The IoT: Self-Powered Solutions for a Sustainable Future

In this video, Everactive’s  co-founder and Chief Technology Officer Dr. David Wentzloff uses Earth Day to discuss the current impact that the Internet of Things (IoT) has had on the environment. Wasted energy, greenhouse gas emissions, and the battery problem have damaged the environment. Everactive’s self-powered solutions show the an alternative path to a sustainable future.


Brian: Here at Everactive, it’s my pleasure today to introduce uh the one of the co-founders and current co-CTOs of Everactive, DAVID WENTZLOFF. Dr. David Wentzloff  is also a professor of electrical engineering and computer science at the University of Michigan. So Dave, take it away for our special Earth Day webinar.

DAVID: Alright, thanks Brian. And let me start off by wishing everyone a happy Earth Day. Thank you for participating in this webinar webinar today. So today we’re going to be talking about self-powered sensors for a sustainable future. uh and here’s an agenda of an overview of my talk. So we’re going to start by talking about the environmental impact of the iot, then we’ll transition into why batteries are standing in the way of realizing a pervasive vision of the iot, and then we’ll introduce some self-powered solutions to that battery problem.

Annual US Energy Use and Emissions

So I wanted to start with a couple stats on the um annual US energy use and emissions shown here on this slide. and so on the left uh we’re showing total uh u.s energy consumption by sector, and on the right um u.s um co2 emissions or greenhouse gas emissions. These numbers are from 2018 uh because that’s the most recent data available from the the doe and the epa. and just starting with energy consumption in the u.s, the total energy consumed in 2018 is a 101 quadrillion btus or british thermal units. and um i just wanted to maybe put that number in perspective for a second because a quadrillion kind of sounds like maybe a made up number, um so first off what is a quadrillion? a quadrillion is a thousand trillion btus, or you can think of that also as a million billion btus. and just to put that number in perspective uh one 20 gallon tank of gas in your car stores two and a half million btus of energy so 101 quadrillion btus is about equates to about 40 billion tanks of gas which is uh you know an enormous number, there’s seven billion people on the planet uh so about three tanks of gas per person. that’s the amount of energy we used in the u.s in 2018. this pie chart breaks down where that energy goes and you can see most of that goes into producing um electricity, uh the second largest consumer consumer of energy is in transportation uh and then the third largest is industrial, at about a quarter of the total u.s energy consumption. uh and then residential and commercial makeup make up the la uh the remaining part, uh but notice that’s a much smaller fraction uh compared to the uh those big three.

if we then focus on co2 emissions again the total um co2 emissions in the u.s in 2018 was 6 600 million metric tons. and um just to put that number in perspective as well your average passenger car consumes about 4.6 metric tons of co2 per year and so 6600 million metric tons of cr2 equates uh to about one and a half billion cars car emissions, uh again co2 emissions per year in the u.s. and if we uh look at those same sectors if we break that down by sector for emissions um you see the um you know the big three transportation power generation and industrial are are the largest. here um commercial and residential are grouped together and then the the fifth bar is actually agriculture as and that rounds out the top five sources of greenhouse gases in the us. but the thing to point out here is industrial is about a quarter of the total total co2 emissions in the us as well.

Environmental Impact of the IoT

So if we look at the environmental impact of the the iot, or how how iot has impacted um the environment today i wanted to to pick on a few examples that i think a lot of us are probably familiar with for starters and then we’ll dig into some others that maybe we’re not as familiar with. uh and the first one is smart homes i think many many of us are familiar with you know nest sensors, honeywell, emerson all coming out with smart thermostats and smart thermostats have been shown to be able to save on average of 30 to 40 percent of the energy consumption in an average household. uh which you know, if all of us were to adopt that that’s a pretty significant savings. similarly, commercial buildings um in cities that have adopted um uh smart building uh technology uh uh can show a potential savings of um on the order of eight to twenty percent kind of depending on what what that building’s utilization is. and then finally uh just talking about pollution in cities, the WHO estimates 4.6 million deaths each year due to air pollution related causes. which is which when i heard that number that’s just a staggering number uh for me to think about um and there have been iot solutions uh to help address this problem: uh for example by putting air quality sensors distributing those across the city, and then using that for example in korea to do pollution based transit routing which i think is a pretty interesting application, but beyond that just gives you better better visibility on what’s going on in terms of pollution in cities. if we focus our attention then on the industrial iot, and remember the industrial sector consumes about one quarter of the u.s energy and produces about one quarter of the u.s greenhouse gases.

