A groundbreaking physical modeling study was published this week in PNAS (subscription required). NSF has a press release here. Christian Braudrick and others at EPS-Berkeley used a 17m flume filled with a mix of lithic and plastic particles to achieve what they call "the successful experimental generation of a lateral migrating, bedload-dominated, meandering channel with repeated cutoffs."
Using the lithic/plastic sediment mix, and stopping flow periodically to replant alfalfa used to create hydraulic roughness and bank cohesion, they created a remarkable simulation of river behavior over a 136 hour run. The photo above shows the remarkable resemblance of the channel and floodplain to a larger fluvial system.
Building on Bill Dietrich’s work on meandering and bedload transport through bends, the authors observed that matching outside bank erosion rates to bar deposition was critical, and that fine sediments (in this case the plastic particles) were necessary to fill in the chutes that form between point bars and the higher floodplain to allow the bar surface to build as channel migration occurred. Very good stuff that beautifully matches with my observations of our models and what happens in real rivers.
I’ve been immersed in this topic lately, so this paper was a very timely and pleasant surprise. Especially since the authors used plastic media similar to that in our Emriver models and concluded that their scaling, both geometrical and temporal, of roughly 1/50 to 1/100 produced realistic results.
Just yesterday (!) I arrived at a similar scale conclusions based on Reynolds number analysis and measurements of our models: Our Em4 is capable of scaling small streams (e.g., with a 7m bankfull width) at 1/50. At this scale the media particles would scale to a D50 of ~40mm (same as the paper). We’re still working on the alfalfa issue (and cohesive banks), but clearly this research supports use of models like our Em4 for research on real world rivers. The figure below shows how Reynolds numbers for a small real world stream would scale to a physical model; below 1/75 or so the distortion is reasonable and you can expect fully turbulent flow and a decent Froude relationship.
The authors addressed time scales in some detail in the suppliments to the paper--very good stuff, because that's been nearly ignored in the engineering literature on modeling because most of that work uses models that are not time-dependent.
You can see a movie of the experiment at PNAC, apparently not behind a paywall, here.
The paper’s title is Experimental evidence for the conditions necessary to sustain meandering in coarse-bedded rivers; authors are Christian Braudrick, William Dietrich, Glen T. Leverichb, and Leonard Sklarb.
Photo NSF and PNAC.