Using the post-processor

The post-processor, FELVUE, is started by clicking on its icon.

Within the post-processor, you will be able to change directory to your user directory, and then a list of output files with .out filename extensions will be displayed. Choose an output file, and the post-processor will first read the mesh data, and display this as in the pre-processor. Click on Continue, to proceed to a display of the results.

Note that the main engines' output the results (displacements, stresses, potentials, etc.) in E10.3 format, so that very large numbers can be handled, and reasonable accuracy obtained with very small ones. However, if you know that all your stresses will be of a certain order of magnitude such as 10-100, you may wish to edit the format statement to output then in F10.5 format, say. FELVUE reads real numbers from the data-file in G10.3 format, so that it will accept numbers to 5 decimal places, and these will then be output to this accuracy in the pop-up windows described below.

It is possible to produce plots which only show elements of a specified material type (LMTYPE). If you want all elements plotted, leave LMTYPE=0 in the dialogue box which offers this facility. The following sections now describe the plots available from the Main Menu.

Scalar-variable meshes

In the case of a scalar-variable mesh, the data read from the .OUT output file comprises:
• nodal values of the scalar potential (e.g. nodal temperatures);
• flow rates in x and y directions at element Gauss points.
The choice of plots (all of which can be zoomed) is:

Yield zones

These are described in the elasticity-plot section below.

Flow contours

The horizontal and vertical flow rates q_x, q_y can be plotted as contours, either using coloured contour lines or with colour fill-in. You can plot either horizontal or vertical flow components, or plot contours of the flow magnitude \sqrt{q_x^2 + q_y^2}. Note that if the mesh is of linear triangles, there is only one Gauss point per element, so the plot will colour the whole of each element in the appropriate colour; the lines'-type plot will not work in this case.

Temperatures

This is the main plot, of nodal potentials U. A contour plot of the potential field, interpolated from the nodal values, will be displayed. When a node is picked using the mouse, the nodal coordinates and nodal value of U will be displayed in the dialogue box.

Flowlines

This is a more useful form of plot of the flow-rates q_x, q_y, as flow-lines at the Gauss-points, showing the direction and magnitude of flow, the latter relative to a reference value which the user can enter (default = 100.0). Decreasing the reference value, increases the lengths of the flowlines. With the Pick Gauss point' menu option, the user can select a Gauss-point by clicking on it, and a dialogue box will advise the flow-rates at that point.

Vector-variable meshes

For elasticity-type problems, the data read in is:
• nodal values of the x and y displacements;
• stresses at element Gauss points.
The choice of plots (all of which can be zoomed) is:

Displacements

There are three ways in which the displacement field (u,v) can be viewed:

Deformed mesh

The mesh is drawn with the nodes displaced from their original coordinates, by the displacements solved for in asdis, multiplied by an exaggeration factor. The initial exaggeration factor is calculated so as to achieve a 10you can then re-draw with a greater or smaller exaggeration as required. If you pick a node, its displacement components will be reported in the pop-up window.

Displacement contours

This is a contour plot of the displacement magnitude \sqrt{u^2 + v^2}

Nodal vectors

This shows the nodal displacements (u,v) as vectors from the node positions. This is a more meaningful plot for analyses where u,v do not represent displacements but velocities, etc. By picking a node, the values of u, v will be displayed.

Plot 1D elements

If the post-processor detects any one-dimensional line elements in the mesh, it will ask if these elements represent beams. If you answer y', the beam element shape functions derived in Chapter 5, together with the nodal displacements and rotations, will be used in plotting these elements. This is the case both for the standard 2-noded cubic beam elements, and the higher-order 3-noded quartic beam element described in Chapter 5. If you answer n', the standard 1D linear or quadratic shape functions will be used with the nodal displacements; this should be used when the elements represent bars or trusses; this should also be used if the isoparametric beam element described in the text by Hinton and Owen (see Chapter 10) has been programmed.

The Plot 1D elts' option will plot the deformed 1D elements (bars, beams and joints) together with a deformed boundary of any 2D mesh; this can be seen in the example datafile eladbm1.dat, of a beam part-embedded in an elastic block. The 1D elements are not plotted in the Displacements' option, so as not to complicate the display.

Stresses

Stresses at Gauss points can be displayed as stress crosses, or extrapolated to a stress field which is contoured element-by-element, for a chosen stress component. Remember that FELIPE uses the compression-positive sign convention, so that the major principal stress will be the most positive (or least negative) in this sense. If both principal stresses are tensile at a particular point, the major principal stress will be the smaller tensile stress.
The first three stress components read from the .out file are taken as sigma_x, sigma_y and tau_xy for plane stress/strain. In axisymmetric analyses these will be sigma_r, sigma_z and tau_rz respectively. A fourth component is also read, which will be the out-of-plane stress: sigma_z for plane stress/strain, sigma_theta for axisymmetric. If the analysis has not produced values for this final component, it will be read as zeroes.

