3.7. Solvers

3.7.1. Static

Go through all loadcases, solve each one and then combine them to generate the final results.

For each element calculate the element stiffness matrix and assemble the system stiffness matrix using the nodal degree of freedom. The DOFs of a node is based on the position/index of the point in the points list

The loads defined for each element primary/primitive loadcase is combined into a single load vector. The load vector for each element loadcase is then assembled into a global load matrix. Multiple global load vectors are solved for at once using gaussian elimination.

Note that to simplify the FEA solution, all elements are simple beams, including bends and reducers which are approximated by multiple beams chained together.

3.7.1.1. General Procedure

  1. Define NDOF indices for element nodes.

  2. For each element

    1. Construct the local stiffness matrix.

    2. Construct local force vector, one for each primitive loadcase.

    3. Transform local stiffness and force matrix.

    4. Add global element stiffness matrix and the global element force vectors to the global system stiffness and forces matrix, respectively.

  3. Apply the boundary conditions using the penalty method.

  4. Solve the global system using guassian elimination techniques.

    AX=B

    Where A is the global system matrix.

    B is the matrix of force vectors, one for each primitive loadcase, and X is the matrix of solutions vectors for each primitive loadcase.

  5. Use the calculated displacements for each primitive loadcase to calculate the element force and moments.

  6. Use the results from step 5 to calculate nodal reactions and element forces. Finally use the element forces to calculate code stresses.

3.7.1.2. Reducers and Bends

Both reducers and bends are topologically curves and therefore approximations. The midpoint of a bend is created as a side effect of defining the bend when it is created. All the other vertices are inaccessible from a nodal standpoint. In other words, the user does not have control of the other underlying vertices because only the vertex that corresponds to the midnode is referenced by the midpoint.

The static solver preprocesses all reducers and bends such that a point is made for each underlying vertex at runtime. Once the solution is generated all the temporary points are deleted along with the nodal data corresponding to each point. For the case of a bend, the solution for the midpoint and end points are kept. For the reducer, the results for the end points are kept only similar to all the other element.

3.7.1.3. Tee Flexibility

Under Construction!

3.7.1.4. Skewed Supports

Implemented using local stiffness transformation.

3.7.1.5. Spring Algorithm

The program tries to place a rigid support, variable or constant spring where a spring support is specified by the user. The algorithm must satisfy all the operating cases such that it does not bottom or top out. For each operating case a linear deadweight analysis (W+P+D+F) is performed and a +Y support is placed at each supports. Remove the force from the weight case at each spring support location with the equivalent upward force and run the operating case with the temperature. Determine the upward movement of the pipe at the spring support. Use the applied force (i.e. the hot load) and the movement to pick a hanger from the manufacturer catalog.

3.7.1.6. Master Slave (CNode)

Constraint equations.

3.7.1.7. Non-Linear Supports

For each loadcase, for each nonlinear support, per each solve:

Initially a non-linear (+Y) support is converted to a linear full Y support. If the solution shows a (-Y) load on the support, the support is modeled correctly and “active” for the loadcase. If not, the program marks it as “inactive” and gets rid of the support altogether and tracks the displacement at that point. If the point shows a (+Y) deflection, this support is modeled correclty for this loadcase. Else, the stiffness matrix is reset to a full Y. If any of the nonlinear assumption proves incorrect reanalyze with the updated stiffness and force vector (if required). Continue the process checking all the “active” supports for loads and all the “inactive” supports for displacement. When all “active” supports have a (-Y) and the “inactive” suports show a positive up displacement. A status change flag variable is used to indicate a support that that initially shows a (-Y) load but then starts to show an uplift.

Note

A static nonlinear analysis is performed by iterating until the solution converges.

3.7.1.8. Friction

Under Construction!

Note

A static nonlinear analysis is performed by iterating until the solution converges.

3.7.1.9. Loading Sequence and Non-Linear Supports

Non-linear supports use an iterative approach to determine the final support loads and pipe configuration. The sequence in which loads are applied matters as superposition is not valid for non-linear analysis. Each load is applied and the displacements extracted. These movements are then used for the next step, ie. the model mesh is modified to incorporate the new point locations. As a result the system stiffness matrix is updated each iteration.

3.7.3. Harmonic

Under Construction!

3.7.4. Spectra

Under Construction!

3.7.5. Time History

Under Construction!