IoT at Alaska’s DoT: To De-ice or Not to De-Ice?
It's still early days for the Internet of Things, much in the way of IoT platforms and infrastructure must evolve before “the connected society” encompassing tens of billions of connected devices comes fully into form. But having said that, IoT systems are already in place handling critical workloads.
In Alaska, at the state’s Department of Transportation, there exists a IoT implementation that DoT managers rely on heavily, one that can already be considered “classical” IoT. Sensors are attached to DoT trucks that travel the roads throughout the Fairbanks area collecting meteorological and road condition data; the data is then sent to the cloud (Microsoft Azure) for analysis and decision support regarding whether to de-ice the roads as a snowstorm approaches or is taking place.
It’s “classical” in that this is just the kind of role IoT is meant to fulfill. Data from the truck sensors comes in multiple formats, and it is wrangled, analyzed and visualized on dashboards in real time; further, according to the customer, the implementation was surprisingly fast, inexpensive and relatively straightforward, reflecting characteristics of the solution, from Boulder, CO-based startup Fathym Inc., that may point to the interoperable IoT infrastructure of the future.
For the DoT, deciding whether to de-ice is non-trivial.
“There are two parts to this: one is sending plows out when you don’t need them, and that’s a major expense,” said Billy Connor, director at the University of Alaska Fairbanks, which is consulting on the Fairbanks project, and former chief of research at the Alaska DoT. “But not sending the plows out when you need them is a major risk for the public. So what they’re trying to do, within the budget they have, is spending the money they have more effectively. They’re trying to optimize the benefits to the public. That means we send the plows when they’re needed and where they’re needed, and apply the anti-ice chemicals when it’s needed and where it’s needed, and making good decisions is critical to being able to do that.”
De-icing decision making has historically been as much informed intuition as a science. The terrain in and around the city makes it doubly difficult because downtown Fairbanks lies in a valley surrounded by hills and mountains. While the general Fairbanks area receives an average of 70 inches of snow per year, snowfall amounts can be highly variable based on localized “microclimates”: it might be snowing in the hills while the downtown area gets none.
“The area superintendent would look at the weather forecast from NOAA (U.S. National Oceanographic and Atmospheric Administration), and make his best guess based on his experience,” Connor said. “Then they’d go out and drive the roads during the storm and try to assess where priorities might be. It was more or less seat of the pants and experience.”
DoT needed a way to verify the conditions in the various areas around Fairbanks. The concept of mounting sensors on DoT trucks is not new, but according to Connor it has not always been successful. He said the DoT initially worked with another vendor several years ago, but he said their sensors, which cost between $3,000 and $4,000, and the data they generated didn’t integrate well with DoT’s Maintenance Decision Support System (MDSS).
“We weren’t really happy with those sensors,” Connor said. ”And we had a third party involved to take the data from the other vendor's system, translate that and try to get it into the MDSS system. The data transfer never really was what we would consider successful. It was not reliable. Sometimes it was there, other times it wasn’t.”
Then Fathym came on the scene. Cost was a major difference: Fathym sensors cost about 10 percent of the previous vendor’s. In addition, Connor said, data transfer to MDSS was seamless. He also said customization of DoT dashboards, which visualize the data collected from the 40 DoT trucks installed with sensors in the Fairbanks area, was relatively simple. With the success of the Fairbanks pilot project, he said, other cities and town in Alaska also are planning rollouts of Fathym systems.
According to Fathym, a prototype IoT system can be up and running in a matter of a week or two.
“What we have is an end-to-end solution using sensors that we build capturing that data, transmitting it into the cloud and into a data lake,” said Doug Harless, SVP of business development. “Then we take our tools, apply them to the data, make that data understandable.”
In constructing dashboards, rules can be set up that alert users as the customer sees fit – such as when there’s a system or equipment malfunction, a temperature change, system overload and so forth.
“We can take data off any type of device-generated data stream or even a compute data stream, bring it into Fathym, consolidate it into a single data set and then turn that data set into something that’s actionable,” Harless said.
The advantage, he said, is in the data consolidation, a “single pane of glass,” which gives users a comprehensive real-time view from multiple data feeds, “so an operations group can take action on a correlation of that data rather than looking at four or five different systems.”
According to Glen Allmendinger, president of consulting firm Harbor Research, Fathym is a “disruptive startup" in the IoT space for its ability to “leverage unique graph databases and easily developed applications (that enable) developers to quickly build tools that allow for interoperability.’
Fathym, Allmendinger said, “provides the building blocks for customers to develop custom solutions in a rapid prototyping fashion while also providing a scalable platform that is agnostic to data types due to the relationship of its application development tools and the graph database structures the company has created.”
The result is “true interoperability” encompassing existing hardware, applications, data management, analytics tools, or infrastructure/cloud, an ecosystem “that allows for the rapid, affordable creation of end-to-end IoT solutions, from hardware to application.”
With the Alaska DoT implementation, each of the 40 trucks on the roads around Fairbanks collect myriad environmental data – air temperature, barometric pressure, humidity, road pavement temperature, as well as a sensor on the windshield wipers (if they’re swiping it’s snowing). With the data flowing into the MDSS and displayed visually at DoT headquarters, managers can gauge (at either an overall picture or down to the individual truck level) weather and road conditions; they can assess the various microclimates in the Fairbanks region and decide whether, and where, to de-ice.
Connor said data visualization on dashboard is the key to the IoT system.
“We don’t want to see the data,” he said, “we just want usable information.”