if we just focus on industrial applications and where the iot has helped out i’ll start on the left with steam systems. so in particular steam traps, which are condensate relief release valves in steam systems, that are a common point of failure and also a um a large contribution to energy lost or waste energy in steam systems. steam by the way just um, just a quick side note, is a is still a widely used uh method for transferring energy and industrial applications. um you know it’s been around for over 100 years and it’s still widely used today. uh it turns out it’s a very efficient way to transfer energy, um and every steam system has steam traps, and we estimate there’s around 65 million steam traps worldwide, but that’s a pretty difficult number to estimate. and on average these fail every three to five years, um if you do the math there that’s that’s a significant amount of uh waste steam that just essentially gets recirculated into the condensate line. um wasting energy if you’re using natural gas to produce that steam uh wasting water and producing excess co2.

um so shifting our attention to rotating equipment i think a pretty interesting stat here is that worldwide industrial electric machines use about 53 percent of the world’s electricity. so think about that for a second, you know if we could monitor the condition of all of those industrial electric motors, and identify ones that are not running at peak efficiency and let’s say just improve the efficiency by one or two percent on each one of those machines, we could have a significant impact on the on the complete worlds electricity utilization. um so just because you know these these applications are so widespread even small savings can have dramatic impacts. and then lastly on on tanks and piping on average about 3 million gallons of oil are spilled each year. and the iot has been applied to this problem by monitoring the thickness of for example the walls of pipes, uh or looking for corrosion and identifying or predicting if there’s going to be a leak in in either of those where someone could go do some preventative maintenance there and prevent one of these spills.

Greatest impact from pervasive monitoring

okay so so those are some some spot applications of where the iot has been applied to sustainability but really we won’t realize the greatest impact of the iot um on the environment um until we have pervasive monitoring. so really think about this as you know a world where we have real-time visibility across all assets. and what that allows us to do is is not just focus on uh individual applications and improving the efficiencies of those, for example steam systems or rotating machinery, but be able to but to be able to aggregate all of that information at let’s say a facility level or a regional level um in order to make uh better decisions um on how to operate those facilities. so again really that comes from pervasive monitoring so uh monitoring all assets not just a fraction of them.

so that begs the question: uh how do we get to a world where we have pervasive sensing ?and um that ever active we firmly believe that batteries aren’t the solution, that batteries won’t get us there. and i’ll use this this plot really to talk to that point. um so if we think of an iot device as really the next generation of computing platform where other compute platforms included desktop machines laptops mobile devices and wearables for example. um if we think of iot devices as the next generation of compute platforms um you know really to get to a world where you have one device per thing we’re talking about a world with potentially for example a trillion iot devices out there monitoring these and um today um you know we we’re we operate on the order of about 10 devices per person if you kind of add up all the devices around you and maybe you know how many batteries you you’re in charge of replacing or recharging when they die it’s about 10 per person you know so maybe 20 30 per household you kind of think of it like that. and that’s a more manageable number let’s say so it’s it’s more manageable to operate those devices from batteries um and replace or recharge them when they need it. but when you think about a trillion batteries i’ve got a statistic on the right side of the slide: if each one of those one trillion devices had a three-year lifetime we would still be changing a billion batteries per day. and with seven billion people across the planet that means every man woman and child is replacing a battery once a week across the entire globe, and and that’s assuming each one of those iot devices has a three-year battery lifetime. and how many devices do you own today have a three-year battery lifetime? um probably not too many of them.

um so really this is this is the scaling problem i think that that our entire electronics industry faces and uh analysts projections have reflected this uh reality as well so back in 2012 um analysts were projecting that we would hit 1 trillion iot devices by the year 2015. and you know we’re well beyond that now and we are we’re far short of that 1 trillion number. in 2017 that was revised to 50 billion devices by 2020 and then in 2018 that was revised again to 25 billion by 2025. so those projections have consistently been revised down and further out in time and again um i firmly believe this is because of this battery uh this battery problem that that everyone faces when deploying these devices.

