGT-generated power in the daytime - probably not enough to cover the round-trip losses of the battery.

There also seems some dispute over Powerwall round-trip efficiency. Tesla's own datasheet claims 90% "at beginning of life". For the industrial-size PowerPack they claim 89% at 4-hour rate, or 88% at 2-hour rate (stated to include losses for cooling etc.). Various websites reviewing Powerwall2 claim 89%, ">90%", 90%, 92%. I'm not sure if this means Tesla have quoted different numbers at different times, or if the journalists are just making them up!

 

[WannabeOwner]

But these events are currently, not a regular occurance

Thanks. Yes, I definitely want "overall" rather than "I cherry-picked a howling gale".

Chucking into the mix (although expect its in your reckoning, and @arg alludes to it too) it would be nice if my battery charged at night used renewable (without displacing someone else to Gas) AND the benefit that I will not be using those same kWh at peak the following evening, thereby definitely reducing demand-response top-up Gas

Maybe I am helping during Peak more than I am achieving non-displacement to Gas during over-night charge? But perhaps both are going to be the same Gas turbine, in which case that would be disappointingly "broad as it is long" (and then there are the battery losses ...)

if you use it in a predictable fashion then you become part of the forecast

That's something to look forward to.

 

[kuruma]

Hi @kuruma is it possible to charge a completely empty PW to 100% on the cheap night time Octopus tariff and then basically run your house during the day

Yes it is. All our car and PW charging is complete by 4am unless we're super low on the car battery which is rare. We run the house off the PW and hardly use peak rate anymore. With solar increasing now we're finding that even our off-peak grid use is falling.

BTW, the PW learns how much energy you typically use and also knows what the weather forecast is. As a result, it only takes from the grid what it needs, relying on solar to give you enough to meet your usual demand. We're very impressed with it.

 

[nightowl]

BTW, the PW learns how much energy you typically use and also knows what the weather forecast is. As a result, it only takes from the grid what it needs, relying on solar to give you enough to meet your usual demand. We're very impressed with it.

Ooh that's smarter than I expected it to be. I thought it would involve manually adjusting it a lot to ensure that it made the best use of solar whilst also topping it up off-peak. The use of weather forecast info seems like it could be very useful in this case. I'm in the process of trying to make my dumb storage heaters consider the next day's forecast to determine how long to run them for (rather than adjusting the store-a-nonspecific-amount-of-heat dial manually and getting it wrong.)

 

[WannabeOwner]

also knows what the weather forecast is. As a result, it only takes from the grid what it needs, relying on solar to give you enough to meet your usual demand

And (just for completeness in case anyone wasn't aware) gets set to Storm Mode OTA when there is risk, which keeps battery at 100% all that time, in case of a powercut. That worked twice for me, on successive days, during the first storm.

 

[WannabeOwner]

if you use it in a predictable fashion then you become part of the forecast

Saw this Patent from Tesla. Seems obvious to me, but I only glanced at it, but maybe something to look forward to ...

Scalable hierarchical energy distribution grid

 

[MrBadger]

Apologies, I had started writing this this morning, but decided that before posting I needed to double check some facts with a contact of mine at National Grid who works on the balancing system. As it will probably take a few weeks, and a few beers/coffees, and for me to ask questions that he can answer, I deleted most of the post. The remainder is below, but half a day out of date.

Still a decent breeze in last 24 hours, and CCGT still an almost perfect match for demand. In fact, a spookily good match.

So absolutely clear what source is the main source of marginal generation.

But... (this is where remainder of post was deleted)

I certainly agree that (currently) CCGT is the major medium-timescale demand-response mechanism 24x7. However, it's not as simple as that, due to forecasting of demand. The CCGT have to be spinning in order to be available to throttle up and down to fill in - so once the forecasted demand is in, some proportion is going to be supplied from CCGT in order to provide that spinning reserve, even if that means deciding in advance to curtail wind generation.

