Ripped this off another forum. Found it of value so it's being passed on.

"Peter G. Engle, PE" <Pete@ALMOSTHOME.COM> wrote:

Yeah, that one really got me screwed up when we did it in college thermodynamics. But here goes:

Insulation acts in a linear fashion for heat transfer through the material. In math terms, H=kxT, or heat loss equals the thermal constant (U value) times the thickness of the insulation times the Temperature difference between inside and outside. More simply – if you double the thickness of the insulation, you cut the heat loss in half. This, we pretty much all know intuitively. But it’s only true for flat surfaces. There’s a lot of simplification that goes on to get to that simple formula.

When you start insulating cylinders, you can’t ignore the effect of curvature. If the insulation is thin, the effects of curvature are small, and the formula above works fine. But as the insulation gets thick, the surface area of the outside increases with the square of the thickness while the surface area of the inside stays constant. Effectively, the heat has more “pathways” to escape. And since these pathways increase in a geometric way while the thickness only increases in a linear way, there is a crossover point where the heat loss actually begins to increase with thickness. It turns out that for any given set of inside and outside temperatures, insulation factors, and cylinder geometry, there’s an optimum insulation thickness where heat loss is minimized. If you apply any more or less, the heat loss increases. It really works this way. We worked through the math, and it really does happen. Go figure.

What’s pretty cool is that the water heater manufacturers aren’t dumb – they study basic thermodynamics, too. They already insulate water heaters to the optimum level for the type of insulation used. To get more insulating value in modern water heaters, they haven’t increased the thickness, they’ve changed the insulation. Insulation with a higher R-value gives you more insulation in less thickness, and the magic crossover point moves to a lower total heat loss level, even with reduced insulation thickness. The ideal insulation would have infinite R-value, so the thickness could approach zero, and still have zero heat loss.

But I digress. The bottom line is that with water heaters, if you go slapping on a loosely fitting blanket of fiberglass, chances are that you will actually increase the heat loss. Pretty neat, huh?

Peter G. Engle, PE

Almost Home, Inc.

www.almosthome.com

Replies:

Here in the Water Wonderland of Michigan, the utility companies actually paid people to go door to door and install water heater insulation blankets. As the blanket insulated and held heat on the gas control valve, the "rubber" seals got hotter than designed for and started leaking.

We have found gas leaks on abt 90% of insulated gas water heaters at the top function knob; on - off - pilot knob. I think the valve was designed to be a little above room temp and adding insulation too close raised that temp 20 degrees or so. As a former slum lord it was not uncommon for me to spray the knob with WD-40 and stop the leak, for a couple months.

I like Peter's explaination and it makes sense.

It's never made sense to me to insulate a gas water heater unless it's a direct vent type and then only the tank should be insulated not the control valve.

In order for a gas water heater to work properly, you have to leave off enough insulation blanketing around the draft diverter to allow air in, you can't cover the control valve, the access cover for the burner plate or the air intakes on the underside of the tank or on the sides. Once you've left that much of the tank uninsulated, there's not much point in trying to wrap it anyway.

My point to homeowners is that the tank manufacturers have figured out how they can achieve maximum energy savings with the way they've insulated their tanks. Adding more insulation might give one a warm and fuzzy but it's probably money wasted.