If you operate a fleet, you probably pay attention to fuel costs. In fact, reducing fuel is one of the most common motivations behind installing a telematics system: the technology helps you keep an eye on speeding and idling, and it can highlight vehicles with unexpectedly low fuel efficiency.
In the last few years, our Catalytix team has taken this one step further. Many of our customers have asked for help to put MPG figures in context. After all, we track over 2 billion miles driven each year. From these journeys, we have real-world MPG data on hundreds of vehicle makes and models, since we can read actual fuel consumption data via our patented CAN bus technology.
A working example of telematics data usage
Using this data, let’s say our Catalytix team has told fleet manager Paul that six of his vans are performing well below benchmarks. Five of these have real issues that Paul can see. The drivers are idling too much and accelerating too harshly. But the sixth driver seems to be doing all the right things.
This is where a simple benchmark doesn’t tell the full story. The sixth vehicle typically makes a multitude of short, slow trips around Sheffield, so it’s not too surprising that its MPG falls below our benchmark.
Here’s a figure that shows a few examples of how journey type affects MPG. Each clickable diamond represents a vehicle, and Paul’s sixth vehicle is shown in green. You can see a typical day on the right-hand side.
While a simple benchmark is helpful, sometimes it’s necessary to dig a little deeper. So, how do we make sense of all this information, in order to provide more meaningful benchmarks for our customers?
I started with three months of journey data for commercial vans. This included over 20 million journeys. My first task was to identify all of the different roles these vehicles have.
Let’s look at two examples. The first is Paul’s van making short and slow trips around Sheffield.
From this chart, you can see that the vehicle usually does a number of short journeys each day. You can also see, from the darker colours, that his average speeds are pretty low. We call this an “urban delivery” van.
Here’s another example. This van (marked in blue on the first figure) takes multiple high-speed, long-distance trips each day. We call this a “long-distance travel” van.
To compare millions of journeys, I made use of unsupervised machine learning algorithms to automatically classify different vehicle roles using 22 variables, including journey speed and distance. Using this type of analysis, I identified seven different clusters — including the “urban delivery” and “long-distance travel” roles, and, unsurprisingly, the same LCV model performs differently in each role. The two vehicles we mentioned are very typical in their own groups — even though they’re on opposite sides of the global average.
So we’ve achieved our initial goal: we can now give fleet operators more accurate benchmarks, depending on each vehicle’s job type. This will help operators focus on vehicles that can stand to improve their fuel efficiency.
Digging deeper still, this heat map shows fuel efficiency for different LCV models depending on job type. By clicking on the cells on the sample below, you will notice how, in the first column, LCV model #4 has the highest efficiency, but the same model performs the worst in the second column.
What goes into a vehicle purchase or lease decision? While MPG isn’t the only factor, it’s an important one. And average MPG doesn’t tell the whole story – especially if it comes from the manufacturer, instead of real-world data.
With Masternaut, you can get real-world MPG information for the vehicles you’re considering – specifically for each job type. We estimate that choosing the right vehicle using these improved benchmarks can reduce fuel usage by 5%. That can be worth £15,000 each year for a fleet of 100 vans. And that’s on top of the fuel savings you already get from using a telematics system.
We’re just getting started with data science here at Masternaut, and we know there’s a lot of business value in the hundreds of billions of data points we collect each year. For example, we can use this information to recognise our most efficient customers with CO2 Certification.