Totally agree on that and always have. What I need to understand from my contact is the feedback loop of how demand forecast is derived from semi random acts, such as ad hoc charging, or topping up a battery.

There also seems some dispute over Powerwall round-trip efficiency. Tesla's own datasheet claims 90% "at beginning of life". For the industrial-size PowerPack they claim 89% at 4-hour rate, or 88% at 2-hour rate (stated to include losses for cooling etc.). Various websites reviewing Powerwall2 claim 89%, ">90%", 90%, 92%. I'm not sure if this means Tesla have quoted different numbers at different times, or if the journalists are just making them up!

I've seen some of those too, but I have also seen some real world data, albeit in limited dataset and potentially not very reliable - probably correct (I was surprised and did question one of the results), but not obtained in a scientific fashion - much like many users anecdotal experiences. These users, in real UK household consumption, put their losses at around 20%. Its very easy to find a sweetspot efficiency of an inverter (usually close to max capacity) and give readings for that, but for me, real world data is key, and that will usually be well down on optimal efficiency - very similar to charging car at low amps is less efficient than at higher amperage. What would be good would be to get more data and sounds like a few Powerwall users on here that can potentially supply - basically energy stored vs energy retrieved - not sure what mechanism Powerwall has for this so may be easier said than done.

 

[WannabeOwner]

not sure what mechanism Powerwall has for this so may be easier said than done.

I haven't looked at it, as yet, so might not be useful, but I saved this link - a description of the API

vloschiavo/powerwall2

 

[im85288]

Ooh that's smarter than I expected it to be. I thought it would involve manually adjusting it a lot to ensure that it made the best use of solar whilst also topping it up off-peak. The use of weather forecast info seems like it could be very useful in this case. I'm in the process of trying to make my dumb storage heaters consider the next day's forecast to determine how long to run them for (rather than adjusting the store-a-nonspecific-amount-of-heat dial manually and getting it wrong.)

Just to be clear here as far as I understand the powerwall does not use weather forecast info (apart from storm watch) to predict how much solar it thinks will be generated on a particular day. Instead it keeps track of historical data and looks back on that when deciding what to do. For example the Monday before last was a great day for solar generation and not much usage of energy in the house, consequently the Monday just gone it used that data when deciding how much to put in the powerwall from the grid...unfortunately it was a poor solar generation day so came up a bit short. Tesla need to add a bit more control on this for advanced users and ideally start including the upcoming weather when deciding on how much to charge from the grid.

Overall though it does do a very good job, it’s just the odd occasion especially in these months where it can get mixed up slightly.

 

[arg]

I've seen some of those too, but I have also seen some real world data, albeit in limited dataset and potentially not very reliable - probably correct (I was surprised and did question one of the results), but not obtained in a scientific fashion - much like many users anecdotal experiences. These users, in real UK household consumption, put their losses at around 20%. Its very easy to find a sweetspot efficiency of an inverter (usually close to max capacity) and give readings for that, but for me, real world data is key, and that will usually be well down on optimal efficiency - very similar to charging car at low amps is less efficient than at higher amperage. What would be good would be to get more data and sounds like a few Powerwall users on here that can potentially supply - basically energy stored vs energy retrieved - not sure what mechanism Powerwall has for this so may be easier said than done.

Yes - while I'm entirely willing to believe that Tesla's efficiency figures are optimistic one way or another, it's extremely difficult to see how you would actually measure the real-life efficiency. Any kind of ad-hoc measurement is going to have far too much error (VA vs Watts for a start), and even MID-grade meters aren't that accurate over a wide range of powers, so I wouldn't trust a measurement based on taking readings off two separate meters and subtracting them. Numbers generated by the device itself aren't necessarily going to be very accurate either. Probably the best would be to get a modern meter with separate registers for import vs export and put it between the powerwall and everything else; however, I can't imagine many people have one as there's be no reason other than specifically to measure the powerwall efficiency.

