Using Data to buy a Generator


Has anyone used the information gained from Sense to figure out what size generator to buy? Prior to my Sense, I had no idea how much wattage that devices would use. The generator company wattage calculators seem pretty off. For example, when my hot water heater runs, it’s about 260 watts. But on the generator site, it said I’d need about 4500 watts for that. My home uses about 16 kWh per day, so by that math, I shouldn’t need anything more than a 1,000 watt generator. Am I approaching this correctly?


Do a search on “Generator” on the community forum. Several threads in different places on this topic, but no strong determination on whether one needs to size the generator for the absolute worst case power spikes or for the 1/2 second root mean square (RMS) power consumption that you see in the power meter. Large motors have huge inrush current and power spikes when they start up, before the windings energize, although some have run-start or compensation capacitors that reduce the spike. There’s also the question of whether you want to size for whole house needs, or just critical circuits. Typically critical circuits are motor-heavy - refrigerators, furnaces, pumps.


For the most part, home standby generators need to be sized based on the peak load they’ll see, not the average load over the span of hours or days. First, if you’re installing a whole-house automatic standby generator with a transfer switch that connects the entire house to the generator, then the NEC mandates that the generator be sized to handle the entire load that can be present. This often results in a generator that’s oversized for typical use, which is a problem not only for the installed cost of the system but for fuel use during an outage. The electrician pulling the permit to install the system will run the load calcs in this case.

If you have a manual transfer switch, or a separate panel containing generator-only loads, then you have more flexibility. Are you looking to back up the whole house or just a subset of your loads?

BTW, the discrepancy between your water heater usage and the 4500W guideline is likely due to gas-fired vs. electric tank water heaters.

1 Like

I’m presuming your hot water tank is not electric? That’s where the 4500w quote for it comes from - that would be about right for an electric tank.

So, yes, as was mentioned, you do need to keep in mind inrush current. Things like a refrigerator may only draw a few hundred watts when running, but upwards of 800-1000w when starting. The same goes for other inductive loads in your house like the HVAC blower motor, etc.

That said, yes, I will agree that more often then not generator companies will drastically over quote you on the size of generator they suggest you need, versus what you can actually get away with. I ran a good portion of our house through several outages with a 3000W generator, and for shorter outages I have a 2500w inverter that I hookup to one of our EV’s that gets us through without ever having to bother with a generator at all.

Now, with 2500-3000W…are you going to be running heavy loads like your stove/oven, al electric hot water heater, central air conditioning? Nope, not even close. But I can run our furnace (heat in the winter is way more important than AC in the summer), our refrigerators, lights, and even the TV and internet/networking hardware. The rest, IMHO, are “frills” that we can do without until the grid electricity comes back.

1 Like

Thanks! I appreciate this feedback. I just looked at my Sense data to see the “startup spike” and I do see it for my refrigerator, but don’t see it for my gas-powered water heater.


Presuming your gas water heater is a high efficiency one with a power vent, that load you’re seeing would be just the small electric ventilator motor coming on before the burner ignites. That’s easily going to be operated by even a small generator.


Yes, it’s a super high efficiency one, rated at about 98%. All the service techs just shake their head at it. I just bought the house and the previous owner installed it, likely as a selling point. It worked. :slight_smile:

1 Like

One more comment: You are likely seeing the tip of the iceberg WRT inrush spikes with Sense. If you were to poke into the motor starts with a finer resolution measurement, the Fridge spike would start higher and the hot water heater motor would likely show a spike, but the RMS power calculation over 1/2 second averages things out over that time period.

1 Like

Ahh, great point. I’ll go back to my firehose and see what I can get that way. I was getting a ton of data by putting a proxy between my browser and the web site, and capturing all the json data in the web sockets. I’ll check that out too. Thank you!

1 Like

Yes there is certainly going to be an inrush spike even with a small motor like that. However, we are still talking pretty small wattages in the grand scheme of things as those vent motors are pretty small.

Also keep in mind that with something like a hot water tank where you have a significant amount of hot water stored at the time the electricity might go out, simply turning the tank off for the duration that you were running on a generator is another option. Load shedding can make even the electrically hungriest house run much leaner, if you see what I’m getting at.

1 Like

In the US, the National Electrical Code sets the requirements for generators. The average load does not matter. What matters is the maximum load. To run a whole house from a generator, on a single transfer switch, the generator and wiring must be sized for the maximum load of the whole house.

