 Rwyf, wilt eich pryd yr hi'n ei gael amser yw bod fel. Dwi wedi bod cymien o'r busn o'r ddwnys. Felly, y chyfr sefyddiwch arall yma yw'r eich cyserio'r ddwnnall yn mynd i'r ffyrdd ychynig o'r ddysgu o'r ddiwch ar hyn o ffyrdd cymryd yn y gyfyrdd cymryd gyunedol yn gweithio'r ddichu ymwhysgol. Ac mae yna bod ddigon i'r Fythryd Cymru, eu oeddwn i'r Fythryd Cymru, bydda'r cyfnodd hynny'n amgylchedd heddiw o cefnaintu Llyfridol. Felly bydda'r cyfnodd hynny'n amgylchedd wedi'i awdd. Nid a'n deall, maen nhw y ffrinsio a'r twfynodd y cyfnodd hynny, ac yn dweud dddwyd arweith y byddch chi'n CO2, ac mae hynny'n ymdodd hynny'n iawn, roedd yn ddiogel. Mae hynny'n cael ei ddweud ddwy i'r cyfnodd hynny ar y modd. I'm going to talk specifically that the excuse for this talk was something called heating as a service, but if you remember that my aim here is to cut carbon and are kind of, we have missions and visions and so on, all that's a nonsense, but basically this is not for big companies or something, this is for normal human beings who aren't stem or early adopter to help them cut carbon in a way that's easy, preferably makes their life better as well. So we have indeed, we're an overnight success, I started this project in 2012 and we've nearly finished starting and we will actually hopefully have a fair amount of product out this winter and we're happy to accept orders in principle, but the bigger mission is what can we do to help ordinary people save carbon The thing that David Mackay got me on was the lowest hanging fruit is domestic space heating in the UK, so something like half of the energy that every single one of you puts into heating your home in winter is wasted, you could not have used that energy and be just as warm. Now there's all sorts of reasons for that and some are better than 50, some are worse than 50, all sorts of reasons, but we were trying to tackle that problem could we do it better. Now I'll talk about Radbott in a minute, but we just finished an innovative UK project about heating as a service. So what's all this about? Well I've stood in front of the CEO of Centrica, which owns British Gas, and so to him you know you are going to have to stop selling gas fairly soon, which is true because if he doesn't we're all going to die, he has a problem that he has shareholders that would like to go on having revenue, you know dividends. So how does he square those two? And so heating as a service is part of the puzzle. So at the moment you're a householder, a typical householder in the UK, one of the 20 or so million households on the gas grid, then you buy gas by the cubic meter or 100 cubic feet or if you're feeling really advanced by the kilowatt hour from your, normally you're buying by the kilowatt hour from your retailer, you're then, most people don't understand how it works, they don't know what a kilowatt hour is and they don't know why it's not the same as the electricity kilowatt hour and across all those bridges later, and you then run the piece of equipment you don't much like or understand as well called your heating system and by the way half of the, we don't have really very good building regs at the moment, they've also fractured so it's not UK wide, but even English building regs says a house must have a minimum of thermostatic radiator valves, we'll come back to that in a moment, a house thermostat and a timer, less than 50% of houses in England have all those three. So you're running a system you don't understand, it's probably cranky, it's old, it doesn't work very well and you don't know how to use it and it's in British gases interest to sell you as much gas as possible. So the point about heating as a service is to not pay British gas for the gas but pay British gas to keep you warm and then it's in their interest to use as little gas as possible to keep you warm. And a kind of classic example of this is a local authority that needs to light its streets. So normally they have to, oh they don't have anyone really technical but they don't need to have this many lumens and they need to hire an engineer to tell them what the lamp post should look like and they put in the lamp post and then they try and make it efficient and they try and negotiate with the electricity company. How much better, and this does happen sometimes, the local parish council goes to EDF and says you put up our lighting and meet the legal regs. Suddenly it's in EDF's interest to use as little electricity as possible to do it so this model does work, you can make it work. We were trying to find ways of bringing heating as a service to the UK environment to start with. Now I'm pleased to say we've got funding this year so we've had lots of help in the past from Innovate and Bayes and DEC and so on. But we kind of jumped the shark at the end of the project because we'd actually got investment in before the project finished. Now we haven't solved the problem and I can't tell you what it is but one of the energy companies has put down an order to try our stuff out. It's not our full order but this is exactly them trying to find ways of providing service in a different way to end users which does not naturally involve selling more carbon pollution to you. So