[0:00] Spyros Sakellariadis, ICONICS Global Director of IoT Business Development
I want to play a video of the implementation that's used for building management at Microsoft. Microsoft has 770 buildings, give or take, worldwide depending on the day, and about 125 in Redmond. There's actually a debate on whether it's 125 or 160. The difference is what you call a building depends on whether people are in it. So, if a garage doesn't have people in it, it gets called a structure as opposed to a building. So if you see different numbers, the reason is that different people count them in different ways. But what I'm going to show you is a video of Tearle Whitson, who is the guy that managed all those buildings using ICONICS.
1:30 Tearle Whitson, CBRE Senior Building Engineer
Good afternoon, welcome to the rock. This is the Redmond operation center. This space is primarily used for command and control. And that's where we host a lot of that out of the building automation systems. We also have used this space and leveraged it as the deployment and operational side of the energy smart building suite. So our energy smart building suite allows us, which is the back end of the ICONICS platform, and polls and aggregates all of the building data from the building automation systems into an Azure hosted SQL database. That SQL database is then synced up to a data lake. And also on the front end allows a software suite to run set algorithms to analyze the existing fault detection process within the buildings and to determine where their energy and money saving opportunities. A lot of the information that you get this information that was present in the building automation systems prior previously, but it now aggregates it puts it into one system allows you to look at a lot of that information in different areas in different places all together at once. So, one of the things that we can see in this area is allowed to see baseload and what's the live demand for the campuses at any given time. This is where you're going to get base information on total power consumption. This point this is segregated on Puget Sound. But we have portions of that for our Asia campuses. And we're currently working on deploying the European campuses right now. So, what you'll see in live data is a real nice piece of snapshot information that normally you would have to pull utility data from a utility company pulling bills, and you get it 30/60/90 days in arrears, and then you never get real. This is real time consumption for the campus. And I can actually historize that information, demonstrate it, look over time to see how the campuses come on and off. Has it shut down actively over nighttime periods? What did it do over the course of a weekend? Do I see a trough where a weekend should be? Or did I have a large amount of assets coming on at any given point here that weekend. And for even some of these smaller anomalies, I might want to explore looks like I had a small startup in the later half of Saturday. And I may want to evaluate that I can double tap, pause the pause, and zoom into that area to start to evaluate how much did I change was it you know, up to a megawatt in that small area that I might want to evaluate even in more detail. I can take this information and go out to a trend analysis for you or to look at it in more detail. But this gives me a really quick high level a piece to kind of analyze the campus at a glance. And again, some nice built-in metrics just to be able to look back. What did we do over the course of the month? Is it significantly varied week over week, month over month? It's nice to see especially when we take seasonal shifts when you come from colder to warmer weather warmer colder. How did the campus handle that; same thing with our fault data? What's been the fault count over the campus over that time, and it looks like one of the things I know I had noticed, I’ll push it off to a year we'll see how well it responds with this. I think actually we only had the aggregated number and faults on this dataset since August. But you can look at the data we've been finding is that over the cooler months we’re finding a lot of extra faults. We're finding also a lot of change in comfort index over that same period of time. I have grown up on this; that's on another display. But average comfort has been dropping over the colder months as well. And it's a nice piece where all of these sets of data are pieces that technicians and engineers can evaluate. But with this dashboard, it allows you to at a glance, not just my trained technicians, but facility managers, business executives, anybody coming through, data scientists coming through, can look at this and they go, “Oh, I can do this with that dataset”. And it really allows a lot of different eyes and minds to say, “What can I do with it”. And this becomes really kind of the cornerstone for where the smart building and the smart city process progresses out. “What can I do with these pieces of information as I get drill in, as I am able to navigate the system back down into Puget Sound?” There are a lot of different methods of aligning this data and viewing it. The touchscreen allows for a lot of cool parts of it, where you can just really zoom in and zoom out. But some people like to operate out of a nav tree and hierarchy and see how things are connected. A lot of times we found that different people prefer to just being able to zoom in and look at the campus at a glance.
