Re: Bilge pump non return valve.
I work with centrifugal pump design every day. What's missing here is our basic tool, the pump curve supplied by the manufacturer. A centrifugal pump always runs on the pump curve, for every head condition along the curve there is a corresponding flow rate. The left end of the curve is the maximum head at zero flow, called shutoff. If you look at a pump curve you'll gain a whole new appreciation for the difficulty in applying centrifugal pumps.
You can learn a whole lot about your bilge pump by putting an open hose on it and running it in a shallow pan. Start with the open end down low and watch how much flow you get. Slowly raise the open end and watch how fast the flow rate falls off. At the height water stops coming out of the hose you have reached shutoff head. Measure the flow into a bucket and draw the curve for the pump. Some pumps have relatively flat curves downward to the right. A small amount of additional head pushes the pump back on the curve to the left and flow falls off very quickly. I attached a pump curve just to show what they include, curves for different impellors, efficiency points, horsepower curves etc.
I see here talk about having a loop well above the heeled water line. The fact is you want the loop as little above the waterline as will gaurantee you won't back flood when heeled. Any more greatly reduces the flow that you can actually pump. You might notice they rate bilge pump flow at about 2 feet of head. Lifting water 3 feet can cut the flow in half or more, and you might not get any flow at 4 feet!
There is also talk about venting the loop, to prevent a back siphon. This is true, however a forward siphon from a loop well above the water line to a discharge point above the water line actually will increase the pump flow a lot. So a tall vented loop can defeat your pump as well. When you do the test above trying raising the middle of the hose way up in the air with the open end near the ground. You'll see that the pump will move a lot of water way up high as long as the siphon on the downhill side pulls the water along. Drill a tiny hole in the top of that loop and the flow will fall way off.
A really important point missed here is discharge hose size on the pump. When you have only 2 feet of head to work with you really must avoid any friction losses. So jumping a 3/4" pump discharge up to 1" or even 1-1/4" can make a big difference, especially if it must run more than a couple of feet.
The comment about a small hole in the hose right at the pump discharge is a good one. In the instructions no one reads for commercial submersible pumps there is always a statement about drilling a 1/8" hole to prevent air locking. A bubble of air forms in the pump housing and the impeller can't grab enough water to build pressure enough to push water up the pipe. In a 44 story building in Manhattan I had to pull two pumps out of 18 foot deep tanks to drill the two holes that no one thought was important because those pumps air locked all the time!