that's what heating is a service about. Now I've got some notes on here. If you want to go and see this by the way, I put notes up which is roughly what I'm going to cover. You can go to my website which is earth.org.uk and you'll find there's a column of new articles and a top new article is this one, the notes on this talk. But I don't know how many of you remember the sort of slash dot or use net joke where someone, there'd be some terribly naughty social problem and some teenager would say, oh but you could just fix it by, you know, and then some small technical solution. And then the joke which goes, you know, propose technical solution, oversimplifier. It won't work for all of these reasons. You don't understand the issues. No one will cooperate with you. Local countries involve the politicians hate you and so on. You can't solve social problems with purely technical solutions. And one of the issues we have with the way we run heating for example in the UK is that people are familiar with the way things run. Also people are busy and I don't know about you but normal people don't want to spend more time with their heating. So how can we find ways of making things work for people in a way which achieves the outcome I want because I'm selfish about this. I want to save the planet. I want them to be doing things which do what I want. If I give them a new boiler, we know that a technical measure will go on saving energy for the life of that boiler. It's pretty well understood. Say 10 years is a reasonable estimate for any technical measure. Apparently if you look at the research, if I have a measure which requires the user to do something different all the benefits stop at a year. Well actually that's because all the studies stop at a year. So if you reinterpret the data a bit, maybe they last four years. But the issue is we want to try and do things which line up with what the users want to achieve in their lives anyway. So if you're a marketing person you want to know what for that user are the pains, gains and jobs that your products might solve for them. You want to line up with those, they'll go on doing it. So for example in a school and we put in a project request for a big project request with DEC for a school system. Teachers are busy. They're absolutely maxed out doing a teaching and learning thing. We are not going to spend a lot of time learning how to end your... I'm a resources chair from my local primary school. And it used to frustrate the hell out of me that I come along on a windows afternoon to collect my kids and all the windows are open. Because they're a bit chilly in the morning. They turn everything up to max. Forgot about it. Massively overheated and opened the windows. You'll see why all this is connected to Red Bot in a moment. And yet you do need to open the windows sometimes because when the air gets stuffy it's actual levels of CO2 which cause people to lose concentration. And interestingly we have just built a new annex on the school and they have put in something which automatically opens the window when the CO2 level goes up. Now the nice thing is that's in line with what the teacher wants to do anyway. The teacher wants to have it easy to teach. So if you can line up your energy saving with what they want to do anyway, bingo, then I think the methods will last. So what's all this got to do with open source? So Red Bot, so I'll show you what an old Red Bot looks like. By the way we did some thermal testing the other day and it turns out that you have to be careful where you're pointing the heat gun. This source is rather more open hardware than what you'd like because it's melted under the heat gun. That was a predecessor. This is what the new one looks like but I now have to kill you all. So we decided that because people in this room, and we have to accept that none of us are normal if we're in this room, no one in this room wants to programme their heating system a week in advance in half hour increments. I met the chap who does, he works for the government as it happens and showed me proudly. But most of us don't and I was propelled into OpenTRV because I bought a market leading programmable radiative valve and we know from research that we can save 30% just with programmable radiative valves. Now I love to tell people I have a master's in theoretical computer science and found that thing rather difficult to understand. So if I find it difficult to understand how will people who have actually useful things to do in their life like looking after their children or something going to spend the time to understand it. Also many people can't programme things so what we were doing with Radbot and if you go and look online you'll see links to our repositories and so on for the open code and the hardware is a system which programmes itself. So our whole point is you don't need the internet, you don't need a smart phone, you don't need to programme anything. The hardware is cheap enough that it pays for itself as a year. Did you know that 25% of people in the UK live in rented accommodation, the average tenancy length is a year and the payback time on any of our competitor products is at least three years. So you can see why