The maximum load can be calculated by adding up the amps of all the circuit breakers, in the worst case. If loads are not “concurrent” (e.g., a winter heater and a summer air conditioner) you may include only the larger of the two. Some specific loads actually use less current than their circuit breakers. There are other rules, and it can get complicated…

NEC section 220.87 allows an alternate method of determining the load. That method is to measure how much power (amps) are actually used, every 15 minutes for at least a month. The highest measured value, gives the minimum size of the generator. I am pretty sure Sense has that data. It is used to display the “meter” page on the website. If you go to the “meter” page, and hover your mouse over the peak time (spike), it will show the load at that time.

There are two problems. NEC wants the measurements “per phase”. Most houses have two phases. The Sense “meter” display, only gives the sum, not the individual phases. The second problem is, I don’t see any way to download that data. To get a building permit, you have to submit a printout or csv file of all of the measurements – every 15 minutes for the entire month. (More often is ok. Less often is not.) A printout of the graph image, might be acceptable to some permitting authorities, but the actual NEC text wants the individual measured numbers.

So, I suspect a software feature to “download” from the “meter” page, could get the required data… But it does not seem to be there, quite yet.


Very helpful. One thing that’s not clear from your description, is whether the 15min sample is an instantaneous measurement, or what period it should be RMSed over. The 15 min sampling indicates to me that spikes are likely ignored since the probability of landing on a half second (as reported by Sense Power Meter) is very small.

1 Like

Is the cost difference between models worth all of this effort to work out exact details?

Assuming that you can’t(legally) do the work yourself, is the cost difference between generator models significant compared to the electrical, mechanical and permit costs? I prefer to do things right once, even if I have to pay a little more or wait a little longer.

Surely you don’t want to get a model that just fits, only to have issues with a new appliance in the future.
I would think that a good ballpark figure with a reasonable safety factor should be enough for most people.


Or you could just do like a lot of people…but a portable generator that’s of adequate size to power the essentials, and when needed…just run an extension cord through the door, plug in a power bar, and power the essentials that way.

Don’t get me wrong, if money is no object…the hardwired automatic standby generator with an automatic switch is most certainly awesome and all. But we’re talking thousands (or tens of thousands) of dollars for that sort of setup. A 3500W generator and some extension cords on the other hand, can be had for a few hundred bucks. :wink:

For me it would come down to how often you’re power is out. If you live in a rural area where a mouse farting in the woods is enough to knock your power our for hours or days, a high end standby generator makes a lot of sense. If you’re in a suburban area where outages are very uncommon (albeit inconvenient) a cheaper standalone generator where you can just manually plug the essentials in for the few hours the power is out may make more financial sense.

1 Like

I got that from an article,
A google search for “NEC 220.87” finds lots of related material.

In that article, it says the measurements must be , “based on the maximum demand (the measure of average power demand over a 15-minute period)”. I suspect that means, the measurement should be RMS averaged over a period, not instantaneous, over a period of not more than 15 minutes. (If the periods are shorter than 15-minutes, but the measurement is still RMS-averaged and continuous, the maximum might be higher but not lower.)

Clearly, they intend to ignore the few-second starting peaks for small motors. Those short-duration loads give higher instantaneous readings, but won’t heat the wires to failure. (That’s why they use “slow-blow” fuses and breakers for motor circuits.) It makes sense, because motors are designed to accept a voltage-drop during starting, and new motors have “slow-start” varistor circuits, so a short-term voltage drop really doesn’t matter.

Sense can ask their engineers to read the NEC requirements, and then do something appropriate. There is (always) a lively debate among electrical engineers, inspectors, and contractors, about the proper interpretation of the code. In most cases, reasonable interpretations are accepted…


[quote=“oshawapilot, post:14, topic:5079”]
Don’t get me wrong, if money is no object…the hardwi red automatic standby generator with an automatic switch is most certainly awesome and all. But we’re talking thousands (or tens of thousands) of dollars for that sort of setup. A 3500W generator and some extension cords on the other hand, can be had for a few hundred bucks. :wink:
If money is no object do four Tesla powerwalls for a 20kw Backup source and they would be silent :slight_smile:


This is very useful. As requested on other threads, if sense split the data per phase and allowed downloading finer resolution than 1 hr (15 min would be great, 5 min even better), then Sense data could be used to size generators. Which would be a selling point, since the difference in cost between generator models will be likely much larger than the cost of Sense.

1 Like