I have the following situation: A pair of identical towers, 30 meters high, have 20” overhead lines dropping a short distance down the sides of the towers then running horizontally into a structure. For various reasons the lines are not supported from the towers so the first available support points are at the structure beams.

Inside the structure the lines join at a tee. A single 20” line then continues from the tee to an exchanger. The towers operate alternately; i.e. tower ‘A’ is hot while tower ‘B’ is cold and vice versa. Block valves either side of the tee isolate the cold tower from the process.

I modeled the two temperature cases as follows:
Tower ‘A’ to the tee: T1 = cold, T2 = hot.
Tower ‘B’ to the tee: T1 = hot, T2 = cold.
From the tee to the exchanger: T1 and T2 both hot.

The boundaries of the system are the flanges at the tower nozzles and the flange at the exchanger. All displacements were calculated manually and applied at the boundary points. Zero displacements are applied to the towers when cold.

The designer added good-sized loops in the system. However, using +Y supports only, the hot tower sees a fairly high –Y force and a very high moment load. The solution, I believe, is the addition of at least two springs located somewhere along each of the overhead lines inside the structure.

When I add the two springs at best-guess locations the CAESAR II recommended load cases only define a combination for D1, T1. i.e.
L1: W (HGR)
L2: W+D1+T1+P1 (HGR)
Thereafter, CAESAR II provides OPE and EXP cases for both conditions.

Questions:
1. What exactly does the L2 case do? (The answer, I assume, is obvious: Springs are calculated for the case where tower ‘A’ is cold and tower ‘B’ is hot. The alternate case is not considered.)
2. Is it correct to insert a new load case as L3: W+D2+T2+P2 (HGR)?
3. If the answer to 2 is no, must I make two separate runs changing D1, T1 to D2, T2 in the second run?
4. If the answer to 3 is correct, what is the best method of determining the actual spring settings required?
5. As a possible answer to 4, does it make sense to base the spring settings on the entire system running hot?

Thanks, as always, for any suggestions offered.