there's a little bit of a gap there that we felt we needed to fill. We've had lots of help from people in this room already on things like getting the cost of the hardware down and that's now finding its way out to manufacturing. We're not quite at that one-year payback time but we are. But it's really important, it does not need programming. What I'm after for the end-user gets payback in a year, they get something they don't need to programme, it just works, it learns their patterns. We've got all those buzzwords, IoT, machine learning, blah blah blah, I often have to pitch this stuff and say I didn't put those in just to double the value of the company. We solved a problem and realised it had those buzzwords and then we got lots of money out of the government to have those buzzwords. But the interesting thing is if we get this right we can save 200kg of CO2 per radiator per year. Did you know that there are a billion radiators across Europe? Half of them are in homes, that's really quite a lot of stuff that can be saved and our target is to save multiple percent of the entire EU carbon footprint with stuff that people can fit themselves and will pay for itself in a year and we wouldn't achieve that without open software and hardware. So that's what we're about, that's open rights now. So an interesting thing discovering doing this, doing lots of research and going to conferences and so on, is people's... it's easy again, us non-normal people in the room, we presumably have a scientific and technical background, many of us and so on and we understand, okay, well heating is energy, lighting is energy. Well it's quite interesting, one of the things it turns out is that it's of an age like me around 50 who remember these switch-it-off signs that used to be by every lights switch in our classrooms and so on, exactly. Ken, you're not a day on this, anyway. And apparently those lessons have really stuck and so if you go and interview people in offices and say, do you save energy? They say, oh yeah I turn the lights off when I go if it's one which is not automatic. But then they say, but why do you leave the window open? Oh because the radiators are a bit hot. And it's quite interesting, people have extremely poor folk physics so in this particular area it is no good expecting people to learn physics to understand that heating is like lighting. And so again that's another reason to automate and again, you know, our algorithms are open, at least the reference versions of them, we can talk about it. You have to understand it's a bit like this categorisation of designs, you don't necessarily know that people even know the right questions to ask. So it's kind of important to do as much thinking for the user as you can. From my point of view I love distributed algorithms and I like the idea that my valve controlling algorithm might be on several hundred million European radiators. And furthermore, not only controlling heat, we're not all just about valves, remember we are ultimately about saving carbon. Here's another stat for you. Roughly 20 million homes on the gas grid in the UK. Everyone of them has at least a 50 watt circulation pump in. If we can drop the heating when the grid is having a bad hair moment, that's a gigawatt of power we can take off the grid. So, you know, that's quite an interesting little tweak to an algorithm we can make somewhere. So a radiator valve sounds like a very mundane thing and when I started this my first implementation was in Pickaxe Basic and it was, you've only got 2,000 bikes or something, it was a couple of hundred lines of basic was the first implementation and it runs okay on that. But it's got kind of small sophisticating, we've dealt with more edge cases and we've made it more stable. So now, when we last totted up, our code base is 100,000 lines of fairly dense C++. Now, for scale, when I joined Lehman Brothers Fixed Income trading in London, somewhere in the 90s, their entire library stack that we were helping the world with was about a fifth the size of that. Now, I was the first person to write any unit test for any of their stuff at all and may that gloriously continue. But the fact is you also need to test it here. The trouble is you'll note that it's warm outside now which is a problem for someone testing a radiator algorithm. So another good reason to do it, there's a very good talk at Fozdem. The guys at Bosch who are developing heating control stuff for boilers is build models of the world, get them as accurate as you can, run your tests against them like you would run software tests so you can test your algorithms in the normal way, occasionally test them against the real world and you're testing both your algorithms and your models. And that's something we've taken quite a heart and we've got, can't tell you how much data we've got scrolled away and our unit tests chomp through every time we check in the change. Very important. So we're making our hardware development a bit more like software because we've got a faster testing cycle and more thorough testing. And then just this last thing I want to say. How am I doing on time? Get off. You've got eight minutes. Okay, all right. Oh, okay. And so our design