Stress crosses

The stress tensors at the Gauss points are resolved into the principal stresses, and a stress cross' drawn at each point. The lengths of the arms of the stress cross indicate the magnitudes of the major and minor stresses (in proportion to the reference stress which you must type in), and their directions indicate the principal stress directions. Compressive stresses are drawn in black, and tensile stresses in blue. The initial scaling factor is set to give a reasonable-size stress cross for the average stress in the mesh, but where there are areas of high concentration this will result in very large stress crosses at those points; you should try another, smaller scaling factor in this case.

Stress contours

Contours of stress are also drawn, interpolated within each element from the Gauss point stress values. You can contour major principal stress \sigma_1, minor principal stress \sigma_3, the stress average (\sigma_1 + \sigma_3)/2 or the deviator stress (\sigma_1 - \sigma_3)/2, as well as for each individual stress component. The contours can be filled-in in colour, or drawn as coloured lines (which is the option used for PostScript printing). The scales are written along the bottom. Note that the contouring is performed element-by-element, based on the element Gauss point values; thus, there will be discontinuities in stress at the inter-element edges. In a well-designed mesh, these will be small.

Yield zones

In Advanced-level analyses, the main program may test the Gauss point stresses against a yield criterion, such as the Mohr-Coulomb criterion, and use some nonlinear constitutive algorithm such as plasticity where yield occurs. To indicate Gauss points where yielding has occurred, a yield code' may be written in the results for the stresses (see format list below). This is used in the plasticity main engines' PLAST, PLADV, VPLAS, but the elasticity programs ELAST, ELADV can also test the minor principal stress for yield against the tensile strength. Under this option, you enter a yield code, and Gauss points with that code will be highlighted. A yield code of 0 means the Gauss point has not yielded. The FELIPE main engines' use a yield code of 1, but if you have developed a more complex plasticity program and wish to distinguish different types of yield (e.g. where several different yield criteria operate independently) you can assign a different code to each type.

Plots of nodal degrees of freedom

Following the plots described above, the user is offered the chance to view contour plots of each nodal degree of freedom. This is to allow for extra degrees of freedom (temperatures, pore pressures, etc) existing, though contours of displacements in the x-direction can also be obtained here, for example. If the extra d.o.f. exists only at corner nodes (as with the poroelasticity application CONSL.FOR), the contouring will take account of this when specified by the user.

The output data format read by FELVUE, including such extra d.o.f.s, is:

 Block Format Variables R1 I5,4E10.3 N DISPX DISPY disp3 disp4 R2 I5,I2, I3,6E10.3 L IJ KODE GX GY SX SY TXY sz
where KODE is an integer yield code (Non-yielded points take code 0). This is used in PLAST and VPLAS, and is also available in ELAST and ELADV to test if the tensile strength has been exceeded; in these programs, yielded points are indicated by KODE=1. Of course, you can develop your own main engines' and use other values of KODE to indicate other types of yielding, or indeed you can use this facility to distinguish other types of behaviour besides yielding.

Note that the G10.3 format is used to read real values. The main engines’ provided will output real values of displacements, stresses, etc. to the .out file in E10.3 format, so that very large and very small numbers can be coped with. However, greater accuracy for values of a particular order of magnitude can be achieved by editing the main engine’ source code to write in, for example, F10.5 format. This will be read equally happily by FELVUE.

Following the examination of d.o.f. plots, the user has the option of returning to the Main Menu if desired.

Postscript file plots

When you have finished inspecting the results, the screen reverts to Text mode, and you have the option to produce PostScript files. If the output file were called test.out, then the PostScript file of the deformed mesh would be called testa.ps, a PostScript file of the stress crosses would be testb.ps, a PostScript file of the stress contours would be testc.ps, and a PostScript file of the yield zone would be testd.ps. These plots can also be zoomed, by specifying a window' to plot. This is done by typing in the x coordinates of the left and right sides of the window, and the y coordinates of the top and bottom of the window. If you want to use this facility, therefore, you should note the coordinates you want to use, while viewing the graphical results on screen earlier.

Plotting further result sets

It is possible to make a main engine' output further sets of results, separated by a title line. This is useful in a time-stepping or incremental analysis, where results after certain times or load increments may be of interest. This is used in the soil consolidation main engine' CONSL.FOR, where the initial, undrained result is output, and then the final, drained solution after completing the timestepping is appended to the .out file (see conslex3.dat).

When the post-processor detects that more datalines exist following a set of results, a pop-up box asks if you wish to view the next result set. Answering y' will return you to the main menu, with the next set of results.