Truck telematics are transitioning from historical data “analytics” to “business intelligence” platforms that not only gather data from what the truck has done, but also incorporate data outside of the truck—everything from current and future road conditions and construction to predicted traffic and weather patterns—into a predictive and pervasive model that will tell you what your trucks will do. It’s exciting. It’s interesting. It’s also an immensely complicated moving target right now. So let’s put our expectations in check.
“Predictive maintenance, or predictive failure, is the holy grail for everyone,” said Matt Pfaffenbach, director of connectivity for Daimler Trucks North America. “It’s my personal view that some of those things are still science fiction, but everybody is on a path to try to develop that capability.”
“Can you point to an example that is definitely out of the realm of possibility right now?” I asked.
“Let’s take the classic example: I want to predict when an injector is going to fail. That, to me, is still science fiction,” he said. “There are so many factors that are happening within the power train of a vehicle that can give you a false indication if you’re trying to predict the failure of an injector.”
A false indication is the exact opposite of what you would hope to gain by implementing a predictive maintenance solution. That would mean you’re replacing components that are completely fine—increasing downtime and equipment maintenance costs. However: “What is true is that you can make some reasonable projections about a certain mileage range, based on preventative maintenance modeling, in terms of when a service event may occur,” Pfaffenbach said. “Certainly, those projections are becoming more intelligent as we provide operational characteristics about an individual vehicle. Now, are we taking advantage of that in the marketplace yet? The answer is no.”
The then, the now, and…
Last year, TMW Systems announced Software-as-a-Service (SaaS) versions of its TMW Data Warehouse, Data Warehouse Explorer and Visual Analytics solutions as part of its new Reveal Series of BI tools. These tools enable users to blend and analyze data from multiple areas—operations, maintenance, finance, mobile communications and more—to uncover actionable intelligence. Within that offering, TMW is transitioning to taking historical “what happened data” and formatting it in a way that would inform your real-time decisions, which would be much closer to pervasive BI.
What is pervasive BI? I put that same question to Brian Larwig, vice president and general manager of optimization for TMW Systems.
“It’s not, ‘Here’s what happened yesterday and feel free to give yourself an A to an F score based off the data and analysis,’” he said. “It’s asking yourself, ‘What can I do today, based on the information available to me right now and the historical data points? What can I do with my equipment operation—adding equipment, removing equipment, adding a couple loads, removing a couple loads? As a fleet manager, you’re saying, ‘show me, within my own network, what I have available, what I can do and what decisions I can make better.’”
The “pervasive” adjective describes how intelligent truck data services can work their way into the decisions you’re making right now as opposed to looking at the performance of the decisions you made in the past. Once you start incorporating data-based intelligence into your daily decisions, it starts to pervasively work its way into deeper levels of your fleet, beyond equipment usage decisions to equipment maintenance, driver training and retention and logistic services.
The next layer of business intelligence is predictive BI, which incorporates what happened before, what’s happening now and what is likely to happen in the future. Let’s go back to the brake pad example and look how predictive BI could impact your shop’s stock.
“You used X number of brake pads last year,” Larwig said. “Next year, you have the same number of customers and you expect to use the same amount of brake pads. However, depending on your level of sophistication with BI, you could pull in data as crazy as predictive economic impact to determine your projected number of loads or predictive weather impacts for the next year—it’s a La Niña year, which means there’s going to be more wet driving conditions, which could mean more brake usage.”
Incorporating data points and predictions both inside and outside of the truck would potentially allow you to set more accurate brake pad stock expectations.
Predictive modeling incorporates the then, the now and a prediction of tomorrow based on a seemingly limitless data set. However, the goal of advancing business intelligence solutions isn’t to throw more spreadsheets and crunched data analytics at you, but to provide information that will shift your decision making from looking in the rearview mirror to what’s on the road ahead of you.
Jason Krajewski, manager of connectivity insights within Daimler Trucks North America’s telematics department, brought me back down to earth with an example of how Detroit Analytics is evolving to improve the quality of data that fleet managers interact with and how that could impact your decision making.
It starts with the quality of the data that is being collected and the depth at which it can be analyzed. Consider fuel economy: Every fleet in the industry is spec’ing new equipment to improve MPG. While many of the world’s largest fleets have long had the data warehouses necessary to collect and analyze the data to prove the ROI before waiting to see what ends up on the their balance sheets, offerings like Detroit Analytics and TMW Reveal are leveling the data intelligence playing field to allow a fleet of any size to take advantage powerful data.
“The abstraction of data from our vehicles needs to be in a high enough resolution that we are able to clearly explain to the customer why they are not seeing what we think they should in terms of fuel economy,” Krajewski said. “You have to take a little statistical thinking into account when you’re trying to do this kind of thing. For example, we can take a lot of data off of two vehicles. That would be helpful for those two vehicles. Right now, we’d have a hard time extrapolating that specific data out to a fleet or region. Being able to take high-resolution data from enough vehicles in a large enough quantity allows you to make more robust assumptions about what’s going to take place with a fleet.
“Think about it this way: If you had a truck driving over a 30-mile stretch of road and you took data frequently from that truck, you could paint a pretty good picture. If that truck drove that same stretch of road again and performed differently, then you might be able to infer the reason. Now, if you take 5,000 trucks driving over that same stretch of road, with different configurations, different powertrains, different loads—but with enough data at a high resolution from all those vehicles—then you could apply different statistical tools, or filters, and see what’s driving change between different vehicles.”
Do fleets trust the data they capture?
“To me, it’s not the data itself the customers run into trouble with, it’s how it’s interpreted—because the data is what the data is,” Daimler’s Pfaffenbach answered. “It is critical to understand the source of the data and what it is designed to communicate to you. For example, you can measure fuel economy multiple ways from the powertrain, but you need to understand the nuances of it. If the person who is providing the interpretation understands how the intelligence system was designed, then customers find it reliable data information to act upon.
“Our customers are extremely smart,” he continued. “If we were careless with our information reporting, they would quickly call us out on the carpet. So, we are diligent in making sure that we provide them with fact-based information.”
As data-driven telematics solutions evolve toward a more pervasive and predictive business intelligence future, now is the time that to understand your current truck data systems and grow more comfortable with the way the data is collected and analyzed.
“The best way to think of stepping into more intelligence-driven decision-making is to say, ‘At the end of the month, my goal is to incorporate intelligence into 10% more decision-making,’ be it through an analytics platform, algorithmic solve tools or whatever you may have just started implementing. It may be for a small area of your business or your entire business,” TMW’s Larwig said.
“A human can make better decisions than a data-driven system, but you only have so many hours in the day. You only have so many people,” he continued. “You have unlimited amounts of data that you can do whatever you want with. At the end of the day, the data is there for the decision-making person to look at and make the best decisions based off of available intelligence.”
Regardless of how you’ll interact with the business intelligent solutions of the near future, there is one constant throughout. As the fleet manager, you are making the decision. Data and business intelligence solutions are just new tools at your disposal. It’s still your call.