Detecting Energy Theft With Situational Intelligence


In 2010, BC Hydro estimated that energy theft cost the utility up to CAD $100 million annually in lost revenues. That’s enough power to supply 77,000 homes for a year.

As described in this EY article, much of the theft supported illegal marijuana growing operations hidden in residential buildings across the BC Hydro service territory, an area the size of the UK and France combined.

At the time, BC Hydro had just finished installing smart meters for all of their 1.9 million customers. Smart meters meant no more human meter readers walking routes and reporting suspicious activity. But smart meters also meant a valuable source of big data for detecting and prosecuting energy theft.

BC Hydro also installed smart meters along feeders and segments on the power grid, upstream from consumption meters on houses and businesses. By using situational intelligence applications that combined spatial, temporal and network analytics, BC Hydro began to detect which locations had abnormally low energy consumption, which could indicate power theft.

In particular, BC Hydro could compare energy delivered down a feeder with energy billed from meters along that feeder. If those two numbers weren’t equal, then energy was being lost and potentially stolen. Other factors such as weather, season, time of day, size and type of building, and so on enhanced BC Hydro’s ability to identify likely instances of energy theft.

Advanced visualizations helped revenue analysts and field investigators prioritize, locate, and research suspected theft cases. If a case warranted legal prosecution, the data and visualizations provided compelling evidence at trial.

Today, theft has been reduced by 75 percent. Plus, customers are safer, with fewer dangerous hacks into the power grid. BC Hydro continues to use advanced analytics and visualization to reclaim revenues lost to theft as well as technical and non-technical loss.


Waterway Management and Situational Intelligence



When I say “water transportation,” your mind might conjure romantic images of the canals of Venice, Italy, with gondolas and vaporettos ferrying locals and tourists down the Grand Canal. Or maybe you’re the cruise ship type.

But did you know that

Commuters, tourists, trade and cargo move along these and many other waterways around the world, which are vital parts of their surrounding economies.

Like their terrestrial highway cousins, waterways are linear assets. Linear assets pose different challenges from fixed assets, such as power transformers, and mobile assets, such as ships. Highways and waterways share a similar set of concerns about quality, environmental impact, accommodating mixed use, seasonal weather and traffic patterns, and more.

Instead of parts like a machine, linear assets have segments. The size and numbering of segments can vary widely. Characteristics of the asset can change over distance, such as water depth and rate of flow for a waterway. Different characteristics can change at different points in and across different segments.

Plus, there are relationships with parallel linear assets, such as power lines that supply energy to the fixed assets along a waterway. There’s constant interplay between the fixed, mobile and linear assets based on time of year, time of day, weather and other factors. All this change and variability means that analyzing linear assets quick gets complicated.

Situational intelligence applications, with the combination of spatial and network analytics and visualizations, work well for monitoring, managing and optimizing linear assets such as waterways. They offer the breadth and flexibility of data integration, analytics and visualization to handle the complexity of waterways and the myriad other related assets—fixed, mobile and linear. Users can analyze and visualize characteristics in real time across dynamic segments.

I apologize if I’ve ruined your romantic conceptions about water transportation.

(Image “Venice grand canal rialto bridge 2012” by chensiyuan. Licensed under GFDL via Wikimedia Commons)


Can Connected Cars Learn to Increase Their Operational Efficiency?


You become more adept at driving familiar routes over time. You learn the best times to leave to avoid traffic, which highway lanes move faster, and where large puddles form after heavy rainstorms. Could your connected car also become more adept over time?

Using the spatial, temporal and network analysis of situational intelligence, a connected car could learn your normal commuting routes and times, correlate those with weather records and your fuel efficiency performance, and offer suggestions for how to drive the route more fuel efficiently. For connected hybrid cars, such data could be transformed into optimizing when to draw on battery charge versus gasoline for more fuel efficiency. Similar analytics might make alternate suggestions if you prefer to make the trip as quickly as safety allows, regardless of fuel efficiency.

Connected cars could perform their own version of air conditioning demand response. On hot days, cars could analyze the forecasted outside temperature along your route, or draw on outside temperature readings from other connected cars ahead of you. Those other cars could also share their operating performance related to specific stretches of road. Your car would then combine this with your interior comfort preferences and your current fuel supply and usage.

Based on all this information, your car could control your air conditioner in much the same way your utility controls air conditioning during demand response event, pre-cooling the cabin on flat stretches of road then throttling back cooling while driving up hills. This would mean better fuel efficiency for you without sacrificing comfort while you drive.

Demand response is just one way that connected cars could learn from one another to increase operational efficiency. Crowd-sourced predictive maintenance might be another way.

Suppose that past data shared between connected cars shows that drivers in your city with your make and model of car tend to need their water pump replaced after 57,000 miles of in-city driving, regardless of highway driving miles. Based on this data, your car could warn you of a potential pump failure before it happens.

Your car could even, with your permission, share this water pump information with others. Your mechanic might receive the update and order the part ahead of your next appointment, saving you time at the shop and preventing a future breakdown while out on the road. The car manufacturer could use this water pump information to improve product design and reliability and inform dealers of potential issues. Makers of after-market car parts might pay to receive this sort of reliability information to help them design and market products.

Connection, analytics and machine learning are still on the horizon for the average car, but that horizon is fast approaching. I think we’ll find great new conveniences and operating efficiencies as that horizon draws near.


Is 2D Dead?


Does the rise of immersive, 3D visualization through gaming technology, virtual reality, and augmented reality mean that 2D displays are dead? After all, why click and scroll through a 2D map when you can virtually visit a 3D landscape?

Which view of the Brandenburg gate do you find more engaging?
Which view of the Brandenburg gate do you find more engaging?

3D does have benefits over 2D. 3D depicts the height, volume, and contour of objects to give viewers a sense of spatial relationships between objects and a more nuanced understanding of the texture of objects than 2D can provide.

But 3D also presents challenges to displaying, understanding and manipulating data. When you add the Z-dimension, you increase the amount of spatial data available about an object – more data, more challenges to determining how and what data to present, and how to avoid overwhelming a person’s ability to take in, and make use of, more data points.

Also, when an object has volume, it takes up more visual space making it more likely that nearby objects get obscured. And alas, we just haven’t gotten that good yet at displaying 3D information in ways that are smooth and feel natural to manipulate. The HoloDeck from “Star Trek” is a nice start, but we’re not there yet.

There’s also the challenge of managing level of detail versus height of viewing. When looking down on a city block from a simulated elevation of 100 feet, you can take in details about individual items. But if you move to a simulated viewing elevation of 10,000 feet, all those details become too much to display both from a comprehension point of view and from a software performance point of view.

Decisions need to be made about what information is important at 10,000 feet vs. 100 feet. Those decisions depend on who’s doing the looking. Making software that intelligently handles these changes in viewing elevation requires forethought and deep understanding of the user’s tasks.

Designers can compromise using 2.5D displays. These visualizations add perspective to a 2D display, but don’t go all the way towards three true dimensions.

Synchrophasor video 01

You get the sense of height perhaps, but not of volume. This can be useful in making items stand out more against a 2D surface, plus it looks a little cooler and affords the ability to rotate displays in space to understand better the spatial relationship between objects.

As with many aspects of visualization, the choice between 2D, 2.5D, and 3D display comes down to design thinking. What task are users trying to perform, and what’s the best way to accommodate and accomplish that task? It all depends on your user’s point of view.

(Image courtesy of Flikr under Creative Commons License.)