The easiest way to layout the PEX collector is to use a serpentine pattern of PEX pipe starting from the top and extending to the bottom of the collector. Each straight run of the serpentine pipe layout is sloped slightly down to encourage water to drain completely from the collector when the pump stops. The collector is protected from damage that would result from water freezing in the collector by making sure that it always drains when the pump is not running.
I was a bit concerned that the drainback process might not be reliable enough for a serpentine collector to insure a complete drainback in all cases. In use of the prototype over a couple months, I did not encounter any failures of the water to drain back into the tank, but I was still concerned that the drain back might not be complete under some circumstances. I did the test described below so that I could actually observe the water draining back from the collector and see how robust and repeatable the drainback process is and to observe what happens when collector tilt and/or slope are changed in ways that might interfere with reliable drainback.
I built a section of serpentine collector using transparent vinyl tubing so that I could actually see the fluid draining. The pictures below show the setup and show the drain back process.
The picture below shows the setup. The water was pumped from the orange bucket to the bottom of the serpentine collector tubing. The line from the top of the serpentine collector drains water back to the bucket -- this line must terminate above the water line in the bucket.
Some ink was used in the water to
make it a little more visible.
This shows the test setup. Water is pumped from the orange bucket up through the transparent serpentine tubes, and then drains back to the bucket via the line from the top of the serpentine tubes. Each of the straight runs is sloped slightly down -- about 0.5 inch over the width of the collector.
For most of the tests, a little bit of detergent was added to the water to reduce the surface tension and (hopefully) improve the drainback. The amount used was about one drop of dish washing detergent in the full bucket of water.
When the pump is stopped, the drain back process starts with air advancing up the drain tube that goes to the top of the collector. The air-water interface advances up the drain tube, and then down each of the serpentine runs until it gets to bucket. This advance of the air-water interface happens fairly rapidly (less than 30 seconds for the full setup). I repeated the process many times, and it appears to be very repeatable and robust. In the 2nd phase of the drainback process, the water that remains after the air-water interface passes drains down along the bottom of the tube section. The first phase drains most of the water and proceeds rapidly. The 2nd phase drains the remaining water, and takes more time. At the end of the 2nd phase, there is very little water left in the collector.
The air-water interface is visible progressing along the upper tubing run from left to right. This interface advances rapidly down the tubing. Most of the water drains during this phase. The water that remains after the air - water interface passes drains more drains more slowly.
I did several tests in which the one edge of the bottom of the collector was propped up such that some of the lines did not have a down slope, or even had an up slope. I don't see any reason to subject your collector to this kind of abuse, but I just wanted to see how sensitive the drain back process is to not having a good slope.
With the collector adjusted to make half of the straight runs level rather than down slopping, the first part of the drain back process (the advance of the air-water interface) was essentially unchanged, and this phase drained most of the water out of the tubing. The 2nd part of the drain back process in which the remaining water drains out resulted in some small pockets of water left in the tubing at the end of the straight runs (see pictures). These remaining water deposits were small and in no case blocked the full diameter of the tubing.
Its hard to see in the picture, but there is just a bit of water left in the tube at the start of this straight run that has no down slope.
It would not (I think) be enough to cause any sort of freezing problem.
I did some more runs in which half of the straight runs actually slopped upward. These actually drained surprisingly well. The first phase was pretty much unchanged. The air-water interface did not seem to mind advancing slightly uphill at all. The 2nd phase resulted in larger collections of water at the beginning of the upward slopping tubing sections. In some cases, there was enough water to fill the full diameter of the tube for short sections (see picture below).
This shows the deposit of water remaining at the start of an upward slopping tubing run after both phase 1 and phase 2 had completed. If this were to freeze, I doubt that it would cause an problems, as there is room from expansion of the ice into the air pockets in both directions.
While its reassuring that the collector still drains pretty well even without the down slope, I would strongly encourage you to make sure that the collector is built with a down slope on every single run.
I also tried a test in which the tilt of the collector was reduced to about 20 degrees to see if this impacted the drainback process.
The low tilt collector still drained reliably, but took a little longer.
This shows the setup for the shallow tilt drain back test. Although the drain back time was a little longer, the process occurred in the same way as for the higher tilt cases, and appeared to be equally predictable.
Gary August 31, 2008