From an experimental point of view, the first thing to avoid is superficial tension. Thus, the channel should not be too small. A minimum of 50 centimeters wide is recommended. In order to study effectively, the varied flow, it is necessary that the length of the channel is sufficient. A length of 6 meters will allow you to visualize all the scenarios. The height of the canal from 50 centimeters to 60 centimeters would be ideal. The volume flow rate must not be less than 30 liters per second. Obviously, all this relates to the rectangular channel. With these minimum dimensions, you can make good tests.

Regards

I have seen quite large flumes, but most of them were wave flumes, I have seen a flume ~3 meter wide and ~90 meter long (approximately). You do not need such big flumes in most of the cases. Your controlling factors are:

- Fully developed boundary layer length is needed (you need boundary layer hits the free surface) and 100 times of the depth is very nice and decent flume.

- Wide flumes decrease effect of wall effect on one hand but you may see 3D effect on the other.

- Some works like dam breach modeling, or non prismatic spillways, or meandering and gully formation may need a wide flume.

- For large flumes you usually cannot circulate the water, in San Antony Falls of Minnesota, Iowa IAHR, Utah Water Lab in Logan and most important hydraulic labs, they are either directly next to a river bank or under a dam to get the water of large flumes from a source. Also you need a very big pumps for those flumes.

- Very large flumes are mostly not tilting (you cannot control the slope) and their walls are concrete or brick wall.

My suggestion is you do not necessarily need a big flume, based on your experiment maybe a mid size one works very well.

Perhaps this link might be useful https://www.openchannelflow.com/flumes/hs-h-hl-flumes

At University of Missouri Research Forest we used 3 foot H flumes. The concrete wall constrained the flow into a concrete channel that was maybe 3-6 feet wide and 20-30 feet long, with flume at outlet. The flumes were installed and functional when I was hired, but constucted to standards in the 1960s. The size of the intermittent catchments was 10-15 acres. To my knowledge, they never overflowed, and we had some severe intensity storms, approximately 5 inches or more per hour during one storm, and it was so hard it seemed to take my breathe away. I uploaded my thesis (1975) a few years back into researchgate that may have other details or useful reference materials if doing experimental catchment studies. I would advise design to contain at least to the upper confidence limit of 100 to 500 year flow, and have constructed concrete dimension with transducer backup to help measure any extreme flow above design or in case of equipment failure, as lost record recovery is difficult and time consuming. Its best to avoid circumstance where you might have extreme historic peak flow, and your design was insufficient to capture the flood. The H flume needed regular servicing to remove sticks or other debris and check function of equipment.

I have done experimental work using 31cm wide and 7.5m long flume and it has given great results for various experiments including RVF analysis

In Hydraulic Laboratory of WARSAW UNIVERSITY OF LIFE SCIENCES – SGGW (WULS-SGGW) we have and work on flume 20 m long, 0.6 m wide, 1.0 m high and smaller others.