See the Change Log for other minor fixes.  This table lists major fixes which may have affected results in earlier versions.

Non-convergence at any analysis stage will halt the analysis and output will only be available up to the failed stage.

If the lack of convergence is in Stage 0, this is often due to the application of a UDL at ground level.  This creates a numerical anomaly at the top node, because in Stage 0 the wall has not yet been inserted.  Try moving the load to a level between the top and second node.

For later stages, try increasing the number of iterations, if the iteration progress messages indicate that the errors are reducing.  The number of iterations can be edited in the Convergence Control dialog.

The reason for this is explained in the Help file topic :

Detailed Processes in Frew -> Creep and Relaxation

You should apply wall relaxation instead of manually changing the wall EI and this will correctly model the change from short to long term wall stiffness, and you’ll see the effect on the moments and deflections.

You just need to copy the bitmap to the Windows clipboard first.  Once it’s there, Paste will put it on the titles dialog.

If the wall elements are very short and stiff, the program will not compute the stresses / bending moments accurately from the displacements. The problem can usually be overcome by increasing the distance between nodes. Also please make sure that spacing between the nodes does not vary greatly(max spacing to min spacing ratio should be typically limited to  around 2).

This problem was exhibited In one example with a spacing of 0.2 m between most of the adjacent nodes, with occasional spacing of 0.1 m. If the spacing was increased to around 1 m, this problem does not seem to occur.

This problem was particularly severe in pre-17.8.6 version of Frew (and DOS Frew) which had lower precision calculations.

In the undrained stages – drop the “data” pore pressures completely out of the undrained material – if they are absent, the effective stress check should be bypassed.

Frew 19.0 and later versions deal with this automatically and this error should no longer occur.

• Neglecting or simplifying the likely effects of wall friction
• Simplification of the action of structural supports (by treating them as springs fixed at one end) Very early versions of Frew considered only a “smooth” wall, which would tend to overpredict displacements.
• Consideration of vertical shear on the wall was added in the early 1990s and significantly improved the results when compared with finite element analyses.

• In the initial stages the ground may exhibit higher stiffnesses which are not represented by the linear elastic stiffnesses used in the program-
• The ground may be partially saturated and have its strength and stiffness enhanced by suction It is difficult to give an estimate of the amount of over-prediction, given the variability of each individual site and local conditions.

This can be found in the Frew help manual under topic Method of Analysis-> Total and Effective Stress -> Undrained to Drained Behaviour  and Undrained to Drained Example.

For automatic calculation of undrained pore pressures, specify a material from which to obtain effective stress parameters for each undrained material.

 Or to effect a manual switch from undrained to drained behavior without using Frew to calculate the undrained pore pressures automatically refer to the help topic:

Detailed Processes in Frew -> Undrained to drained behaviour – Manual Process

If you want to model pre-stress alone, then the stiffness is set to zero. In this case, the strut force will not be affected by the wall movement. If the stiffness is also entered, then the strut force will change due to wall movements.