Calculating seal pressure 1

[kwilli19]

I'm building an enclosure for some electronic equipment to take it underwater. Nothing complicated, a 6" cylinder with flat end caps bolted over a neoprene gasket. Is there any rule for calculating the the amount of clamping force needed on the enclosure seals to keep out the water? For example, If the end cap seals have 50 lbs of force on them, how much water pressure will that keep out?

 

[IRstuff]

The requirements dictate the design, not the other way around, so how much pressure do you need to keep out?

TTFN

 

[kwilli19]

We'll be diving to a max of 100 feet, so I believe that's around 45 lbs/sq in.

 

[IRstuff]

So you'll have 45 psi * pi *(3in)^2 = 1272 pounds of foce on the endcaps

TTFN

 

[gwolf]

If I can add a little wrt IRstuff. What IRstuff is alluding to is that you should use the water pressure to compress any seal rather than worrying about any pre-load.

The usual approach with this sort of seal is to use the force from the water pressure on the cylinder end-plate to force a seal profile against the cylinder (either radially or axially). If the seal/housing contact pressure exceeds the local water pressure then no water will come in - voila.

It is common in such circumstances to have a V type seal and angled abutment. The abutment angle on the seal works like a wedge or lever and dramatically increases the contact force between seal and cylinder, thus giving a higher seal/cylinder contact pressure for a given cylinder end-load.

 

[kwilli19]

The water pressure at 100 feet is 58 lbs/sq in. So why wouldn't I match that pressure with the pressure the bolts put on the seal. The seal bearing area between the cylinder (1/4 in wall thickness) and end cap is about 4.7 sq in. Multiplied by 58 and that's 272 lbs. So I torque the bolts to put minimally 272 lbs clamping force on the seal.

 

[gwolf]

As IRstuff says, you have 1272 lbs pressing the end plate onto the seal. The bolts are just there to stop the end plate falling off in free air or shallow water.

 

[kwilli19]

OK. So I'm basically limited by the strength of the materials, ie crush depth. Either way, I'll still need a good bit of testing before I put anything expensive inside of it.

 

[gwolf]

Yes that's right. If you want to run some basic design data past me like materials, diameters, thicknesses and such I would be happy to do some guideline calcs for you, they don't take long.

You should also select a good seal, probably off the net, suitable for such an application.

Also beware if you need to have any external switches or controls with wires coming out of the main enclosure. It is here that the water is most likely to get in and you would need a purpose-built (but probably still off the rack)switch interface to deal with this.

 

[kwilli19]

Sure, see what you think. The end caps will be 1/2" lexan, 6 1/4" square. The body will be 6" OD plastic drain pipe, 1/8" wall thickness, 7" long. I was planning on cutting the sealing washer myself out of 1/4" neoprene, 50 duro. The bolts go through the corners of the lexan, end cap to end cap, outside the body. No other holes will go though the enclosure.

 

[gwolf]

OK, leave it with me for 24 hours. The static strength looks fine but I need to check some buckling equations at work.

 

[gwolf]

kwilli19,

OK, I did some basic calcs using some equations that I am not too familiar with - ie test what I suggest first without anything expensive inside. Also make sure that you bang the sides walls of the tube gently at depth to double check for buckling instability under rough use.

Assuming your pipe is made of some sort of Nylon and the end plates are lexan I predict that the plastic dran pipe walls will buckle with the pressure but the end plates will be OK (some comfort :-( )

I suggest using a minimum of 3/8" thickness pipe, preferably 1/2" to give you some margin for rough use.

Anybody else like to comment on my sizing suggestions?

Regards,

Gwolf.

 

[kwilli19]

Yes, I'm looking at cast acrylic tube with a 1/4" wall. No doubt stronger and clear too. Thanks for your input.

 

[IRstuff]

Seems plausible, although some bracing might make the original tube sizing work.

TTFN

 

[gwolf]

If you attach big flanges on the ends of the tube then 1/4" nylon would work. No idea how stiff cast acrylic is but it smells stiffer than nylon so maybe 1/4" acrylic will work.

 

[kwilli19]

An update: I've got all the materials cut and ready for assembly and testing. Changed the end caps to hexagonal and put stainless rods through each corner, still using 1/2" Lexan. Gasket will be .070 neoprene, the 1/4" one was unneccessarily thick. The 1/4" wall, 6" OD cast acrylic tube for the body turns out to be quite strong, I loaded 155 lbs sideways on a 3 1/2" length piece of it and it barely flexed, although it is brittle compared to lexan. I found formulas, discussions and data around the web for reference and from this I estimate the crush depth for this vessel to be between 500 and 800 feet, so my original goal of 100 feet should be well within limits.

 

[gwolf]

Great, let us know when you have taken it down.

 

[kwilli19]

Initial test in the sink OK, no leaks. Must experiment next with bolt preload, gasket material doesn't compress that easily. I tightened the bolts until the gaskets were visibly compressed, which bent the corners of the end caps. Might not need that much pressure. Ends of the hull cylinder were wet sanded down to 1200 grit.

 

[arto]

Look in the ASME B&PV Code Sec.VIII Div.1 Appendix 2 for Gasket Seating Load Design, etc.

p.s. O-Rings are "self Energizing" - don't need bolt force to establish initial seal

 

[kwilli19]

True. I'm used to internal pressures, seeing compressed gaskets, got to get that out of my head.

But, I also don't want further compression of the seal from the water pressure to let the bolts get loose, since the bolts will also be used to attach handles, legs, lights etc. I think there's a way around that though if I add neoprene washers under the bolt heads, so they'll expand as the gasket is comoressed, keeping it all tight.