Summer 2015: Testing Arctan

Arctan has now air shipped off to Australia and team members have flown to meet it, so we have time to take a breath and reflect on our busy summer!
Like previous summers before WSC, the team unveiled its new car to the public and set about testing. Unlike previous cycles, we knew from the start that we would be air-shipping our car to Australia, and thus spaced our test drives throughout July and August accordingly. This gave us time for three full-weekend test drives in the Central Valley, a day at the track at Thunderhill Raceway, and two back-to-back mock races designed to test the strategy model by mimicking actual race conditions. We also had numerous drives near and on campus, in addition to bench-testing of all critical systems.
The team learned a valuable lesson from the success of its 2013 cycle with Luminos, which we now use in hashtags, signage, and the lettering on our trailer: Test It Again. One of the greatest contributors to the no-stops race and reliable vehicle that we had in 2013 was the extensive testing leading up to WSC. Following the unveiling of this year’s car, Arctan, we set off on a rigorous summer to develop the same or greater degree of reliability. Each test drive had a purpose and came with valuable lessons, chronicled below.

Central Valley Test Drive One: July 18th

Test Drive 1
Our first Central Valley test drive was unfortunately cut short to only one day of driving due to an unexpected hail and lightning storm as well as an issue with our battery pack. While driving, we noticed a constant, steady-state difference in lowest and highest cell voltages, which indicated a missing weld in the battery pack. This was easily remedied once back at VAIL, but could have been a major failure point had it gone unnoticed. The telemetry system did its job and warned us of this problem, but we still modified the display to make any future similar problems very obvious.
In a post-test drive debrief we noted several teamwork practices that needed improvement, mostly organizational things such as role assignment, communicating timing, and checking the weather forecast. All changes were implemented for the next test drive, which ran much more smoothly.

Central Valley Test Drive Two: August 1st to 3rd

We performed two sets of controlled speed runs over the course of the weekend to characterize our strategy models. The speeds we used were compensated to account for new Michelin tires which have an 8 mm smaller diameter than what was programmed into the Tritiums motor controllers.
Our controlled speed runs were driven between 50 to 90 kph, in 10 kph increments. We drove at each speed in both directions and in random order to minimize the influence of wind speed, direction, and solar intensity based on time of day.
We also performed a battery pack test to verify bottom-of-pack behavior. During this test we were also able to verify the accuracy of our state of charge measurements to estimate when the car actually would run out of charge and shut off.

Central Valley Test Drive Three: August 15th and 16th

Central Valley early morning

We dedicated all of the 16th to characterize our array. We noticed that a few of the MPPT’s occasionally latched onto local maxima with lower voltages. The code on the trackers have been since modified to run full curvetraces so we won’t have this problem during future test drives. We should also see an increase in array power in the future.
We also managed to get through another set of controlled speed runs, which agreed with our previous models taken during our second Central Valley test drive.

Thunderhill Raceway: August 24th

Arctan_thunderhill_small
Like in 2013, we were able to join the Dynamic Design Lab on a trip to Thunderhill for track testing. We only had a few brief sessions on the track over the course of a single day, but we used this time to ensure all of our drivers could pass the WSC-mandated braking tests, in addition to light cornering and slalom testing. The controlled environment of a closed race-course allowed our drivers to safely gain handling experience in these more dynamic situations–something we feel is important in developing an experimental vehicle. It was also very interesting to see the other cars from VAIL in action.

Mock Race: August 27th to 30th

mock race
Over the course of four days, we completed two 500-mile mock races, which we used mainly to test our strategy model and let the race crew and drivers on the timing of stops during the actual race. Our course was set up to have similar solar angle to the actual race and have thirty-minute control stops approximately every 200 miles. This left us with more free time in the Central Valley than our typical test drives, which don’t have the set stop-times that we’ll encounter on the race.
At the end of the summer, we packed everything up back at VAIL, cleaned off all of the Central Valley dust from the car and supplies after a total of nearly 2800 miles on the car. See you in Australia!

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *