It is early spring in 2018. Most major fleets have become such sophisticated users of data to do prognostics or predictive maintenance these days that they don’t even use the predictive word anymore. It is just part and parcel of routine maintenance—business as usual.
Prognostics, of course, had been talked about for more than 20 years before it became a reality, but there was never the data to actually make it work. Now trucks all roll off the assembly line connected to the cloud from the first time the engine starts until the last. And the industry is awash in data, quality data with the analytical horsepower to spin it into informational gold.
Or maybe not….
Fleet Owner had the opportunity to individually interview numerous maintenance, big-data and analytics experts on the future of predictive maintenance and where it stands today, in early 2015. Their perspectives and insights are collected here in a virtual roundtable discussion. The panel included:
- Renaldo Adler, Vice president of development, asset maintenance systems group for TMW Systems
- Fritz Ahadi, General manager, commercial and government fleet sales, Ford Motor Co.
- Mark Alsbrook, Senior product manager, Omnitracs
- Mike Cerilli, Vice president and general manager, Connected Vehicle, Navistar International Corp.
- Bill Cooper, Vice president of customer acquisition, WEX Fleet One
- Dean Croke, Vice president, Omnitracs Analytics, a business unit of Omnitracs
- Conal Deedy, Director of connected vehicle services, Volvo Trucks
- Greg Dziewit, Vice president of commercial OEM, Telogis
- Thomas Fansler, President, Vusion
- Bill Frykman, Manager, product and business development, connected services, Ford Motor Co.
- Rich Glasmann, Vice president-OEM strategy, sales and marketing, Omnitracs
- Matthew Pfaffenbach, Director of telematics, Daimler Trucks North America
- Michael Riemer, Vice president of products and channel marketing, Decisiv Inc.
- Dave Reed, Fleet management consultant, Dossier Systems Inc.
- Mark Wallin, Vice president of product management, Telogis
- Mike McQuade, Chief technology officer, Zonar Systems
Q. Vehicle connectivity is seen as essential to predictive maintenance. Where is the industry today, and where is it headed when it comes to connectivity?
McQuade: The future of every vehicle is connectivity from day one. If you really want to blow open innovation for fleets, this is what it will take—and it is inevitable. There are already something like 8 to 10 computers on a heavy-duty truck. In fact, I think of trucks as “diesel-powered computers.” You and I would not buy a computer that couldn’t be connected and the same thing will be true for trucks. In a matter of months, I believe, people won’t buy a vehicle that is not connected to the cloud.
Twenty-four months from now, there will be more connected-to-the-cloud vehicles than there are today. It is a very steep growth curve. Daimler Trucks, our partner with Virtual Technician, already has more than 100,000 vehicles connected to the cloud.
And every car company on the planet is trying to copy Tesla when it comes to connectivity. Every Tesla is connected to the cloud when it rolls out of the factory; customers have no choice. If a car needs a software update or there are system enhancements, they are pushed out to the car over the air.
At Zonar, we bucked the current trend and decided to deliver all the data to the cloud. If there is too much data, you can always add filters and filter out what you don’t want to see. Now we have several years of data available and are beginning to analyze and predict how long parts will last.
Cerilli: The connectivity space is growing considerably. In our industry, it is already a $10.5 billion business and it will grow to north of $17 billion by the end of 2018. It is a big growth category in the trucking industry. The largest mega fleets have been the early adopters, but smaller fleets are coming onboard as costs drop. Connectivity is also moving out to buses and medium-duty vehicles.
Fansler: Today’s trucks are really computers on wheels already and connectivity is certainly the future. That does not mean, however, that all data has to be passed to the cloud. An engine generates about 3 gigabytes of data per month, or about 5,000 data samples per second, so it is not even practical to think about passing all data to the cloud.
This is not just a problem, but also an opportunity. With onboard analytics and a very high sample rate, for example, we could coach a driver on shifting just as he or she requires the information. You could not do this without missing the window of immediate opportunity if you had to send the information about the driver’s shifting from the vehicle to the cloud for analysis and then back again.
