This week's demo has been a little late in coming - but I've got a few moments to get this up. Video is still processing though...
The plastic media that comes with the Emriver stream table is made of ground-up plastic. The stream model's physical behavior is largely determined by the relative difference between the density of flowing water and the sediment (there's also the viscosity of water, but that's another set of posts).
The above graph shows my student's results from soil mechanics lab last week. Students' results (red points) are a bit more varied than mine (blue). Aside from one errant point, student results lie along, or to the left of my own results - this suggests to me that it's a measurement error, rather than very different material. The leftward distribution points to less water than expected, rather than more. This could be the result of letting too much water dribble down the side and not into the measuring container, or waiting for all the water to stop dripping out of the spout. Or that the coarser fraction doesn't displace water as easily as a more graded mixture. Still, for 10-15 minutes worth of lab work, not a bad set of data.
This graph shows the results from a more extended set of measurements. The lab had students using a small (~100ml) overflow beaker - small errors like missed drops end up having a very large effect. So I tried using a much larger overflow beaker. My results were very consistent (basically a SG of about 1.50 ±0.01). But these values are a bit lighter than what Steve Gough (head of the LRRD) had for their color-coded material at 1.7. It's possible that my method allows for too much material to cling to the top of the beaker. Or the air bubbles trapped next to the surface of the plastic result in lighter-than-actual measurements.
So why graph the data this way? I can measure the plastic media's mass very accurately. And the amount of water displaced is proportional to the specific gravity of the material. So by making a bunch of measurements of two values, I can define a third as a linear function of the other two. It also saves some time on the calculation side of things.