Inside GM’s Expanded Electric-Vehicle Battery Lab
The battery lab tests a host of EV-driving scenarios and conditions, and up to 600 kW of continuous power.
General Motors aims to sell more than one million electric vehicles (EVs) annually in North America and China by mid-decade. The pathway to success leads directly through the company’s recently expanded, 100,000 sq-ft2 Estes Engineering Center – Global Battery Systems Lab, located at its Global Technical Center in Warren, Mich. Doug Drauch, lead engineer at the lab, along with Anthony Modafferi, engineering group manager for battery systems development and validation, designed the facility.
Drauch explained the central importance of the lab’s battery-cyclers, high-power devices that feature a 480-volt, AC, three-phase power connection to utility lines. There also is an AC-to-DC converter with output levels for voltage and current and controlled by an external PC. Finally, a DC-to-AC inverter draws current from the battery pack or cells being tested. “The battery-cyclers allow us to change the amplitude and direction of the power flow,” said Drauch. The power is controlled in 10-ms increments.
When a battery pack or cell is ready for testing, it’s placed inside environmental chambers that range in size from a couple of cubic feet to more than 1,000 cubic feet. The temperature in the chamber can be set between -68 deg C (-90 deg. F) to 85 deg C (185 deg F). The relative humidity can be controlled from 5% to 98%. Most of GM’s battery packs use liquid thermal-management systems, so a using a “chiller” allows engineers to also manage the temperature and flow of the coolant through the pack. The chiller essentially mimics the liquid thermal system used in a vehicle.
Replicating pack performance
The lab maintains an extensive library of drive-cycle files for every imaginable type of EV-driving scenario. GM produces the files by instrumenting vehicles during typical driving events, such as an errand to the grocery store. “If you record all the information, we can play it back through the battery-cycler,” said Drauch. Engineers also can create drive-cycle files from a math model based on known power requirements and losses.
GM uses a separate rack to supply inputs and outputs mimicking discrete signals – such as wake-up and run – that a battery would receive in the vehicle. This allows the testing software to communicate with the battery pack the same way a vehicle’s controllers would. There’s also a device that mimics the vehicle’s serial communication channels. Those channels deliver information about the state of the car, such as charging, ambient temperature, cell voltages, current and faults.
In early 2020, Drauch and his team upgraded 16 test cells with 600-kW pack cyclers. As of mid-March, 13 of these test cells were commissioned and operational. When the remaining three test cells are put into operation, expected by the end of May 2020, there will be a total of 19 high-power cyclers in the lab. They will utilize 38 separate test channels. The latest cyclers, which can run 24/7, upped their capabilities to 600 kW of continuous power but can peak to 660 kW for up to three seconds.
The increase equips GM to test batteries that might be used in the GMC Hummer EV, for example, due in 2021. The electric Hummer will have the option of a 200-kWh pack capable of about 400 miles (644 km) of driving range and will have an 800-volt electrical architecture. GM also evaluates battery cells from different vendors. “We put them through their paces,” said Drauch. He explained GM tries to obtain every commercially available cell – from suppliers, emerging startups and cells pulled from the packs of competing vehicles.
Lifetime of road abuse
But what about replicating the abuse a battery pack experiences when pounding the pavement? For that, packs are placed, one at a time, in one of the lab’s three “shakers.” Battery packs are shaken and rattled along three different axes. The shakers also can approximate different drive cycles and roadways – for example, as if the vehicle was rumbling at speed over Belgian blocks.
A typical vibration test takes 87 hours, with three 29-hour periods for each x-, y-, and z-axis. The process utilizes complex algorithms to determine what frequency and energy level to apply throughout the test cycle. “You find out if anything breaks or comes loose,” said Drauch. Eighty-seven hours is enough to evaluate the level of real-world abuse a battery pack would experience during its lifetime. Including the intricate battery installation on the shaker and its removal, the process requires roughly five workdays.
In a separate evaluation, a “nail pen fixture” utilizes a PC-controlled press with a chuck that holds a nail. The device’s press – using a tightly controlled speed, force and linear location – rams a nail through the side of a pack and into a cell to a precise distance. That “propagation” test approximates what would happen if road debris penetrated a cell to evaluate if any voltage or heat spreads throughout the module.
GM’s battery lab requires a lot of juice. If all 19 cyclers are in use at the same time, the peak power requirements are roughly 25.7 Megawatts, according to Drauch. That’s based on pulling AC from utility lines. In the unlikely event that all channels in the lab, pack and cell are also maxed out – plus the PCs, chillers, and I/O racks – the draw could climb to as much as 34.4 MW.
But Drauch explained that the lab’s inverters reduce the overall power bill by constantly pulling power out of the tested packs, converting it to three-phase and pushing it back into the grid. “We’re pulling power out of one test cell, then pushing it into another and pulling it out of here and pushing it over there, bouncing it all around within the lab and the building,” explained Drauch. “We’re recycling those electrons.”