I've been trying to decide what a good scale test would consist of. Maybe talcum powder and thanks to your above mentioned addition a little flour to bind the concoction. Wet it and form it into a basin then let it dry. Afterwords, add water to the basin until it breaches the rim. I think I would have to maybe add some high frequency vibration to simulate earthquake conditions. What do you think? My brother has a machine shop so we could probably make a nice scale test. I even considered using something thin like alcohol to partially overcome the mismatched scale viscosity of water.
I'm not sure what you're testing for there. If it's a simple erosion scenario from the top down, I think you'll be disappointed.
I don't want to discourage you; rather I'd say you need to plan carefully or be prepared to spend a lot of time. I'm confident there's lots to be learned.
I suppose I had a somewhat unique game as a child. Used to find people who watered their lawns on schedule. Eventually, a slow stream of water would be produced in the gutter, making its way toward the storm drain. I used to build dams out of sticks, grass, dirt, pebbles - whatever was handy. Scale was limited because I wasn't allowed to play in the street
, which I interpreted as the asphalt.
Now the shapes mud takes (dirt turned out to be the least leaky material) when the dam is breached are identical to the shapes of the various hills (mesa, butte, etc.) one sees out west. Scaling isn't even an issue; they match perfectly. When I got older and they told me dirt and sand piled up over the years, and the wind shaped it so... I was unconvinced. When I later discovered flood geology, I just essentially said "there you go. It's obvious." It's such a perfect fit I didn't see how anyone could think differently for an instant. But then it occurred to me that not everyone builds dirt dams...
Investigating the canyons may be a little more difficult. The flow of trickling water on a flat surface is unpredictable. If you arrange your surface properly, you can largely dictate the initial course it will take.
Dealing with muds & pastes of different consistencies will produce different results. Another problem is how to contain the mud while allowing water to escape. The flow of streams within the mud itself is probably the most important issue. An ideal model would freely permit caverns to form if the consistency is right.
I can see such experiments going on for quite a while, employing various combinations at different levels. Could be a lot of fun if you set it up right. But if you're just out to get a quick answer from a short-term project, I think you'll be disappointed. If you don't think it'll be a fun project, and worth spending a few years, I wouldn't bother.
You also need to recall that the "top" of the Grand Canyon wasn't the top surface when the water began to flow. Higher layers were there, but they were almost entirely washed away. Their remnants are the features which match the results I obtained when my dams were washing away.
Scaling won't be much of an issue for you, except when it comes to water velocity. I'd recommend controlling the volume and tilt, and letting the water do its own thing.