The point about inverter efficiency is also an interesting one. Certainly if you just calculate an overall efficiency number it's going to be hopeless at very low powers. Typical efficiency curves fall off dramatically at the low end due to fixed costs (gate drives etc. which are constant for a given switching frequency regardless of the output power), reach a peak somewhere in the middle, then tail off gently at the high end as I²R losses dominate, and made to look more extreme by plotting the power on a log scale. So if you strip out the fixed costs the curve is much closer to flat.

So for a powerwall it's probably more meaningful to think of it as having a fixed cost of a few watts while it is operating, plus an efficiency that's reasonably constant. Obviously the powerwall is going to have some fixed cost even just sitting there with the control settings causing it to neither charge nor discharge; I have no idea how that power compares with actively discharging at a very low power level (indeed, how sensitive is the import/export detection? How much do you have to be drawing from the grid before it can be bothered to start up?).

In fact, thinking about that leads to another issue to wonder about - how does it react to power factor? Traditional export limiters (immersion heater diverters and the like) typically have a low bandwidth measurement from the import/export sense coil (not correlated with voltage) and just try to drive that current to zero, so they will not achieve zero power if the load has poor power factor. Probably that's not a big deal for most domestic setups that won't have much of a traditional inductive/capacitative load (the powerwall could in theory offer power factor correction for such loads). However, domestic loads will have a large component of consumer electronics with switchmode power supplies in them with poor power factor caused by harmonics rather than phase lead/lag. It's not clear to me what the powerwall can or should do with such loads, but they will certainly mess up the sort of measurements we are talking about here of the efficiency at low load. The couple of hundred watts background load of a typical home nowadays will be almost entirely made up of these ill-behaved devices.

 

[WannabeOwner]

or a powerwall it's probably more meaningful to think of it as having a fixed cost of a few watts while it is operating

Sorry, bit of a tangent, but my PowerWall is in the potting shed. That's at the very far end of the "external stuff". Its all modern / well insulated, but the PowerWall is providing a smidgen of background heat, such that the frost radiator has not had to come on at all. The radiator in that room, right down at the far end, has the longest pipe runs and an old fashioned low loss header thingie is involved too ... so the moment the thermostat calls for heat there is a loss that makes driving off with a stone cold battery in mid winter look trivial! So perhaps a little something for the "credit" accounting column.

I'll have waste heat in Summer of course ... but excess PV at that time, so not sure I will care (potting shed doesn't overheat in Summer, tiny window etc. so if the few continuous watts of heat starts to matter I'll open the window).

 

[nightowl]

Now that we're starting to see a fair bit more sun, I've had the opportunity to try out the Eco+ mode on the zappi. Standards apparently state that the charge must occur at a minimum rate of 1.4kW. I'm generally finding that I do need to allow it to import 25-30% from the grid to cover the brief dips when clouds go past etc. but it does still stop charging occasionally due to longer periods of cloud cover, meaning I have to wake the car up when I see it is sunny enough to continue charging. Not that I need to charge much anyway at the moment as I'm only driving to the supermarket once a week, but still I wanted to see how effective it was.

One possibility I suppose would be to charge the car up overnight and just use the solar to feed a PW once my immersion heater has taken its share (usually around 8kWh but I am looking to replace that with something a bit more efficient as the insulation on the tank is quite poor.) I'm generating around 15kWh a day at the moment, I expect even more will be generated in summer. Obviously there's a 10% loss when drawing on this power though.

I also read that apparently there might be an increased chance of power cuts whilst the coronavirus situation is still ongoing (presumably due to less staff available to support the energy infrastructure?) This, combined with the previously mentioned ability for the PW to respond to upcoming storm alerts to charge itself in anticipation of outages, seems like it could be useful to have (particularly as I am working from home at present.)

The company that installed my zappi can install PW but they are currently not doing any installs. I see that reservations can be placed via the Tesla website, but who would contact me? Do they just pass my info on to an approved installer? I assume I would probably end up having to wait until the lockdown ends (thus making my point about power cuts a bit more redundant but hey... for storms it's probably useful anyway.) What sort of timescale would I be looking at if I ordered one?