Batteries would be a environmental catastrophe

so shifting gears a little bit you know another downside to uh deploying a trillion devices that all operate on batteries is that you know that would that would mean that there’s a trillion devices a trillion batteries out there that we’re now dealing with and potentially disposing of. and um you know just to add some more statistics around that so for example um in a in a mining town in chile, the percent of water that is used for mining lithium in support of lithium ion battery production is around 65 percent and you know a number like that really just doesn’t scale if you think about um scaling the number of batteries you’re going to be producing by um let’s say a hundred to a thousand x. um likewise if we continue on our current uh projections um we’ll be disposing of of 11 million tons of unused lift of unrecycled lithium by the year 2025, and if i go back to my three year lifetime um assumption for a trillion iot devices we’re replacing about 333 billion uh batteries uh per year. um so really this uh beyond just you know deploying these devices and maintaining the batteries there’s an environmental impact of actually producing and then disposing of all those batteries as well um.

so you know one question is well could we rely on uh battery improvements to uh basically get us out of this jam? um and uh here i’m showing a plot of the battery battery energy density and how it has scaled over the last 100 years, so from 1900s to uh to 2010. and um you know the the um uh the takeaway from this graph is what we roughly see is about a seven percent year-on-year improvement in battery energy density, energy density meaning how much energy i can store in a given volume for a battery; like how much energy can i put in a double a battery form factor? so that’s improved by about seven percent year on year and there have been step function in improvements when we introduce a new battery technology but those steps are not several orders of magnitude those steps are you know more like single-digit or two-digit percentage improvements. so if you contrast that to how the semiconductor industry has scaled um over the last 30 years or 40 years now moore’s law scaling has shown a 50 year-on-year improvement for for multiple decades. so a dramatically higher rate of improvement in the electronics than what we’ve observed in battery technology okay. so this begs the question um can we um can we operate our iot device without a battery? can we run it from a renewable energy source or from harvested energy?

Renewable energy sources at a micro scale

So, i’m going to call this renewable energy sources at the micro scale. meaning you know rather than solar panels on your roof you know think one by one centimeter solar panel or solar cells on a posted stamp form factor iot device. um so you know begs the question uh is it possible to operate these devices from harvested energy? and so for that we have to look at the supply and demand of power in an iot device uh and from a energy harvesting source. so first um let’s let’s talk about on the demand side so how much power does an does a device consume and what is the uh what is the resulting lifetime if i were to power that device uh from a battery? so the plot on the left um shows power on the y-axis and time on the x-axis for a an electronic device an iot device that i’m powering from a double-a battery or a coin cell like a small coin cell like what you might find in your key fob in your car. and you can see in order to get to a one year or even 10 year lifetime on those batteries we need to be operating in the power range of of 10 microwatts to 100 microwatts average power. and again that’s to achieve a three let’s say three to ten year lifetime um at those power levels but that also means after three to ten years you’re changing the battery you’re disposing the device or changing that battery somehow.

um okay so let’s look at what you can do with micro scale energy harvesters so the table on the right compares a few different uh energy harvesters and how much how much power you can expect uh from each of them. so starting with outdoor light uh which is which is really there’s an abundance of energy from a iot uh device perspective um we estimate you know for a square centimeter square centimeter photovoltaic cell you you you can um assume about a milliwatt of power. so you know on my graph on the left that’s about the midpoint um the 1 milliwatts about the the middle of the graph but harvesting from other sources such as human motion like like uh heel strikes for example in smart shoes or vibration like on uh rotating machinery or um thermal like heat gradient um uh temperature gradients uh or harvesting from heat. um and then finally indoor light um you can see rough numbers you know you assume about a few hundred to maybe uh down to a couple tens of microwatts per centimeter square is about how much power you can harvest from those sources. so putting those things together um i’ve drawn this harvested power floor around 10 microwatts where if you could develop an iot device that has an average power of 10 microwatts or less you’ve got a good shot at operating that device continuously from these small amounts of harvested power from these micro scale sources.

