Babel Buster Network Gateways: Big Features. Small Price.
EMAIL INTERVIEW – Dr. Igor Mezic, John Morris and Ken Sinclair
Dr. Igor Mezic,
Chief Scientific and Technical Advisor, Co-Founder
John D. Morris,
Vice President, Marketing and Sales
Sinclair: How do you describe Ecorithm?
John: To hit all the
buzzwords off the bat, we’ve developed Ecorithm’s True Analytics™
software platform for the automated analysis of complex, dynamic
systems, with a particular emphasis on building HVAC and Internet of
Things (IoT) applications. Ecorithm’s software as a service (SaaS)
offering is differentiated by its highly accurate automated fault
detection and predictive analysis – a product of our patented
Sinclair: Igor, you’re a professor of mechanical engineering and mathematics at UC Santa Barbara. Can you tell us a little about your background and how it led to the development of Ecorithm’s technology?
Igor: I grew up in
Croatia, where my father was a Captain of the massive French CMA-CGM
cargo ships that are essentially floating cities (I sailed with him on
a bunch of those trips when I was in my teens) and my father-in-law
owned an HVAC business in which he designed complex HVAC systems for
resorts and industry. I helped my father-in-law a bit when I was still
an undergrad and he is one of those building whisperers that we
referred to in the December article.
So, I have a great appreciation for how really skilled, experienced
people develop an intuitive feel for diagnosing mechanical systems and
the key data that they need in order to make quick decisions. That
probably sparked my interest in figuring out a way to practically apply
artificial intelligence to very complex dynamical systems.
I first came to the US to do my graduate
work in mechanical engineering at CalTech and then did my postdoctoral
research in mathematics at the University of Warwick in England. When I
returned to the US, I joined the faculty at University of California at
Santa Barbara, which has an outstanding engineering department and a
community with an exceptional commitment to energy efficiency. To that
end, I was a founding member of the Institute of Energy Efficiency at
UCSB, which has been very successful. In fact, one of our faculty,
Professor Shuji Nakamura, won the Nobel Prize last year for his work on
LED. It is a great environment for developing new technologies.
Sinclair: John, you’ve been around the building and controls industry for a couple of decades now. What makes Ecorithm a game changer in your view?
John: For 20 plus years
I’ve been involved in innovations in building controls with a focus on
networking devices – effectively making buildings a prototype for the
Internet of Things. Once all devices in the building were networked,
‘big data’ became a real possibility for the first time. When I was
first introduced to Ecorithm’s technology, it was clear they were in
uncharted waters as far as making sense of the data building systems
were capable of delivering.
Sinclair: What are the big problems that Ecorithm is solving?
Igor: From a technical standpoint, I’d say the combination of scalability and specificity of our True Analytics™ platform. Building analysis is tough sledding. Historically, retrocommissioning would take months of study by a team of engineers with specialized knowledge. Every building is unique in terms of its equipment, operation, environment, and usage. Plus, they are very dynamic – everything is changing all the time, and each change can have a ripple effect. The question of how to automatically and accurately identify existing issues in a large building (as well as opportunities for predictive/preventive maintenance), while at the same time balancing the needs of the demand side (occupant comfort) with the supply side (energy efficiency), is hard enough. When you think about extending that across a portfolio, and ultimately billions of square feet, and that many of these buildings will have patchy or incomplete data, it’s a seemingly intractable problem. That makes it really tough to design an intelligent platform that is specific enough to handle granularity down to the individual device level, while also scalable enough to handle the diversity of equipment and building configurations.
To come up with the solution, we had to think at both the micro and macro level and design the software platform to accommodate everything in between. We came up with a very modular design in which the underlying foundation filters through the noise of the massive data sets and recognizes key patterns. On top of that is a layer of domain expertise that includes the physics of how the ‘healthy’ devices and systems are supposed to operate. And resting on top of that is an interface to quickly tailor the spatial and physical connectivity of devices in the virtual database to match the configuration and operation of each physical building. That means exceptionally quick start up and customization, highly detailed insight and root cause analysis, and easy integration of new devices or changes in configuration. Also, this makes the platform readily extendable beyond buildings to other complex systems as well.
