Testing - Cooling Methods and Temperature Distribution under High Loads

Before diving in to manufacturing the modules, we need to have a solid understanding of how our modules heat up, how much cooling we need, and where that cooling will come from.

This testing was originally started to evaluate the effectiveness of cooling our modules through the catamaran, but has since taken on more of a 'let's get more cooling' attitude.

Testing Setup:

Switch acquired form Amazon (200A or 300A battery disconnect switch).

Fuses: littlefuse Midi 200A

Highcurrent cables: 8AWG, silicone insulation, from Mouser

Anderson connectors from the bay.

Items attached to wood with mainly wood screws (we don't seem to have very many short ones - had to dig through the misc. bin).

Anderson connector attached with 3M grey VHB tape.

Method:

The temperatures of all the cells were logged once per seconds with the 64 ch DAQ, during a discharge at 100A through the DC load.

We ran the test with and without the fan running, and with different fans.

Data was also collected for a charging test at 20A and at 30A.

Results:

We first graphed the temperature difference between a cell in the center of the module and a cell on the corner of the module, with and without the fan running. While the cells to reach an uncomfortably high temperature, this is after a long time discharging at 100A. We only plan to hit 100A going up hills, and for brief periods of acceleration, so this may not be an issue. We will continue to analyze the data to examine rates of heat rise and see how much heat we are really pulling out of the module.

Note: The slight 'concave up' section near the end of the graph with the fan is likely due to the cells' internal resistance rising near the end of the discharge cycle, thus producing more heat.

We then graphed the temperature distribution throughout the module. I had some fun figuring out the matplotlib pcolormesh, colormaps, and interpolation to smooth it out.

Note: these graphs were produced using an arbitrary log point of the data that showed us what we wanted. It is not the max temperature that the module reached (still working on the python script for that). It should also be noted that inside the module chamber cardboard box, there is more room towards the end shown on the left of the graphs. This left end is the intake end for the fan. The data was only collected for one half of the module and reflected over the other half. The thermistor wires do come out the side of the module and will have some effect on the airflow touching the cells, but the hottest cells are not restricted too much.

From these graphs, it looks like we will need either more airflow, a more directed airflow, or a channel to get more air through the middle of the module.

Cooling the module with a carbon fiber plate proved to be ineffective. The carbon fiber did not raise in temperature at all. This was verified by attaching a heater to the carbon fiber plate, and observing where the carbon fiber heated up. We measured the temperatures of the carbon fiber plate and concluded that the thermal conductivity through the plane (through the layers of carbon, foam core, carbon) was much greater than across the plane. The temperature on the opposite side of the plate, underneath the heater was fairly hot (almost at the temperature of the heater) and the temperature of the end of the carbon plate on the same side as the heater did not noticeably rise in temp above ambient.

A few articles on the thermal characteristics of carbon fiber epoxy are linked below:

https://www.christinedemerchant.com/carbon_characteristics_heat_conductivity.html

https://www.researchgate.net/publication/262007032_Thermal_properties_of_carbon_fiberepoxy_composites_with_different_fabric_weaves

We will experiment with some airflow guides to direct the air through the module instead of around it as well.