 From San Francisco, it's theCUBE, covering OSISoft, PyWorld 2018. Brought to you by OSISoft. Hey, welcome back, everybody. Jeff Frick here with theCUBE. We're in downtown San Francisco at the OSISoft called PyWorld 2018. They've been doing it for over 15 years. There's about 3,000 people here from all types of industries, really sharing best practices, about using this software solution and the data that comes out of it to basically find inefficiencies. And we're excited to have for our next guest these Josh Reigns, a senior metering engineer from MIT all the way out in Boston. Josh, I'm glad we could get you out of the snow. Great to be here. Absolutely. So a little bit about what you do at MIT. I deal with the campus metering, campus energy metering. MIT brings in electricity and natural gas and then makes electricity, chill water and steam and then distributes that to the rest of the campus. I deal with all of the physical meters in the building, the data acquisition hardware, as well as the PI system that hysterizes all of that information forever in a day. And just for people, kind of general scope, how many buildings, you know, what are some of the kind of top level numbers of all the systems that you guys are keeping track of buildings, et cetera? We've got about 120 buildings, I believe. At the moment, we aren't metering that many. However, we are pushing out a lot more meters within the next three years to do exactly that, to really get a good solid grasp on exactly what every single building is using, every watt that goes through the wall, every BTU that goes through the wall. So it's interesting, because buildings are kind of a living organism. I think most people, if you're not in that business, right, you see the walls, you see the glass, you think it's pretty static, but there's actually a whole lot of stuff going on. And I wonder if you can talk to some of the, you know, obvious inefficiencies and opportunities to make those buildings perform better, performs the right word, and maybe some of the less obvious ones that you've discovered using the Pi software or just other ways that you've discovered opportunities. Calling a building a living entity is actually a really great example. We'll have buildings that will almost completely shut down between the hours of about 10 o'clock at night and maybe about six or seven o'clock in the morning. And then you can actually watch and using the Pi software is phenomenal for this. You can watch the building wake up in the morning and come alive. You can see in the summer, the doors start to open, the internal temperatures start to rise as people come in and out, and the chill water usage go up as that air conditioning starts to kick on. Some of our newer buildings, we have done some predictive analysis on in the building management side of things. So the air conditioning, I should come on about 30 minutes before people start to come into the building and try and pre-cool and get ready for that influx of heat as people start arriving. It helps maintain the overall temperature of the building and you don't get some of those big swings that would then propagate back to the central utility plant. This allows the central utility plant to even out their chillers, maybe bring on a larger chiller a little bit ahead of time and not have to then bring on two or three chillers in order just to deal with that surge of heat coming back in. Just curious, one of the really interesting topics is happening all over right now, right? With the rise of intelligent machines and artificial intelligence, as you know, are the machines going to take over the world but really, consistently we hear, it's really humans making better decisions with data that's provided by the machines and the system. So I wonder if you can share some examples of that where you've been able to take some data, find the pattern without maybe some really crazy big data analytics or running all kinds of crazy analysis but actually relatively straightforward trend lines or anomalies that really pop out of the data once you have the data presented in an easy way to consume. There are actually two scenarios we had on campus within the last year that popped mine real quick. One of them was in a building where we had simultaneous heating and cooling we found. In the same building. In the exact same building. How big was the building? It's actually one of the central buildings on campus. I can't remember the square footage but it was. Like two stories, eight stories. About four stories with a large mechanical subbasement to it. And we found that, we actually found the simultaneous heating and cooling and were then able to track it back and find a three-way valve that was completely broken and allowing both hot and chill water to float into the coils at the exact same time. Just by finding that, fixing the valve, we were able to bring that under control and reduce the wasted energy going into that building. By how much? Or is it back to 10%, 100%? I want to say it was 5% on that one. Okay. One of the larger improvements we made. We had a building that was returning a chill water delta temp of somewhere in the 0.2 to 0.4 degree range. So we were supplying chill water at 42 degrees and getting it back at roughly 42 and a half degrees. Ideally, we're striving for a 12 degree differential to actually pull the heat out of the building and bring it back to the plant. So it should be hotter water coming back. Exactly. To the air conditioning, okay. Once we found this, we realized that the control valve was not working in any way, shape, or form that we were supposed to, it was basically stuck open. Once we were able to identify that, we were able to fix the valve, start controlling the building better. The savings actually necessitated the amount of chill water on a gallon per minute basis going into the building was roughly 1200 gallons a minute, full flow. We dropped that down to close to 150 gallons a minute. That necessitated, almost shutting off a chill water pump at the plant. Estimated savings over the course of a single year, I believe, anywhere from $60,000 to $80,000. Wow, but what's interesting about that story, because the actual valve itself that was broken had no sensor on it, right? It's just a static, whole piece of equipment. But you're able to determine, based on the other data, to track it down. Yep, correct. That's a great story. And it really ties to another factor, which I'm sure you already talked about kind of evening things out and we hear a lot now, right? In the popular media about Tesla batteries, you stick them on the side of your house and now you can kind of manage your consumption off the grid when it's cheaper and put it back on when it's expensive. Because it's not a single price that you pay for that kilowatt, right? It's just a highly variable. So I wonder if you've been able to really take advantage there, too, to avoid some of that peak consumption patterns that's going to cost you a lot more than if you can even it out. Actually utilizing the Pi data in the past was one of the pushes towards MIT creating or revitalizing their cogen system and bringing in an entire new cogen building, increasing their existing electrical output from 25 megs up to theoretical 40 megs in order to reduce how much we are pulling off the grid at any one time. Wow. So what's next? What's next? What are some other opportunities that you see that you can leverage these tools to go find some more inefficiency? One of the things, and actually one of the reasons I'm here at this conference this year is to work on a way to pull in high speed PMU data. And be able to analyze that after an incident happens or as an incident is happening to determine where analytical fault may be occurring, whether it's in our system or whether it's coming off the grid and make determinations as to, do we need to replace equipment? Do we need to go into island mode and do we need to disconnect and just source all the power directly? Are there particular buildings that we need to isolate and figure out why are they performing so badly immediately and could be a detrimental cost to the campus? So it's really interesting because you're finding all kinds of opportunities just to fix things versus I would imagine at some point somebody looked at a number and said this is completely inefficient like your other building, we need to overhaul the whole system. Yes, and we've got an entire system engineering group that is doing exactly that. They are taking data after the fact that they are analyzing it over the last year, two years, 10 years and determining how a building was operating 10 years ago. We may have made full building renovation. How is it operating now? Did we do better? How, if this building is almost equivalent in usage, in size, in location on campus, direction of where the sun is and they renovated this building but they haven't done this one yet, can we expect to see the same energy improvements on this other building or should we do this other building in order to get to the same energy profile? Right, really cool stuff. So Josh, I really appreciate you taking a few minutes to stop and buy. Happen to be here. All right, he's Josh, I'm Jeff. You are watching theCUBE from OSISOF PyWorld 2018 in downtown San Francisco. Thanks for watching.