Engineering | Liam Grace-Flood

Winter break of my senior year, my friend Nadav and I built an electric motorcycle for Nightshift Bikes. We started with a 1975 BMW R75/5 that was flooded in Hurricane Katrina, and in 3 weeks, stripped it, sand blasted the salvageable parts, and built a fully electric city scrambler on that frame.

See more photos of our final product

& photos of our build (1) (2) (3)

More photos can be found on instagram @nightshiftbikes

Instead of designing one piece of furniture, here I designed a set of constraints and relationships defining a family of objects. More here.

Drawing above by Summer Madox.
For a Thayer engineering course, Digital Electronics, Nadav Hendel & I used VHDL to design digital circuits. For our final project, we programmed an FPGA to work as a multicolor Etch-a-Sketch.

Our VHDL code and report can be found here if you want to program your own FPGA, or if you're just curious.

A musical interface I designed, called Hexba, as part of a grant-project exploring new music technology. As part of that, I also built a recording studio.

Fall ’15 I worked with three other Thayer students (Ralf Carestia, Nadav Hendel, and Keenan Wood) to design and build a 6 inch long cantilevered bridge, and to accurately predict the load and point of failure at an astounding 400 pounds.

solidworks prediction images:

our constructed bridge & testing apparatus:

traffic flow matlab analysis December '14: graphs & report

However, the way in which cars tend to slow down is varied and potentially very weird. My code modeled flow around a 10 km circumference track to approximate traffic on a stretch of highway, and initial traffic density was determined randomly for each point on the road, centered around a particular preset value. The few parameters of traffic density, rate of car flow, rate of acceleration, maximum speed, maximum traffic density supported by the road, and most importantly, initial traffic density, drastically affected the behavior of traffic.

If initial traffic density was well below the “carrying capacity” of the road, the random spikes in traffic would quickly diffuse and traffic would assume an evenly spread distribution. However, for greater values of initial traffic density, the initial spikes would grow larger and larger, eventually settling in to a pattern of regular emissions of cars in quick bursts, which are then immediately slowed by the next peaked jam. Depending on these values, there appears no limit to possible patterns- I’ve seen traffic densities travel against the direction of traffic, those strange pulsating sporadic swells, and converging jams.

For initial traffic densities below or even slightly above the maximum, traffic will quickly diffuse, and the plot of position vs density will approach flatness.
For initial traffic densities much higher than the maximum, spikes will become more and more pronounced, and cars will move to the right from one jam to the next. Making traffic density higher or lower as long as its above a certain point will not affect the overall behavior of the traffic. However, we do see the number of spikes increasing as initial traffic density increases.

For both of these possibilities, the traffic density will sometimes appear to move forwards, sometimes appear to move backwards, and sometimes stagnate.

Above is a graph of just one set of initial conditions– if you want to see more, contact me for my matlab code!

In summary, this project was really cool in helping me understand what plays an unfortunately large part in my life: traffic. It also showed that otherwise difficult to comprehend systems can be understood using relatively simple methods: the same tools and models we apply to natural systems can be used to demonstrate the laws of man-made things like traffic, too.