The project entails a 140-vehicle field trial to evaluate customer acceptance and battery performance across a wide range of drive cycles and temperature ambients. The program is slated to rack of 9 million miles of data, with 1.4 million miles in sub-freezing temperatures and 1.1 million in temperatures above 32 °C. In addition to battery again and life cycle analysis, the project is exploring smart grid interfaces; bi-directional charging; second life—i.e., grid storage and load balancing for the battery pack; and integration with renewable energy.

The Ram PHEV is a blended plug-in hybrid, meaning that it doesn’t run exclusively all-electric during its charge depletion period. The fully-charged plug-in starts off with charge depletion with limited regeneration at the high end of the SoC; the team will look at calibrations in the field for full optimization, Gross said.

That ramps up to a full regenerative capability somewhere in the 70 - 95% range and depletes down to about 20%. Once it’s depleted, it comes into a narrow charge sustaining range.

The PHEV life challenge, Gross said, is that a PHEV is expected to demonstrate durability comparable to non-electrified vehicles. This then brings attention to the AT-PZEV battery warranty requirements of 10 years, 150,000 miles. This has been translated into a set of basic battery requirements by USABC:

• 15 years calendar life at 30 °C (recently changed from 35 °C); and
• 5,000 charge-depleting (CD) and 300,000 charge-sustaining (CS) cycles (e.g., microcycles) (75 MWh total energy throughput, for a nominal PHEV-20).

To date, no battery meeting PHEV performance and packaging requirement has been shown to meet the calendar and cycle life requirements, Gross said, although some have come close.

Battery life. For a given chemistry, capacity loss and impedance growth over calendar time will comply with a linear time relation, a square-root time relation, or in-between, Gross said. This can only be determined under controlled test conditions; accelerated testing is used to determine this model fitting.

Probably the most common way [Chrysler] looks at battery calendar life is we do Arrhenius modeling... The Arrhenius models are well done and well understood, its very simple to take this and then say, for example based on some experience with this location, offset them and adjust them for particular solar episodes....this way you can actually start looking at your data, processing it, trying to get some ideas as to how long you are really going to be able to operate this vehicle and meet the end of life requirements.

—Oliver Gross

With the derived acceleration factors, you can begin to do simple prediction on calendar life and the effects of operating temperature on life.

Some of their initial results show that operating 3 hours per day with a 10 °C operating temperature above ambient can lower battery life by 5%—which, for the battery pack in that evaluation, would take it to an end-of-life calendar life of just over 15 years for some locations.

Chrysler’s battery use model blends UDDS, HWYFET and US06 cycles to represent multiple use scenarios. Putting this all together (merging calendar life, energy throughput and an operational temperature increase of 10 °C), resulted in product life estimates that vary by location and by miles driven.

Battery packs in use in Los Angeles, Phoenix and Miami had the lowest estimated calendar lives, at 11.8, 11.6, and 11.6 years respectively at 10,000 miles per year. That dropped to 7.6, 7.5 and 7.5 years, respectively, at 30,000 miles/year.

The longest calendar lives were for packs modeled to be in use in Minneapolis, Detroit and Fairbanks, with 15.8, 15.4 and 18.6 years, respectively, at 10,000 miles/year. That dropped to 8.6, 8.5 and 9.2 years, respectively, and 30,000 miles/year.

These vehicles are now going to be out in the field for three years. We are going to be collecting the data remotely, and we have a good ensemble and team out there helping us process the data and then we are going to apply that to the battery lab cycle and calendar life data to see exactly is this model correct, is it calibrated, and then we can actually apply it back onto a new set of data where we would then modulate the operational performance and begin to optimize the vehicle.

This is what we really want because this is what we are going to use to determine the proper PHEV profiles we will be using in the future.

—Oliver Gross