Five Things to Know About the Sense Power Meter

The Sense Power Meter offers a powerful real-time measurement of your home’s total usage. But the Power Meter is far more sophisticated than most people recognize. Here are a few details:

  1. The main Power Meter (as differentiated from the Device-level Power Meters) is the most accurate view on what’s going on in your home. Configured properly, it offers a half second resolution real-time view of your usage based directly on Sense Monitor measurements. Many of Sense’s other views are built on the incoming data that flows to the Power Meter, but virtually all other views are dependent on additional factors and calculations making them thirdhand views. The only other nearly direct measurements are the Voltage, Power and Frequency in the Signals view, under Settings and similar data in the Power Quality view within the Sense Labs.

  2. The Power Meter shows the results of number of calculations, not direct measurements - Even though it seems like a direct measurement, the Power Meter usage numbers are several steps removed from the actual voltage and current measurements sampled a million times a second by the Sense Monitor.

  • Sense produces an RMS (root mean square) result based on AC (alternating current power). Your home voltage and current alternate at 60 cycles per second (Hz), so even if the voltage and current stayed the “same”, Sense can’t just sample the raw voltage and current to produce a power number. See the diagram below to see what voltage, current and power look like over 1/60th of a second. Sense has to assess many voltage and current samples over a number of 60Hz cycles, probably 30 (= 1/2 second), to perform an RMS calculation using all the data points to produce an RMS power value for that 1/2 second. The RMS calculation produces a value that is the same as if the power was constant for the whole 1/2 second. In the same vein, your house voltage is really 120V RMS. The waveform actually peaks around +/- 170V. But Sense hides all these calculation so we can work with a single number. So people who look to Sense to determine “peak power usage” will never see the true peak usage - instead, they will see essentially the peak average power over 1/2 second.

fig-2_linear-resistive-loads

  • There are multiple components to AC power - Real (or Working) Power, Reactive Power and Apparent Power - Sense outputs Real Power numbers, just like your electric meter and residential utility customers are charged based on Real Power. But if you look at the inner workings of AC power, especially when a large motor is running, the current and the voltage waves can be driven out of phase from one another. That causes less Real Power to be delivered for the same AC current and voltage (the green curve vs. the red dotted curve below), because part of power delivered is diverted into Reactive Power that doesn’t do any useful work in your house. Generally, the motors in our homes are small enough relative to our total usage, that residential usage doesn’t wreak havoc on the grid, but industrial users have to be very aware of how much their loads change the phase relationship, often referred to as the Power Factor. They will often pay increasing amounts for Power Factors that move too far away from 1.0 (current and voltage usage perfectly in phase).

fig-3_inductive-loads

  • Sense’s Total Power combines consumption for both phases/legs in your home - Most homes in the US have 120V split-phase power. That means that there are two different supplies, or so-called legs, to your electric meter and breaker panel with the AC voltage 180 degrees out of phase with the other (below). That’s why Sense has 2 CTs (sensors) to monitor home usage current - each measures the current for one of the two legs. So the Power Meter sums the usage on each leg.

Split Phase

  • Sense Solar sometimes adds another calculation to the Power meter - If your solar CTs are on a load-side feed-in, your Total Power is calculated by adding Net Usage for the mains CTs and Solar Production. This addition helps explain why Sense Total usage can sometimes partially mirror solar production if there are problems (open CTs, calibration issues, configuration issues, all typically handled by Sense support)
  1. The only places in the Sense UI where you’ll see the split-phase RMS voltage and power measurements for each individual leg, are in the Signals section and in the Power Quality view.

BTW - Sense has written a bit more about the Split-Phase Power in this article.

  1. The Main Power Meter in the smartphone/tablet app gives hints about what Sense is actually “seeing” - The Power Meter attaches tags to significant transitions that Sense “sees”. Here are a set of waveforms from when I was descaling our Coffee Maker - you can see a bunch of the tags that quantify the size and direction of the power transition - positive = On and negative = Off.

And if you zoom in you can see even more !

You can see from the Coffee Maker Device Timeline that Sense “saw” and detected all those on/off cycles.

  1. Devices have their own Power Meters - but accuracy can vary. Device Power Meters for detected devices show Sense’s “estimate” or prediction of the usage pattern, but that is mostly based on the size of the on and off transitions, not any complex usage waveform the device might present, and totally misses any Always On component. Below is the Device Power Meter for my Coffee Maker for that same series above, so it is a Sense AI detected device. What’s interesting is that my normal coffee making pattern is one 1200W large spike followed by a bunch of 850W spikes. Even though Sense is seeing a bunch of 1200W spikes in the tags from the very different descaling cycle, the Device Power Meter gets populated based on what Sense “already knows” about the Coffee Maker’s pattern.

For comparison, here’s a view of a Device Power Meter from a smartplug that powers my laptop, monitor and external hard drive. It does include Always On power and isn’t reliant of on an off transitions.

I’m betting you didn’t know all that complexity was hidden beneath the simple Power Meter.

4 Likes

Kevin, your posts are always helpful. I especially appreciate your explanation of how alternating current relates to power, on which I was a little fuzzy.

While one million times per second sounds like exaggeration, it is not. Here is why, as explained by the manufacturer:

2 Likes

Thanks @jefflayman,
That blog is a great explanation of why the Sense Monitor has to collect current and voltage data at millions of samples per second. There are a number of chips on the market that can sample and provide real-time power data - those chips go into power meters and some low cost ones go into smart plugs, but those chips don’t:

  • Sample anywhere near the same speed as Sense. I think the ones I have looked at sample from between 1000 to 16,000 times per second.
  • Do the complex analysis and AI done inside the Sense monitor to distill all the information about every “significant” power transition to a small set of distinguishable features of each transition.

If the Sense Monitor didn’t perform its magic and just relayed the data back to the Sense mothership for processing, the data would be entirely unmanageable. By my calculation, the 4M samples / second would turn into nearly 700GB uploaded per day, instead of the 300MB max that Sense quotes in their spec sheet. Data usage for my Main Sense backs up that 300MB top end.

This data challenge also explains why utility meter and utility companies are so interested in Sense. Utilities definitely want a much better picture of what is happening on the customer load side of the grid, both for their customers’ benefit and for their management of the grid.

2 Likes

While we are on the subject of how many times per second the Sense meter takes measurements, I ran across this. In an interview named Ask the Expert, Sense CEO Mike Phillips says on page 2 (emphasis added):

A new generation of AMI meters can sample data at high-resolution tens of thousands of times per second. With this high-resolution data, machine-learning-based software is able to detect how energy is being used at the device level and can see detailed operation of the grid from the edge — including detecting faults and failures.

At even the lowest-resolution deployments, Sense software processes 50 million times more data than 15-minute interval AMI data.

This makes me wonder if Sense is ever deployed with variable resolution. Another blog from Sense, Unlock new value cases with high-resolution next-gen smart meters, says “15KHz is the baseline for high resolution data with the greatest benefits at 1MHz,” and includes this figure:

1 Like

Very interesting excerpts @jefflayman.

If I do the math, the lowest resolution deployments that Mike is talking about would be around 100k samples / second - ksps), assuming the 1 sample per 15 minutes base (though I know that my meter actually produces a sample every 4-5 seconds). So maybe Sense is offering some kind of lower/slower product tier for the meter guys. That jibes with the region and capabilities just to the left of “Behind the Meter”. Of course the current Sense product we’re using is the “Behind the Meter” product.

2 Likes

One more interesting article from a while back on Sense metering and the whole history of energy usage disaggregation

3 thing that jumped out at me:

  • Sense launched a 500-household pilot project with Wisconsin utility Alliant Energy and consultancy Cadmus to further explore the value of real-time energy data.

  • In December (2021), Sense launched an open-source software development project aimed at creating a common way for multiple devices to share data on how much energy they’re using from moment to moment. That could allow them to act in concert to keep household grid demands below levels that would trigger the need for utility upgrades, as indicated by this graph of data from 3,500 Sense-equipped homes.

  • Sense recently hired a sales executive from Bidgely. Bidgely was trying to do usage identification using only data from first generation smart meters. Bidgely, a Silicon Valley startup with disaggregation tech that analyzes less frequent data pulled from utility smart meters, has also focused on utilities as customers, with nearly 40 utilities and energy retailers using its platform.

2 Likes