 

[ACarneiro]

I’d be astounded if things got so bad as to power cuts becoming “a thing”.

 

[gangzoom]

We had our PW installed Nov last year. Looking back our electricity usage back than was 2689 kWh day rates, 11354 kWh E7 rates since some time in 2018.

So the split for E7 normal was 25%, however this did include washing machine coming on at 5am witth all the noise associated.

Our last meter reading on the 22nd March showed, 2791 kWh day rate, 14327 kWh E7. So since the PW install we have used 102kWh day rate, and 2973 kWh E7 rate, so the split is now 97% E7 rate!!!

We also have a 4KW solar PV system that is split with half the panels SW facing half SE. Up until last few weeks we had little sun to speak off, so the PW software has been set to charge off E7 - hance the massive split. But since the sun had started to appear I've been able to charge the PW off solar more and more. This was Monday this week, virtually empty PW, charge to near 90% by sunset.

This now means I can also charge the car off the PW at night, the fact Tesla gives you so much control over charge rate and battery charge target makes doing this very easy. Am currently doing 12 miles a day commuting to work as am a key worker, I simply set the charge rate 4KW at with target SOC at 50% with charging schedule for 10pm, and that seems to empty the PW and charge the car exactly as what I want.

My current weekly electricity usage as an result is falling like mad, even with car charging. This was a couple weeks ago, and current week!! Sub 20kWh per week use at E7 rates is £1.60 a week in electricity costs!!

This is not a cheap bit of kit, but boy is the tech amazing .

 

[Martinb999]

I had a powerwall 2 and gateway installed last week and have a 3.5Kwh solar array (installed 5 years ago). I live in the South of Kent and we have a lot of sun (normally) and my rebate from the FIT has never been below £150 per quarter (max was £2,314 per quarter).

I've got the thing in self powered mode and I am already seeing my grid useage fall dramatically. We have Economy 7 and I'll look at that again in three months once we have the first electricity bill come in. Our smart meter is from the first round of smart meters so isn't compatible with Octopus so we would have to wait for a new meter anyway.

We have a model 3P and a model X, both of which I hope to monitor their charging from the powerwall/grid over the next three months.

Let's see how we get on............................

 

[MrBadger]

and have a 3.5Kwh solar array (installed 5 years ago). I live in the South of Kent and we have a lot of sun (normally) and my rebate from the FIT has never been below £150 per quarter (max was £2,314 per quarter).

Those numbers don't stack against my experiences. We had 4kW PV installed over 6 years ago and our annual rebate comes in just over £600/year. I've not kept up fully with FiT prices, but have in my head that FiT generation and deemed export we are just over 17p per kWh so probably more than your kWh tarrif with us being older - we missed the higher 40p+ tarrifs.

Whilst we are not optimal generation wise (winter and evening shade), I cannot see any way that a lesser array size on a probable lesser tarrif can possibly bring in the sort of figures (£2314/qtr) that you are saying that you get. Our FiT quarterly payments range between somewhere around £30 and £280, very much season dependent. So even if we had twice the efficiency (which would be impossible), your numbers are way adrift of my actuals and other PV owners that publish their generation figures.

 

[gangzoom]

Those numbers don't stack against my experiences. We had 4kW PV installed over 6 years ago and our annual rebate comes in just over £600/year.

Don't worry you are not missing out, our FIT payments are similar to yours. Those that had the system installed earlier are the ones really racking in the income!

The good thing about FIT is the 50% assumed export which we still get paid for, even though our export figure is now pretty much zero%.

Oh another sunny day means on track for another 1kWh grid usage day post car charging.

 

[AndyS3]

If you watch YouTube 'fully charged'

He had a powerwall installed. Tidy installation. Two things put me off,
first when there is a power outage, the switch over to the power wall is not seamless, upto several seconds downtime. This on/off can damage house electronic equipment.

secondly. You cant power the heating from the powerwall it limited to a several kW demand