Q. There are lots of stakeholders when it comes to connectivity and remote diagnostics. How are the roles and relationships of OEMs, Tier 1 suppliers and telematics providers changing as remote diagnostics development progresses?
Glasmann: Coming from the OEM world, I think it is interesting to see how OEMs are using telematics to develop their maintenance and diagnostic service products, working with telematics suppliers and other partners. Everyone is going to have to start working together, and the reins will be loosened concerning access to currently proprietary data. It is just a matter of time.
OEMs will also be able to use the data [gathered to better manage maintenance] to help them design and build trucks. As we help the OEMs build databases, they can share that information with their Tier 1 suppliers.
Fansler: The modeling of maintenance is very complex. The next step is to have this information shared. In other words, everybody is now becoming more dependent on one another. Think about the idea of an ecosystem comprised of the telematics/analytics vendors, dealers, OEMs and the customers. This is a macro view of where we are headed.
Vehicle OEMs are facing a maintenance polarity. They have a warranty obligation they must cover, so if there is a fault code for a critical event, they want to address it right away to avoid cascading events and resultant costs. The reaction is to get trucks in for maintenance as soon as possible.
On the flip side, fleets really hate to take trucks off the road. So they are thinking, maybe it was a false positive indicator and just a sensor problem. [We can keep running].
That is a big area of work where we will see a lot going on. [To solve the problem, we have to have embedded analytics right from the start.] The challenge is to organize data so that there is real context for any issue. The right analytic tools have to be embedded in the right departments to turn up the information that is needed there.
Engineering needs to be able to predict and prevent failures. Marketing needs to optimize the customer experience to improve customer satisfaction. Maintenance and warranty needs to deal with a maintenance repair issue. There is a huge need for everyone to speak a common maintenance language.
Riemer: The OEMs will either develop their own telematics hardware or partner with third-party providers to capture data from their trucks. Based on current renewal cycles, I would estimate that within 10 years, all heavy-duty trucks will at least offer the option of OEM-provided hardware. When it comes to maintenance, I believe OEMs will become very proprietary in how they analyze and use data from their trucks, which will go well beyond the basic ECM data that most people use today. I don’t see any reason not to believe OEMs and their dealers and select partners would want to have control over this information.
The challenge for fleets with multiple brands will be how to get access to the [proprietary] information they require in order to run their businesses without going to multiple, disconnected portals (a solution that Decisiv actually provides today).
Wallin: More and more, OEMs are providing connected vehicles with information built-in, and every OEM is working to provide better data and maintenance information to improve uptime. Differentiation opportunities exist in the marketplace based on the amount and quality of the data an OEM can provide, the level of data granularity and having richer analytical capabilities.
As OEMs capture more and more data, they need a platform and advanced analytics to turn their data into something meaningful, to make it really useful. You can’t see patterns and context so that you can start truly predicting parts failures, for example, if you can’t look at enough good data and do the required analysis.
Dziewit: Ford Motor Co., in partnership with Telogis, introduced a ‘next-generation’ connected vehicle solution earlier this month called Ford Telematics powered by Telogis. It replaces and rebrands Ford’s Crew Chief powered by Telogis connectivity solution. The key benefit for customers is ease. You can spec Ford Telematics so that the vehicle comes right off the line with all the technology on the truck. Because it is a Ford offering, it can be financed and capitalized with the vehicle and is covered by Ford’s vehicle warranty.
Then the customer just has to go to the Ford Telematics website, type in the VIN (vehicle identification number) and give the vehicle a name. The vehicle then starts ‘talking’ to the cloud.
Ford and Telogis also made the decision to make Ford Telematics available for retrofit on any vehicle. All of our joint customers have existing vehicles on the ground. They can always retrofit a device on their other Ford vehicles; we made a retrofit kit for Ford. They can also add it to any other vehicles of any other make in their fleet. It can even be installed on non-vehicles, such as heavy iron or containers.
Q. What will be the role of big data in maintenance?
Riemer: It is all about data, data, data. Not just any data, but the ability to consistently capture the right data without human intervention. This will be the foundation for a fleet’s competitive advantage going forward. This is true across all fleet operations, but most importantly for maintenance.