Sinclair: How does that translate to customers?
John: This pays off for the customers in several ways. Obviously, detailed insight into a building provides energy savings opportunities. We typically see the payback of Ecorithm’s SaaS to be well under a year, which is excellent. This is where the conversation with new customers always starts, but after a few months our customers quickly recognize the benefits of maintenance savings, tracking the effects of measure implementation, a searchable archive of equipment behavior, optimization of occupant comfort, and a comparison of multiple buildings across a portfolio. Over time and with scale, we’ll be able to quantify the financial benefits in each of these categories. We expect that customers will be able to utilize the outputs from the platform to demonstrate a solid financial (and employee productivity) benefit to new tenants as well as existing tenants up for renewal.
We also capture the unstructured data, e.g., the notes written on a clipboard next to the boiler, and put it into context in a historical record. Extension to the portfolio level provides the opportunity to see the definitive impact of specific control sequences, equipment, and architecture on building comfort and energy use.
Sinclair: You’ve branded the software platform “True Analytics”. Why?
John: Analytics is a broad term, and, as Igor alluded to earlier, building analysis is a tall order for software packages. When we looked at the building analytics market, we were seeing a lot of noise created by products that are either plagued by false positives or really aren’t analytics at all– perhaps just having some rudimentary features, such as comparing energy usage to a historical average, or providing a dashboard to look at basic metrics. Ecorithm’s platform provides the deepest, most comprehensive analysis of a building system while virtually eliminating false positives. One of our customers conducted a trial in an Energy Star 98 building by independently vetting each of the 100 issues identified over a six week period and found Ecorithm to be 100% accurate. If that’s not “True Analytics”, it’s hard to imagine what is.
Sinclair: Are there additional pain points for customers that Ecorithm solves?
John: Many of our customers have tried other analytics and expressed pain with long startup times (as well as the false positives and alarm fatigue that we mentioned earlier). Even for data ready buildings, customers reported that it took weeks to months of custom programming to specifically configure rules for each building. This is a huge time sink and large cost. And the pain is renewed if there are any equipment or operating changes in the building.
For data ready buildings, we measure our startup time in hours or days even for very large complex buildings. And, because of the modular platform, we can very easily accommodate new equipment or changes in operation. In fact, the platform is getting smarter with every building and device that we encounter. For example, if the software uncovers a fault that we haven’t seen before, we can easily incorporate that intelligence directly into the platform so that all of our existing customers will benefit as well, without the need to re-program for each building.
Sinclair: You mentioned that the software is getting smarter and that the intelligence is built in. Is this machine learning?
Igor: Yes, but even more than that. It really fits within an Artificial Intelligence and Dynamic Machine Learning context. Whereas machine learning usually entails learning of static relationships between driving forces (inputs) and system response (outputs), the basic physics of a commercial building is driven by the dynamic forces of environment (weather) and occupancy, which give the physics a time dimension. Fortunately, the driving and dynamic forces, and the resulting responses, tend to follow certain cyclical patterns. Our software is uniquely suited to identify abnormal patterns that might otherwise be obscured by typical fluctuations in operations. When an unhealthy state is detected, the True Analytics Platform has the ability to identify which part of the operations caused it – this is the benefit of the system layer with a learned understanding of which root cause produces a certain pattern. This is in stark contrast to the rules based offerings in which outputs are completely static until a highly skilled (and expensive) building analyst discovers the issue, and a highly skilled (and expensive) programmer manually changes or adds a set of rules to generate an alarm. However, with our machine learning capabilities, the True Analytics platform automatically (and, therefore, inexpensively) gets continuously better and better over time.
Sinclair: Are there limitations? You also mentioned data ready buildings – how much and what kind of data do you need?
Igor: Well, we certainly need data – and that data needs to be operationally relevant in order for the software to deliver meaningful results. It is also important that the building have enough system control to be able to act on, and realize value from, our findings.