How Everactive enables pervasive technology

okay so um so then uh how does ever active enable pervasive? uh remote monitoring well it starts with really our battery list technology. so we develop our own custom uh circuits uh we design them in-house uh and we leverage ultra-low power circuit design expertise that we’ve developed over the last dozen years or so to design wireless um or circuits that include radios processing sensor interfaces memory uh really all the components of an iot device but can operate under that 10 microwatt i’ll call it 10 microwatt floor in quotes. so it really starts with that that that core chip technology. and then from there we build full stack solutions we integrate those chips into self-powered nodes or self-powered devices those are the the plastic devices we can stick on an asset and power it from heat for examplewe connect those uh with an always-on network that we deploy um in in a customer’s environment and then that back calls to the cloud where we can perform additional analytics and derive insights. um to date we’ve taken that full stack system and targeted um specific solutions in the industrial space um and we’ve we’ve really gone after ones that have maximum impact on the amount of waste energy in those industrial facilities which which in turn maximizes impact on the environment in a good way.

so just to dig in a little deeper um to that full stack solution again starts with on the left um it starts with our our custom design chips which includes an always on ultra low power receiver includes ultra low power digital processing so we can do local compute of sense data on the node and also includes energy harvesting power management so we can interface our ship directly to energy harvesters and operate those circuits continuously from harvested power. which is which is another key differentiation between your typical battery-powered device which which quite often does not operate continuously it operates on what’s called a duty cycle or operates intermittently and the reason it would operate intermittently is to try and save power and extend lifetime. in contrast if you can get the power if you can get the active power low enough so that you can run it from harvested power sources you can just operate continuously you don’t have to be intermittent. um so we’ve we’ve developed our own network that then takes advantage of those ultra low power receivers that are on listening all the time to develop an always-on network um that enables uh scalability to um a thousand nodes uh per gateway with millisecond latency um down to any one node um or you know even groups or subnets of nodes. um then on top of that we develop our our self-powered nodes um we call them ever sensors uh these are powered by our custom chips they include energy harvesters they’re designed to operate from harvested energy alone there’s no battery inside so they have exceptionally long lifetimes multiple decades and um and can do all of the continuous sensing um computing at the edge and then uh communicate um through our wireless network back to the cloud. and then on the right at the top level um we this is our really our interface to our customers uh includes uh cloud analytics and a dashboard um that allows you to you know gain visibility into what’s going on uh but also um through our insights as a service model um allows us to you know um deliver um let’s say alerts uh or you know interrupts if you will to our our customers uh to alert them when there is an asset that we have identified as failed or needs maintenance and they can go address that.

so what does that mean for our customers. um out of the back service meaning you know you install a device on your asset it immediately powers up and starts talking to the cloud no additional maintenance there’s no batteries to recharge and replace and lifetimes of multiple decades of our devices and then ultimately reduced energy consumption and reduced emissions saving money.

Stream trap monitoring

so next i wanted to dig in on um our first two products uh just to give you uh maybe ground a little bit what i’ve been talking about um in some actual devices that we’re building and selling today. and the first one is our steam trap monitor so this is um this uh his first product is commercially available today i mentioned before steam is is used in a in a wide range of of applications still widely used today and steam traps are these condensate valves that are a really across industry pain point they’re a common point of failure for steam systems kind of by design for safety reasons and contribute to a lot of waste when they do fail um both in water and co2 emissions. and our um our first product includes um uh monitors uh for determining the state of the trap powers from the heat of the steam and then streams that information to the cloud. and on the right just showing some uh some projected returns for in this case a facility that has a thousand traps in it um a typical facility could run between a few hundred traps to several thousand traps in one facility uh these numbers assume a thousand traps um and just i’ll just call out the the annual co2 savings of 28 000 metric tons by monitoring steam traps for failures and then correcting them once they fail. and again that you know that’s that’s on the order of of around six thousand uh vehicles the equivalent of the emissions from six thousand vehicles that were effectively taken off the road.

