I am working on the seismic analysis of a 2"NPS warmup line (bypass) around the HP steam stop valve for a combined cycle plant. This line has 180mm (7") of mineral wool insulation, and the system geometry is more or less an elevated double expansion loop (up-east-north-west-north-east-down) because of its location. The two connections are about 3.5m apart, the risers are 0.9m, max easting is about +600mm and -800mm. There is a 50kg hand valve in the southern part of the loop.

I have two constant supports on this, one at the south end of the eastern N-S leg and the other at the north end of the western N-S leg. All is fine in sustained and operating load cases, but when I hit this with the equivalent static load appropriate for the project, I am overstressed at the southern connection by about 2% even after applyign the 120% factor for occasional loads.

However, the way we model insulation is to ignore its structural contribution. Here, where the weight of the (mineral wool with binder and aluminum jacket) insulation is 120% of the weight of the pipe on a per unit length basis, and the runs are so short, I am prepared to argue in my analysis report that the insulation here, rather than contributing additional load to the seismic analysis, will in reality provide damping to the line.

Any comments?

I personally would never consider the 'stiffness' of the insulation system, just too dicey for me. But there are other assumptions that you could challenge:

1. Caesar calculates the forces and moments uncorroded, but the section modulus is calculated corroded. Ask yourself if a steam system is going to corrode uniformly or locally, and I believe you can change this setting in the config.

2. You can do an FEA of the tee's and calculate a more accurate SIF (remember it's the ratio of cycles to failure, so you have to calculate the cycles to failure in the FEA). Getting accurate drawings of the tees can be challenging though.

3. Can you move the valve further away from the middle? This should reduce the bending moments.

4. This one is cheesy, but sometimes tweaking the spring loads can make a pesky 2% overstress go away.

Thanks, Paul.

1. is a good point - no point in calculating forces & moments generated by seismic acceleration fo material that has been corroded away.

2. It's a weldolet.

3. Can't move the valve - it's in the HRSG vendor's scope.

4. I have tweaked that SOB until it wept.

Final point - I am NOT attempting to take advantage of the stiffness of the insulation. I AM prepared to challenge the argument that, in a seismic situation, the insulation will follow the movement of the pipe at all.

For a 0.9m run of 2" pipe with 180mm insulation, the OD of the insulation package is 420mm. So the aspect ratio (L/D) of the insulation is 900/420, ~ 2.2. The aspect ratio of the pipe is 15. They are going to behave much differently when subjected to random seismic excitation forces. But the routines in CAESAR II assume that the insulation is an infinite series of small disks firmly attached to the pipe. It just isn't so.

Even for the 2.5m horizontal run, the insulation will react independently of the pipe. After all, it's not glued to the pipe, it's just resting on it. When the pipe moves, the insulation will (a) probably not move up and down much (at least in response to driving displacements of the pipe), (b) probably move axially only so far as to accommodate the movements of the ends or its own mass:stiffness ratio, and (c) move laterally only according to its own whim or when the deflection of the pipe is sufficient to make contact. Don't forget, the ID of the insulation does not necessarily match the OD of the pipe.

While I wouldn't for a second consider adding the stifffness of the insulation to the stiffness CAESAR II calculates for my pipe in order to reduce my piping loads, I do believe that the insulation package will be stiff enough (in this geometry, at least) that it will react to a seismic event with at least some independence from the motion of the pipe. This is a definition of one characteristic operating mode of a damper.