Monday, January 11, 2010

Fountain Advice

So, if I were to design another fountain, how would I do it differently? (This is for you, Diane.)
  • 8 inch flow straighteners. There is no sense in messing around with Reynold's numbers. You want it to be around 2000, and that means you need huge internal cross section. 8 inch PVC is actually reasonably easy to get. SCP in Santa Clara has it and all the fittings you need.
  • Design gimballed nozzles from the beginning. Trying to get this right with careful assembly just did not work, the tolerances are far too tight. I think it can be done with screws between the gimballing nozzle and the glued-in support, so that you could adjust the angle after building it.
  • Rebar connecting the inside and outside rebar curtains. We have a crack all the way around our hot tub which I'm sure is going to make it's way through the tile some day.
  • I would build all the fountain jets to a fixture, rather than just half of them. The fixture worked really well and would have worked even better if I'd designed it to be independent of swelling due to moisture. This can be done -- all the surfaces that locate plumbing have to be on radial lines to the center.
  • Make the fixture locate a center #4 rebar spike, and drive the center spike at least 18" into the ground, and leave it in place while shooting the gunite. This will give the gunite crews something to locate off when they are drawing their circles. One problem that we had was that we kept re-finding the center of the hot tub, and as a result the circles for the plumbing and the circles for the tile and gunite are not concentric.
  • I would have the guys doing the gunite get the gunite surface level and ready for tiling. We spent a lot of time levelling that gunite out.
There were also a couple of things we've learned about the tiling, which would have saved us a bunch of time had we known it a year ago.

The glass tile is 1/4" thick, and can be set with just 1/8" of thinset, but this isn't the stackup you want. There are two problems: the thinset shrinks, a lot, as it cures, and this tends to bend and eventually break the tile. Also, the plaster guys want to be 5/8" thick, not 3/8" thick. They can feather down to 3/8", but they don't like it, as explained below.

So, you want to put down on the gunite 1/4" of some kind of low-shrinkage mortar, and screed it so that it is flat. This is the stuff that is going to take out all the uneveness in the wall. However, you can't be sure it's going to stick really well to the gunite. So, the stackup we used under the last mosaics ended up being:
  • 2 coats of Hydroban, sticking out at least 1" past any of the rest of the stack. This forms a structural watertight barrier. The principle issue being protected against is water leaking through the cold joint between the plaster and the tile, and then leaking into the gunite from there. Hydroban is expensive, but only comes in 5 gallon buckets, which is enough to cover a ridiculously large area. I think they are trying to make sure you use a lot.
  • Some thinset, as thin as it can be, to adhere the mason mix to the hydroban.
  • 1/4" mason mix (either "deck mud" for the tile on the pool bottom, or "fat mud" for the tile on the walls), screed to be dead flat.
  • After that all sets up (ideally about 4 hours so it gets a chance to shrink, but isn't fully hard, think about covering it with plastic to keep the water in), a super-thin layer of thinset on both the mason mix and the back of the tile.
  • Blue tape out the wazoo for anything on a wall.
  • Then cut the stiff but not yet really hard mason mix away from the edge of the tile. We never let the mason mix set up so hard that it was sticking to the hydroban tightly, so this was pretty easy to do without nicking the hydroban.
On our dam wall, we had to build up the wall top with an angle to hold our teflon strip. The easy way to do this would have been to just do it with wall mud, screed with one of those adjustable angle things running along a strip screwed to the side of the wall.

Now back to the reason the plaster guys want to put down 5/8" material. For any kind of exposed aggregate surface, they shoot the mix, then trowel it. The troweling is usually done (on a smooth plaster finish) to bring the "cream" to the surface, but for an exposed aggregate surface they are using the side effect, which is to compact the aggregate below the surface. Then, when they wash away the surface, they expose more tightly-packed aggregate. Since cement but not aggregate will erode from chlorine attack and mechanical erosion, it's better to have a surface which has more aggregate.

The reason they don't simply mix more aggregate into the mix before they shoot it on the wall is that, when the aggregate and sand grains are randomly oriented, they simply can't pack densely enough. If you subtract sand/cement/water, you don't end up with more aggregate in the as-shot mix, you end up with more air. And air is no good because the plaster has to be watertight (and dense and strong). Once the mix is on the wall, they work it with the trowels, which helps settle the aggregate and sand grains together more tightly. This is why the quality of an exposed aggregate surface has a lot to do with the skill of the guys doing it.

In looking at the samples provided, it's clear to me that most folks aren't actually attempting to get a maximum packing of aggregate, and I don't understand why. Why don't they mix larger and smaller aggregate together, so that the small aggregate packs in the spaces between the larger aggregate?