Some results from the Cart-to-Solar project

With a deadline looming (and past) and no students around since the start of summer final exams, I spent part of last week and this weekend putting together a mid-point report to the Foundation on the Cart-to-Solar project. As you all recall, Aaron did a great job getting this funded and on its way to a successful conclusion this summer, organizing a team that carried on the work even when he was not here. However, I was not aware of just how well the project worked out until I processed the data into the form shown below.

The baseline data acquired during the spring are shown with diamonds on this graph. (The two green squares are results of an initial trial with the solar panels and should be ignored here.) I made the three points that were outliers red and excluded them when computing the mean and sample standard deviation shown with the blue band. The yellow band shows the average and standard deviation if all points are included.


You can see for yourself how consistent most of the data were. (The gap in March is Spring Break, and one of the anomalous points appeared when the cart sat idle while the campus was closed.) Aaron did some nice work to try to understand what might be going on with those red points, but I won't include the graph or try to explain it here.

FYI, getting about three miles per kWhr corresponds to an operating cost of about 5 cents per mile for "fuel". (This does not count the cost of amortizing the cost of replacement batteries every few years.) These carts are pretty efficient.

The data acquired after the solar panels installed are shown with squares. There are really three sets of data here: The green squares were some initial tests with three panels, the first set of maroon squares were tests with six panels used in one configuration, while the ones in the region following a break for spring finals, the area covered by the green band, used a different configuration.


You can see very clearly that almost all of the data in this last set show performance that is better than any of the baseline observations. I speculate that the three low points were overcast days but found no weather records to use to test this guess. The green band shows the mean and sample standard deviation for this set (including the three low points). The average value is about a 50% improvement on the efficiency of the cart, which is, I think, about what Aaron had predicted. The band is wider than for the baseline data because of the variability introduced by weather conditions.

FYI, adding the panels saved about $2 worth of electricity over those six weeks. An interesting question is whether they will improve battery life and reduce that cost as well.

In my opinion, the primary goals of the project have been completed but (as discussed at our meeting on July 11, there are still some great things to do this fall to wrap up work with one cart, improve how it works, do the final report, and possibly start a longer-term project monitoring battery life and/or convert other carts.

Good work, everyone!