Machine health monitoring

our second product is a machine health monitor um which um which is powered by the heat of the motor and monitors electromagnetic fields vibration uh really can uh can look at the efficiency and operational state of motors um and stream that to the cloud as well and enable us to make make predictions on the efficiency of the motor identify when um when motor efficiency is being impacted by let’s say the need for lubrication for example is a common maintenance maintenance item for machines or in the in the long term predicting failure. so ideally uh with a continuously monitored motor uh we can uh reduce the electricity consumption of voters and remember uh 53 of the world’s electricity goes to industrial rotating machinery as well as um extending motor lifetime which means fewer replacements and um and less cost there. okay so so kind of summarizing the um uh the returns uh for these two products um on the left for our steam trap monitor again this is uh this product is commercially available today at the end of 2019 uh we had uh 1251 uh traps that we were monitoring and if you project that number over the the um the entire year the impact of that monitoring it equates to a 34 000 metric tons of co2 saved or 7 000 cars taken off the road and 59 million gallons of water saved so significant savings there realized just by monitoring those traps. our machine health monitor um is um we’re currently prototyping uh we’ll be releasing um uh later this summer um the uh the initial production release of that um but you know just again showing some numbers there assuming a two percent improvement in motor efficiency by adopting a continuously monitoring solution like this you can you can see the matrix of in this case dollar saved but um for the case of ten thousand motors which is not unrealistic uh for a a single facility um in 200 horsepower motors um that equates to roughly uh sixteen thousand homes electricity usage for a year of savings just by again improving efficiency by you know a mere two percent um just because uh uh because the essentially the scale is so high.

um okay so just wrapping up um i wanted to talk a bit about um our next generation developments at everactive and and really where we think this can go from here. and i’ve broken this down into and basically two categories and i’ll talk about those separately so on the left .um we are steadily improving the performance of our our custom circuits that are powering these devices and targeting uh what we’re calling an ever active mode at a power consumption of one microwatt so on continuously communicating and with again a power of one microwatt well below that 10 microwatt power floor i was showing earlier. uh which what that does is allows you to develop devices that are smaller form factor use smaller harvesters or can harvest from uh much less let’s say dimmer indoor light or lower temperature gradients for example. um and um our next generation device uh we’re designing uh to be a roughly a form factor of a like a double stuffed oreo cookie and uh within three years uh we’ve got line of sight on enabling devices that are posted stamp size both in you know size and thickness and form factor you pull this together and you have smaller smaller volume and longer lasting sensors really means less electronic waste which is another i think concern you know with again deploying a trillion devices you don’t want necessarily all of those devices to be um large uh and you know producing a bunch of plastic and other electronic waste.

and then on the right side um uh we are we’re also looking at improving the amount of compute we do at the edge so that we can uncover new insights on the device itself rather than shifting that data to the cloud and improving our wireless communication range. um all of this intelligence at the edge uh results in a smaller cloud footprint which means less processing in the cloud and less data storage in the cloud, so reducing the footprint there by pushing all of that out to the edge where we operate those devices from harvest power and um all of this together leads to a more sustainable future.

Alright so with that um i want to thank you again for your attention um and um i’ve included our contact uh myself and brian’s contact information here um feel free to if you have any questions we have time for um questions now if you do have questions please use the uh the q a feature and zoom type your questions in there um and i see some people have also put them in on the chat um and so i’ll hand it back over to brian to moderate the q a session but thank you for your attention

BRIAN:  Thanks Dave. We’ve got a lot of questions here, so let’s see what we can get through here. First question is: what kind of data rates can you support with a kilometer range? Is this different for downlink versus uplink? What sort of reliability can you support?

DAVID:  Yes, so the difference for downlink and uplink is true. The data rates we are supporting today are for our ultra low-power receiver on the order of a few kilobits per second. We primarily use that receiver to offload networking traffic so we don’t use that for primary data communication. We have a separate one, we call it a communication radio, that handles the bi-directional data communication to do things like over-the-air updates of the software on our nodes and to send data to the cloud. Our comm radio operates at 200 kilobits per second and then again our ultra low power receiver is um around a few uh kilobits per second.

BRIAN: Okay, thanks Dave. This next question actually combines a few different questions: how close do Everactive devices need to be to a hub? And what is the network topology with a mesh network hub and spoke etc?