[6:31]
This info allows me to see the heart of Puget Sound and the main cluster of buildings and see some high-level information around, “Do we have utility power for the campus?” If we don't have utility power because occasionally, we take power bumps and power transients, especially during inclement weather, do the generators startup do the life safety generators come on. So, when I'm dispatching technicians and teams, I send them to the right place and not just send them to go drive around campus and listen for generators. So, you find that you're stepping 10/20 years into the future. Where as we were only a few years ago, we were running the same process that might have been going on in 1970s. To find out whether or not was buildings or how were the buildings operate. And now we can actually live aggregate look at data, look at the systems to know how they are performing. What's going on with them? And are there events happening within the building envelope? So, we can really move around and drill into more detailed pieces of information. I can continue to look in my Puget Sound architecture to get my list of building. I can use this as well to navigate and allow it to move me around over the city center. I can select the building; it'll pop open with a building navigation window that allows me to see high level information for that building. And at a glance, pieces of data that can really help you get a good snapshot of what's going on in that building. City Center happens to be our biggest building in the portfolio. It’s about a half a million square feet, a little over 2000 people are assigned to that building. And that's only assigned headcount; it's not live headcount. It's one of those data points that we're looking at testing to see who's on the test that we'd like to track to and be able to populate that with full time, real time. Data quality, how connected is the system? There's always a question of data connectivity on our buildings. Probably the biggest single biggest piece of important information is how connected are the buildings because often equipment coming offline project work going on in the buildings. There are a lot of pieces of data that we miss. Again, this information allows me to look at the energy for that building specifically, and I can expand that view to see it in a larger window to be able to get more detail. But I can still pause it to actually be able to look back and do the same navigation. Or I can drill in and look at specific details. This case is actually a bad point to evaluate the building because the building is starting strongly in energy. A lot of times it's because it's getting too cold outside. And then first thing in the morning, it doesn't soft start the building over time. It's starting all at once and bringing a lot of units on simultaneously causes a really large demand spike in energy. And that winds up being a little bit higher charge to the utility.
[9:41]
Other ways and information that I want to see here could be comfort per floor. And I'll show you a little bit more detail on that in a second. And then we also see fault rules and the active faults for the building all up how many pieces of information of equipment might have active fault detection item currently present on the system. Typical fault rules: it's nothing crazy, it's an” if- then” statement of compare acts. It's just trying to compare data sets, one to another. It might be saying “If this room is supposed to be at a certain temperature, and it's not perceiving that temperature over a specific period of time, I might have a fault whether I'm overheating or I'm under cooling. In many cases, it might compare that I have that damper open, like we saw in the test setup of that damper open and pushing cold air into it. But at the same time, I'm heating it as well. That's a simultaneous heating and cooling. And even though I have an occupant that might be satisfied and happy, at the same time, it’s costing a lot of energy. And so, it's this type of information that is the backbone of what we do with the fault detection and diagnostics. And this has been one of the core pieces that we have focused on over the last three and a half to four years as this system has been live in production, it has led to about six and a half million dollars of straight reduction in utilities spend year over year, which is about 18%, considering the persistent savings from year to year.
[11:11]
So, it's actually been a major achievement for getting at Microsoft energy consumption and meeting Microsoft’s sustainability goals going forward. But a lot of our information and interest now is starting to look at user comfort. And using this comfort information. One of the first things that we did was to look at that at a floor level and be able to see those temperatures and see that comfort level at a format. And this allows us to see what the current trends are on a given floor. And typically, what we're looking for is our normal design ranges, the green zone here in the middle 70 to 74 degrees Fahrenheit. And anything outside of that, we'll highlight by another color and heat map it and shade it and look at that shading. And this lets us see that it may or may not be that hot in the zone. But I might have some small amount of area in there that I want to look at. I’ll bring it down, and I can look to see that we have a set point of 74 degrees, and actually 72 degrees in that space. And we are running at 74 degrees, so two degrees above setpoint. In this case, so the zone is going to continue to try to cool, but it lets me know that that zone is running just slightly above our above it or our range there. So, this information is nice piece at a glance, but it's floor over floor. And this lets me see one floor at a time in the 27-story office tower. That's only one building out of 125 in Puget Sound, and 160 plus that we're now connected to across the globe. If I turn on the fault indicator, I can highlight the zones where I have active faults. So, in that zone, I can select a fault, and I can look to see that I've just got a set point fault. And it may just be the set points for that zone are configured outside of our allowable range, but it allows me to take action on that. Meaning as soon as I see it, I can create a work order. And this information is the baseline data fields that we would submit to the CMMS system and push that out to where we would drive a work order. And then we would have a technician respond to that work order and make a correction.
[13:37]
And it's really about where we want to go with the data; as much as possible from this habit that a technician can have less steps in between of transferring data, looking at one interface and looking at another. If you can come right from this live system, push that false connection out there, then the technician can drive a lot faster directly to a fault resolution and problem than having to open this system, or open another system up, come over here punch in a ticket and then move back out before this can that direct data connection. It carries in equipment type, what's the asset? What's the problem type? What are you? What are the categories? What are the savings numbers and fields in data fields on the other side? But it allows us to see a lot of information at a glance and start to get a good feel for where we need to get people moving on problems and how we promptly analyze bigger problems that might be on campus.