We’ve initially focused on large commercial buildings for a few reasons. Larger buildings are more likely to have building management systems (BMS) with sufficient sophistication to facilitate data acquisition and provide enough elements of control to be able to tune the building. Also, more complexity means more opportunities for optimization and savings. Though large commercial buildings account for only a small percentage of the total number of buildings in the US, they represent the majority of the energy use, and therefore represent a huge opportunity for us to make a significant impact.
In a typical building, we trend between 1,000 to 15,000 data points on 5 or 15 minute intervals. We’re collecting the data points that are most operationally relevant: temperatures, flow rates, set points, damper position, occupancy, etc., depending on the devices within the building. However, we can also convert small data sets into powerful results. From a technical standpoint, we can create significant value for virtually any relatively modern mechanical system (think IoT).
Sinclair: It looks like a very crowded space. How does Ecorithm differentiate itself?
fashionable to attach ‘analytics’ or Internet of Things or IoT to every
offering, so it’s important to take the term with a grain of salt until
you understand what the solution really does. From our perspective, we
see most of the companies in the energy software space falling into one
of three categories: Dashboards, which simply organize energy data,
without any real interpretation; Controls Override, which focus on
optimizing the efficiency of a single device (like a chiller) without
accounting for the system wide impact; and Trouble Ticketers, which
tend to be rules-based software that require extensive custom
programming and skilled interpretation to sort through the alarms.
Comparatively, Ecorithm’s True Analytics™ platform provides a much deeper, comprehensive analysis of root causes of unhealthy behavior across the entire building, without the need for custom programming.
Sinclair: Is this in real-time? How long do you have to wait to collect enough data points to perform a useful analysis, implement solutions, and see results?
John: We can certainly take real-time data and ultimately we expect that the market will move toward intelligent real time control. However, at this point, the term real-time tends to be a marketing buzzword that sounds great but, in practice, doesn’t provide what the customer wants or needs. Most customers already have real-time alarms for critical systems set up through their building management system and are dealing with alarm fatigue. They need a system to cut through the noise of those alarms to prioritize the issues that are persistent and important.
We’ve found that customers often work through newly identified issues with their in-house engineers or through their maintenance contracts on a weekly or monthly basis. They also want to quickly see the overall impact of an implemented change on the whole system. We’ve found that daily updates strike this balance very well.
Sinclair: Is this meant to be a short term solution, or a service that is used for the life of the building?
Igor: Again, buildings aren’t static. Their performance is always changing with the effect of weather, seasons, and equipment wear. Depending on the building, the biggest gains may arise from fault detection at the initial stages. In others, the biggest gains may be from ongoing optimization. But in every case, ongoing analysis is critically important to maintain savings and find new opportunities. Lawrence Berkeley National Labs provided some great information on how quickly the gains evaporate without monitoring. [Source: Building commissioning study conducted by the Lawrence Berkeley National Laboratory for the California Energy Commission (http://cx.lbl.gov/2009-assessment.html.
One real world example: a building that had big energy savings potential due to outdated pneumatic controls and sequences of operations had decreased its HVAC energy consumption by 30%. And then, effectively overnight, the energy consumption jumped back to the pre-analysis usage. A reboot of the building management system had simply reset the system to default and the unintended consequence was to erase all of the accumulated changes and efficiency gains. In this case, the building “returned to the mean” in a single night, but this type of behavior will happen over time in every building in the absence of ongoing monitoring and tuning. The customers can of course decide to discontinue the service, but in most cases that would be against advice given by ROI calculations.
Sinclair: Lastly, what does the future hold for Ecorithm?
Igor: 2016 is shaping up to be a great year for us. Our team is phenomenally innovative and we’re really just scratching the surface. We’re excited about rolling out to larger portfolios that will encompass a variety of different use cases: large and small, office and manufacturing. We’re incorporating predictive analytics into the platform, as well as new features that will allow us to derive substantially greater value from utility data. We’re also working with partners to expand the software library for a variety of IoT applications. We envision that, as analytics are adopted more widely for IoT applications, customers will want to consolidate analytics into a single platform to understand their buildings and operations as a whole. That is exactly what True Analytics is designed for.
[Click Banner To Learn More]
[Home Page] [The Automator] [About] [Subscribe ] [Contact Us]