DAVID:  Yeah um, good question. So our our first product uses our first generation silicon, and the range that we have observed in our deployments, you know i mentioned at the end of 2019, we had roughly 1200 devices. We’ve got many more than that now deployed. The rough range we see is around 30 meters to a gateway and that is in a dense industrial deployment so line of sight is more like 100 meters. We do not use mesh networking, we use a star network to talk directly to the gateway and that has advantages in terms of scalability, ease of deployment, enabling  mobility and reducing latency of the devices. So there are lots of advantages to being able to talk directly to the gateway, rather than having to multi-hop with a mesh network so we’ve adapted the star network topology. I want to mention though our second gen product, our machine health monitor, is built on our second generation platform — which again we’re prototyping today — the range that we’re observing with those deployments again in dense industrial environments is around 250 meters. So from a gateway we can support up to a thousand devices for a single gateway with a radius of 250 meters in a practical industrial environment. 

BRIAN: Thanks Dave, moving on here next question: what’s the lifetime of an Everactive sensor before you have to throw it away?

The IoT has caused there to be a lot of batteries, this batteryless sensor removes the need.
Everactive’s forever sensor, which removes the need for batteries.

DAVID: Forever. It’s a forever sensor. It’s kind of tongue-in-cheek, you know the lifetime. So there’s no battery and there’s no battery to wear out. We do local energy storage on the device, but we do that in a capacitor and capacitors have much longer lifetimes than batteries. We have performed extended lifetime tests on our first generation device and the lifetime from those tests I am told is greater than 20 years, uh limited by the testing that we did not do because the device failed. So we’re trying to build forever sensors and we anticipate decades of lifetime.

BRIAN: Okay, thanks Dave. Moving on for energy harvesting: what happens if the source goes out, the lights go out for example, and there’s no harvesting available?

DAVID:  Yeah yeah, good question. We have been thinking about low power circuits and uh batteryless operations. I mentioned we’ve been doing research in this space for about a decade now, or a dozen years, so we’ve been thinking about that exact problem for a long time. The way to think about that is basically there’s three components to how energy is generated, or stored, or used and really it’s kind of in that order. You have an energy harvester which produces power. We have energy storage which is done in capacitors, not batteries, so we can store energy locally in the device. Then you have energy consumption, or power consumption, which is the electronics like the sensing, the processing, and the wireless communication. And you know, there’s this kind of multiple trade-offs that you can make when designing a full solution which include things like volume, size of the harvester, what you’re sensing and what you’re processing on the device itself. 

But we have tuned those solutions for these first two products that i mentioned uh for the steam trap monitor just to give you some real numbers. For example, for the steam trap monitor, the steam shuts off and all heat goes away. Our monitor will stay active for about a day or day and a half after steam goes away. That’s really plenty of time to see the whole steam system cool down and then sit there cold at room temp once steam comes back. It immediately cold starts, fires back up, and starts talking to the cloud, our machine health monitor, where we’re engineering that to be more like a few days to a week. Once the motor shuts off, heat goes away the device can stay alive for a matter of days but that is something we can engineer in the final solution. 

BRIAN: Alright thanks, Dave. We have time for one more question, looks like we’ll end with this on. How does Everactive compare uh to the other ‘low power iot offerings’ on the market?

DAVID: Well first of all, uh we’re batteryless. Second, we offer a full stack solution. So rather than just delivering a single low-powered chip, or even a low-power sensor, or a low-power class, or I should say a cloud platform, we’ve integrated all of that together to deliver a full CAA solution, for our industrial customers that’s a key differentiator. And then we’ve been talking a bit about the wireless connectivity on the wireless side enabled by our ultra low power receiver. We have the ability to scale to 1000 nodes per gateway with millisecond latency to each node, with a very simple deployment, repairing, and provisioning process. So from ease of installation, um essentially zero maintenance after that, and then full solution for our customers, I think all that together differentiates us from other IoT solutions today.

BRIAN: Thanks Dave, I’ll turn it over to you for any closing remarks, but we’re at time for questions. Alright, anything else to add then?

DAVID: Yeah, sure. Well, I thank you again for your attention today and attending. You’ve got our contact information on this slide, so feel free to reach out if you want to discuss any of this further. Happy Earth Day!

 BRIAN: Alright, thank you!