so far, so there's an interesting thing. So there's an open source project. Now we have, in order, if we're going to get on any of those 400 million radiators significantly, we need to have a business behind it because we're not going to the government for handouts pay to have 100 million of these made or build a Swansea lagoon out of them or something. We need to have this as a self-standing thing which has customer pool. They want it. They put it in themselves. If the government mailed you a clever thermostatic radiative balance and put it on, most people just chuck it away. If, however, it's an attractive product that they actually want themselves, they'll pay attention and use it. So we need to be a self-funding company, a growth company to have get the scale. So there's the open source project is still there and it's still in the 90s of the code is still sitting there, 90%. There's a little bit of secret source on top for the commercial products so our investors can get a return and we can grow the company. But even with that, we have manufactured, we have designed this product so that it's got a nicer to programming pins that you will all know and love if you have an Arduino. It thinks it's an Arduino, no, for example. You can reprogram it. We haven't released the circuit diagram yet but it's not complicated. We will do. So even on this commercial product and there's a delicate balance I have to negotiate, which is still a discussion. But when this thing comes out, if it's available, it will be available to tinkerers as well. And I think it's important because we benefit both ways from having access to that open code base and having feedback. Actually almost all our device drivers, which are not big things, we're running on bare metal, have been written by people in our open source community. Possibly people in this room. You were doing some of the CAD stuff for us, weren't you, for earlier plastics or something. So people in this room have contributed to that project in various ways. So I just wanted to say, and the last thing is someone else mentioned money. If you are wanting to do something technical like this, particularly if it has a good outcome, you guys may have noticed a little thing. You may not have noticed a little thing called Brexit coming up. And small companies like us, 80% of our market may be on the wrong side of a non-tariff barrier in blah blah blah. So the government is throwing absolutely piles of extra money over the wall to small businesses to support technical things. So if you've got something which is commercially viable, there are enormous numbers of things going on. I can't tell you about a current one. But we've had quite a lot of success in the past with getting funding to develop bits of this. So if money is your only problem and you have a viable product, particularly if it's something like clean tech or medicine or something like that, there is stacks of support at the moment. I've never had it so good. I've clearly not Richard Branson except for the beard, but I've been doing start-ups for 30 years and now is as good as it has ever been. So you were hanging back just because of that? Don't. Right, that's sort of it. Now if there's any questions, yeah. So the aim is, it's two AA cells, an aim is for it to last at least two heating seasons. Now if you're in the north of Scotland and I did a little poll again, I said who's still got their heating on on the 21st of June? I, you know, summer solstice. Well the bloke in northern Scotland says what's off? So he might get a bit less life out of it. And the secret I can't possibly tell you is that the aim is probably not even the next version. You know, not the one that's going through manufacturing now, quite possibly not the one after that, but the one after that we're hoping to have actually powered by energy harvesting. So there'll be no need to change batteries, because I see having to remember to change batteries as being a real inconvenience and barrier to being used. You know, they'll fail and people will say, why is it broken? Well, you know, you do have to change batteries every two years, and that in two years time you won't need to do that. Well, so it's interesting. That would slightly complicate our testing regime for safety. The boards are all actually run off internally, although it's two AA's would happily run off USB, and if you take earlier boards of this, they've actually got a USB connector on. A little sort of straw poll survey says that lots of people's radiators are nowhere near their plug sockets, so we would have to deal with it. I think much better is if you can get the energy harvesting just a target in terms of market, although it will vary, is if you walk into B&Q now, you can buy a mechanical TRV that costs £3.99. You're lucky if it doesn't wound you on the way home, but it exists. You can pay 30 or 40 quid for a lovely one with chrome and probably bamboo. Neither case of those do you need to remember the battery. You probably looked at the nest and stuff as you went past and thought, oh, that's like 150 quid. I can't do that. So I'd like us to sit firmly in the middle of the mechanical ones because no battery is needing to be changed at