All your points about frame design are good points bob and the best course of action to follow.
But the math isn't really that difficult for someone who does it everyday. Unfortunately, I don't do it every day, but I've been down this path a couple of times before.... pity it was years ago now.
- You've got known exciting frequencies and a fairly well guessable excitation force.
- You've a known mass of the engine and frame.
- You've a desired maximum amount of motion in the frame
- You've presumably want the minimal amount of force transmitted to the floor.
All those can be put together to get some mounts that have the right characteristics straight up, with minimal tuning and tweaking afterwards. It at least allows you to separate the wheat from the chaff when you're trying to select a mount.
The thing to remember is that for less vibration on the outside of the mounts, there's going to have to be more vibration of the engine. The stiffer the mounts, the less engine vibration, be the more vibration is passed through them to the floor.
Have a look at this chart :
(A bigger version is here :
http://listerengine.com/coppermine/albums/userpics/10075/NaturalFreqIsolation.jpg )
Pick the wrong rubber mount, and worst-case, you'll be at resonance and the engine frame will tear itself away from the ground. Too stiff, you might as well just solid mount it. For proper isolation - of the type that reduces the forces on the ground, you need a mount that has a low natural frequency.
Now, have a look at this airmount :
It's also at :
http://listerengine.com/coppermine/albums/userpics/10075/1X84D_Airmount.jpg , so you can read the numbers.
It's natural frequency at 40PSI is about 1.35Hz. It's carrying capacity at that pressure is about 300 pounds, so four of them will carry a 6/1 + genset + a bit extra mass fine. Looking at that chart further up , with a forced frequency of 5Hz and a natural frequency of 1.35Hz, the transmitted force through the mount is about 10%. If we have a reciprocating force (after a 50% balance factor) equivalent to a hundred pounds of force at 5Hz, that particular mount reduces that thumping on your thin concrete slab to 10 pounds. For 10Hz flywheel imbalances, it's up around 97 percent isolation.
I've thrown together some cad drawings for the frame - basically as you describe bob, but a little longer than a normal compact frame, to allow for some extra mass to help damp it a bit and also to allow me some room to possibly mount an A/C compressor or DC alternator near the generator later on. Consists of twin 3" box rails, separated by 6 3" tubes, one on each end, two at the spacing for the engine and generator. This frame will be plated with something like 1/4" steel. It's reasonably wide (wider than the flywheels) and the base is reasonably long (to allow for risers and gussets/strengthening on each end to mount the 8" high airmount supports at crankshaft height)
A question with the top plating - does it need to be welded all round? Or will solidly clamped via bolts do? Just wondering about access for tuning with sand, etc.
Those times I was down this path before? It was when I was working in a coal analysing lab on site at an open-cut mine. The lab was three floors up, in a small room.... attached to a 7 floor washplant full of screens and centrifuges and pumps , all merrily vibrating at anything from 300Hz up to a few kilohertz. Every now and then, they'd all get into sync and the whole place would shake enough to spill your coffee! I had a balance that read to 1/10000th of a gram - (0.0001g) and I needed that accuracy. Weighing things with the washplant off was fine. Weighing things with the washplant on resulted in variations of about 50 milligrams, which was bigger than some of the things I was trying to weigh! Eventually after mucking around with lead sheets, rubber mounts, springs, etc, we went with wire-rope isolators under the balance, which solved it except for when everything got in sync. Seeing as when it did that it was about one big *thump* every second, there wasn't really much we could do there. For all the other time though, it went from a 50 milligram jiggle down to a 0.3 milligram jiggle... which improved the results we got no end.
All the frequencies are the same then as they are in our challenge now - in fact some were lower (and therefore worse).
I used the same methodology then as I'm doing now, except then I had the advantage of a hired vibration analyser widget which highlighted all the frequency peaks. As we (should!) only have two large peaks at 5/10Hz, the problem here is a bit simpler.