I believe that fleets significantly underestimate the impact of their maintenance operations on bottom line profitability now and in the future, but mostly they lack the data to show this and make better decisions… big data is a buzzword that gets a lot of attention, but fleets should be really focused on getting the right data.
Pfaffenbach: Big data for me is too grandiose a concept. People need information. They are looking to OEMs and technology suppliers to connect different, disparate bits of data from the truck and its operating environment that they can use.
For example, last year we had a polar vortex which brought extreme cold to some parts of the U.S. At DTNA, we experienced some fault codes concerning exhaust aftertreatment that could have caused a misdiagnosis of the problem. However, looking at the fault codes in the context of the operating conditions for the engine told us we had a software issue to address instead. So we were able to update the software [and keep trucks on the road].
We also use big data to help us look for trends in warranty failure.
Q. Who owns vehicle data? And what security and privacy issues lie ahead?
Pfaffenbach: The fleets own their data. We respect and guard that on their behalf and ask them for access to data in order to provide services to them. Data security is a big issue.
Glasmann: There is a whole ecosystem of people who need/want the data, which belongs to the customer, the fleet. They will have to decide who will get access to their data.
Ahadi: We do not have access to specific customer data; it belongs to the customer, even though the system is Ford’s. Some customers let us see their data, though, which gives us the opportunity to coach, etc.
Cerilli: The data is really the customers’ data. We ask to use it to help us do a better job. When dealers have transparency to the data, it can help them manage uptime. The real power comes when all the dots are connected. Most customers, in our experience, are very willing and eager to share data if they see it as a value-add for their business.
Riemer: There is a great argument to be made that during the warranty period, the data is owned (at least the maintenance and related data) by both the fleet and the OEM.
Q. If/when this predictive maintenance vision comes to pass, how will it impact maintenance shops, technicians and even the driver?
Deedy: Analytics eventually will allow truck owners to move from standard preventive maintenance schedules based simply on miles or hours to schedules tailored specifically to their trucks’ duty cycles. Diagnostics will become faster and easier as analytics mature, allowing technicians to focus more of their time on repairs and less on identifying the problem.
In fact, this is already happening with Volvo’s Remote Diagnostics, which identifies what parts will be needed and provides repair instructions even before the truck arrives for service. Remote Diagnostics, on average, reduces the diagnostic time by more than 70% and lowers actual repair time by more than 20%.
Croke: Technicians used to be able to diagnose problems just by ear, by the sound sometimes. That is gone. Now it is all about data and analytics.
For the driver who is not a mechanic, and that is the vast majority now with electronic engines, it has become almost past the power of the human brain to [accurately identify problems and the required remedies]. Say the doser valve is faulty on an EGR system. Now we can tell the driver that his/her truck has a problem and [which location to stop at] that has the part on hand [or what will happen if they don’t stop]. We are trying to get way ahead and predict when the doser valve will fail based on historic data.
Riemer: It will be harder and harder for fleets [doing their own maintenance] to attract and maintain skilled technicians for little more than PMs. On a grand scale [and not counting rural locations and off-highway requirements], I believe there will be major consolidation around outsourced PMs, tires and light repair. A few large players…will probably be able to take on other work and maintain their technicians because of their buying power with OEMs.
Q. Will the ability to benchmark against best practices be enhanced in the new world of predictive maintenance?
Glasmann: Fleets do want to do benchmarking, but it is very difficult to get true apples to apples comparisons to do that. They have a better chance of being successful doing that by sharing data within their own operation.
Pfaffenbach: Fleets always have an interest in benchmarking, but we are hesitant. At least for now, benchmarking probably has to go through an industry organization or association to get wide acceptance and [valid] results.
Riemer: I think there will be more and more benchmarking between fleets, and we are actually working with the Truckload Carriers Assn. to support such an initiative.
However, as you can imagine, the quality, reliability and consistency of data across fleets is extremely variable, but as the trucking service supply chain ecosystem adopts service relationship management technology…the frictionless flow of information between all parties and the fleet will become a great source of competitive differentiation.