all. Did that answer your question? So two things that make it easy. Yes, you could plug USB. Our previous board designs, and the new board design is hardly different to the current one. The only new thing is that we've got a stepper rather than a brushed motor in that one. So yes, you can take one of our existing board designs. If you drop me a line, if I can find a spare board, we're a bit pressured at the moment getting into manufacturing, but if you drop me a line, we've got a spare board I can even send you on. One thing is that our radio interface is modular. We actually don't care as long as you can send a 64-byte frame, 63-byte frame, we don't care what the radio is, so you can easily drop in a Wi-Fi interface if you want it to. That's what it's there for. So it's trivial. So lots of people don't even have a thermostat or anything. There are lovely stories. We tried this out with one of the people who shied out was a member centricus staff, and I said, so what do you set on your timer? So what's your daily schedule? What's your timer? Oh, I make my husband get up and turn it on and then come back to bed for half an hour. So lots of people are not using the facilities already there because they're already too complicated. This has got occupancy sensing built in. So if you get something like a nest, it turns the whole house on or off on the basis of whether anyone's in the house or not, and that will save you about 30% so that it's not on when there isn't someone there. There are various complications, but nonetheless it's a good start. The problem is that why are you hitting the bedroom when you're in the living room all day? So we are providing zoning by radiators. So for most people that's effectively by room. And that makes the difference. Now if you just run these things stand alone, so there's no communication on any of those, they just are replacements for done mechanical valves. We got savings. We expect savings up to about 30%. We did see those in a trial. I'd be more concerned say 15%, but 15% to 30%. If we also, these all have a radio in them. They have encrypted comms. If you have a unit which can listen for them calling for heat and switch your boiler on off instead of the thermostat, that should rise to about 50%. But you're only going to get that by a combination of correctly controlling a temperature in each room and sneakily turning off the heat in the room when the human occupant is unlikely to notice when they're asleep and when they're not there. And that's how we do the extra saving. It's not PII. It's simpler than that. It's just ambient light detection and some cleverness. So... People don't want to go into a cold room. No. The reference version of the algorithm out there in our labs, in GitHub if you really want to look. But it's a kind of risk approach. So very simply put, there are some really strong clues that you're around, which is such as you're switching a light on or drawing the curtains. Those are big clues that you're there. You probably want it to be quiet and cooler for various reasons. Even if you're watching a TV, you probably want it to be quiet and you can sit under a blanket. And it has a rolling 24-hour memory. Which fades out after about a week. So its memory is about a week long. So it's making predictions about when you're likely to be back in the room. But also if the room is light even if it thinks you're not there, it's not doing it by turning the heat off. It's by controlling a setback. So you set a nominal temperature on the dial and the less likely it is it thinks you're going to be in the room. The further it will let the heat setback from there. When it thinks you're going to come back in it reduces that. But even if it doesn't think you're going to be there the maximum setback when the room is light is three degrees. When the room is pitch dark and has been for a long time it's six degrees. If it thinks you might be coming back in the room now it's one degree. Because you're not even going to feel the difference of one degree. And every degree you setback so in theory people draw these lovely comfort diagrams showing you need this sort of thing this is the temperature along this way. I was firmly told at this DEC meeting where David Mackay was there when he was chief scientist no one can be comfortable at less than 19 degrees on any circumstances. Here's our diagram and David Mackay said I keep my house at 14. Nonetheless there are charts that show temperature, humidity and so on. I don't use that in the algorithm but what we do and this one does have a humidity sensor is if the humidity gets high over 75% we raise the setback temperature to a minimum of 14 degrees. Now you know the old joke to engineers three opinions try humidity and condensation risk papers but nonetheless 14 seems a good guess I mean sort of engineering something or other for reducing the risk of humidity. Anyway remember we're measuring temperature at ankle height on something which may not be in the coldest part of the room but in other terms at other times we'll let the temperature drop as low as six. In this case we artificially raise it if the humidity, relative humidity is high in the hope that that will avoid that. Okay.