The idea of automotive “prognostics” (a combination of prognosis and diagnostics) to determine remaining useful life is not new. Motorists long have had a garage technician perform a routine diagnostic check on the car, looking at tread on the tires and condition of the sidewalls, and obvious signs of an impending malfunction, such as a large fluid leak from the cooling system, engine or transmission, and suspension/steering looseness. Finally, a scan tool readout could be used to find any vehicle trouble codes.
The increased use of vehicle sensors for all mechanical and electrical systems, however, provides opportunities for a major expansion of automatically generated prognostics, with contributions from computers in the multiple data buses. Two full-day sessions on prognostics were held at the recent SAE 2012 World Congress, illustrating an increasing interest in the subject.
A comparison was made to the aircraft industry, for which the health of the plane is a more safety-critical issue than automotive. However, as observed by Keith Jackson, a consultant and visiting professor at the University of Sheffield, “a plane that can’t be dispatched can leave 300 angry customers, so there’s also a huge business case.” He said that aviation data analysis has to be done very carefully, noting that an aircraft engine vibration must be compared with the level from an earlier period. Prof. Jackson, an SAE Fellow who does work for Rolls Royce PLC, said that although the greater vibration may still be within an acceptable range, an increase in the level would likely be considered a more valid prognostics indicator.
The safety issue means Federal regulators have an especially tight focus on aviation, said Dr. Ravi Rajamani of Meggitt USA, an aviation engineering company. He said that when he worked for Pratt & Whitney, it was difficult to justify a change that could safely extend a maintenance interval unless it paid for itself within 25 years. But even if a company finds a change can extend service intervals, it must be able to convince regulators—not easy to do.
Although it might seem that many automotive areas can be ignored, for many operators prognostics do fit in. Tim Cavanaugh, Director of Condition-Based Maintenance for BAE Systems, pointed out that an unanticipated failure of a PTO (power take-off) could create an expensive delay in a project.
The U.S. Army also has an obvious major interest in prognostics for military vehicles, observed Christopher Thompson, Program Manager, Remote Monitoring and Diagnostics for General Electric Power & Water. With such detailed information as “acceleration and how fast the vehicles stop, we can predict when the big, heavy wheels must be pulled for service,” added Cavanaugh.
The motorist whose car has one of the factory-installed infotainment systems, probably has been receiving “vehicle health reports.” But the information in these notices is more focused on immediate, obvious needs. They include tire pressures as recorded by the tire-pressure-monitoring system, oil change intervals as determined by a computer algorithm, and emissions-related trouble codes from the onboard diagnostic system.
An important new prognostics scheme, which points to a need in a future with more electric vehicles, is used on the Nissan Leaf. During one of the periodic service visits, the dealer technician connects a PC-based Consult scan tool into the onboard diagnostic system and gets a reading of the capacity of each of the 192 cells in the lithium-ion battery pack. They nominally are rated at 3800 mV but in good condition run very slightly higher, with 3931 mV specified in a Leaf tested for AEI.
A Consult program can flag any comparatively suspect cells. The capacity loss rate should average 2% per year or less, so there is 80% capacity remaining after 10 years. If the rate is excessive, the technician can interview the driver, using a factory checklist, to see if there is some aspect of charging procedure or vehicle operation causing it or, if not, inspecting for an electrical issue.
Originally, the Leaf could have uploaded a lot of data, identifying driving and battery recharging patterns that might contribute to an excessive battery-capacity loss and other potential problem areas. But an Internet blogger was able to download a lot of information that Nissan didn’t even realize was available, and privacy issues were raised. So the software was changed and now only those who sign an access agreement with Nissan can release some of the information, which is automatically processed by aggregator software, so no individual cars can be identified. The privacy concerns and laws have affected all the OE infotainment systems, starting with OnStar. They’re all limited in what data they can “mine.”
However, the potential for prognostics data tied to an individual car does exist, and with a cooperative motorist and the right algorithms, there could be developed a wide range of remaining useful life information. So long as access is within the car owner’s control, there should be no privacy issue.
OBD II 'Mode 6'
Since 1995, all U.S. cars have OBD II (onboard diagnostics, level 2) and one of its features is “Mode 6,” which automatically performs periodic tests of such emissions systems as the condition of the catalytic converter and the ignition system for misfire. It is part of SAE J1979, the standard that covers OBD II communication between the vehicle and a scan tool.
Momentary anomalies can create a “false-trigger.” So before logging a trouble code and turning on a dashboard malfunction indicator light (MIL), the OBD II system “counts” failures, but until they reach a certain level, it just stores a pending code. It’s possible for a technician to read this “Mode 6” data on a scan tool, and see an impending failure.
In fact, if privacy and warranty legal concerns are overcome, diagnostic routines could predict impending mechanical and hydraulic problems with PTOs, all-wheel-drive couplings, and transmissions. The onboard computers could measure slippage through clutch packs by looking at rpm differentials and algorithm-derived torque throughputs. They even could detect needs for increased clamping pressures in CVTs (continuously variable transmissions). Brake-lining wear indicators have long been used to detect a need for replacement, but more advanced ones could predict remaining lining life.
There are thousands of possible trouble codes and PIDs (parameter IDs or data items) in the powertrain and driveline computer systems of a late-model car with OBD II, the comfort and convenience systems (such as chassis ride control, power seats with memoryl power door locks, tailgates and windowsl automatic temperature control, seat heating and cooling), safety systems (airbags, intelligent seat belts, rear camera, adaptive cruise, cross lane and lane drift detection, etc.), and infotainment systems. The codes and PIDs for all these could be integrated into algorithms for numerous prognostics.
Telematics with access to the cloud provides an enhanced opportunity for automotive prognostics. Onboard ECUs (electronic control units) are not designed to do extensive data crunching prognostics, said Louis Scott Bolt, Chief Engineer of Mahle Test Systems, because of their limited RAM, ROM, and CPU capabilities. With more wireless, he added, “it’s easier to get data out.”
However, as Jackson said of an overall approach to prognostics, “accuracy of detection with minimum of false positives” is critical to “avoid an angry customer in the shop.”