Q. Does more data mean more onboard sensors and instrumentation?
Cerilli: We need more and better sensors and actuators; the more we can integrate the better.
Pfaffenbach: Let’s not get carried away with adding instrumentation to the vehicle. If you are going to do that, there has to be an ROI and it has to be practical and not cause new problems even as it solves others. How instrumentation will evolve is not certain, but it will probably be driven by need. Remember, we already have lots of data that we’ve only begun to mine for useful information.
Q. What do fleets want when it comes to maintenance data access/presentation?
Pfaffenbach: How we present information to users is often overlooked. The primary audience is the operations manager or the maintenance manager, but the driver is also a part of the audience and so is the fleet owner. We could even include the fleet’s customers, as well. And each of them interacts with a lot of different people every day. To me, this is a really big topic that I’d like to focus on more. It requires a fair amount of discipline, for instance, to report only what is needed. [With big data and today’s analytics], there is definitely an opportunity to over-communicate.
Cooper: We hear a lot from our customers about the desire for a single platform, a single dashboard that provides a great line of sight into the enterprise. Fleets want to get the biggest bang for the buck. Their question is: How can I make my vehicle assets deliver more value to my business?
Q. Bottom line, will the industry really see true predictive/prescriptive maintenance in the foreseeable future?
Reed: We mandate the use of VMRS codes in the Dossier system if you are doing repairs, so if the codes are correctly entered, it can help take you to the reasons why a failure occurred. I don’t put much faith in prognostics, however, because product changes occur that don’t necessarily change part numbers, but they can change failure rates and you might not know that. You have to keep measuring and testing and watching the data all the time.
Cerilli: Remote diagnostics is really in its infancy. It has only been around for a very few years, but it will be a core competency for OEMs—table stakes. Today at Navistar, we are ranking fault codes and turning unscheduled events into scheduled events, into preemptive maintenance and repair. Other industries are ahead of trucking, though, such as underwater mining and locomotives.
Croke: At Omnitracs, we have aggregated maintenance fault code data now from thousands of vehicles, and we are aiming at predicting service intervals to fix things before they break. [It would be great every morning for a fleet to be able to know] that these are the trucks where there is an opportunity to move a PM up to fix something before it breaks—or move it out to get more value from a still-good part.
The viability of successful predictive/prescriptive maintenance programs are still part of the Great Unknown, though. We still have to go through beta testing to see if this will work. The question is, can technicians go in and fix or replace something that may not appear to need to be replaced and prevent a problem?
Deedy: Volvo’s Remote Diagnostics was launched on Volvo-powered tucks in 2012…. The prognostics stage is the next logical step. An OEM has to have a very high degree of confidence before it can recommend to a customer that they replace an expensive component before it fails based solely on a data profile. Developing that level of certainty will take time, so prognostics will come in stages.
The next step in the evolution likely will be to use vehicle data to modify maintenance practices. In the future,
analytics could lead to adjusting maintenance schedules based on duty cycles rather than on miles or time elapsed.
Ahadi: More and more, customers want to connect to improve service and maintenance. They want to handle maintenance in the best, most efficient way and at the least cost. We launched an oil change minder, for instance, that is a complex algorithm that looks at idle time, etc., and not just mileage or engine hours. Fleets don’t want to change oil too early or too late. That also prevents other things, like a blown engine. We have customers who are sending these oil change alerts to their dealers as well. That allows service managers to schedule better.
Our vision, our underlying strategy is to keep trucks on the road, to foresee pending issues and plan ahead. We are working on that, on prognostics.
Riemer: I think PM cycles will become more predictable and effective as the data becomes more reliable and consistent. I think the concept of prognostics and predictive maintenance could be a reality, but not based on the typical data quality and consistency that most fleets capture today.
Cooper: Two years or so is a good estimate, I think, for when we will have very robust, proactive predictive maintenance and maybe even auto-scheduling. This will have compliance implications for things like CSA and HOS. It will also open up opportunities for maintenance-related suppliers to do things like loyalty couponing, for instance.
