Principle: 

Wire kerf definition (Ray) :

The software is going to compensate for the headroom of melted foam, depending on foam type and cutting speed. As root and tip speeds are known, the program is able to perform the required trajectory corrections in order to match the desired airfoil sections, even with wire glowing effect.

If you want more informations on these effects, have a look over Jacques's document:   Rapport sur effet de rayonnement (French only up to now, sorry)

The wire kerf may be compensated using two ways:

1. compensation applied to the airfoil section
2. compensation over the basic height.

In the first case, compensation is equivalent to a negative skin thickness. The compensation is always perpendicular to the wire movement.
For the root chord, the correction has a constant value, equal to half the melted diameter for the given cutting speed V. For the tip chord, the correction vary with the cutting speed value between two points.

With multiple sections wings, the basic height compensation allows to align exactly cores and bottom foam bed for each trapezium (Refer to drawing). This compensation is only applied on the Y axis.
 

Critical points:

Unfortunately, all those corrections works well but in two particulary cases:

At foam block ends where the wire temperature is not the same than in the middle of the block, because the wire is hotter within the block than outside where it is cooled by ambiant air convection.

Also at the beginning of the cut, when the equilibrium temperature for the wire is not reached: As the wire first penetrate in the foam it is hotter and radiate more. There a 1.5 to 2 cm cut range without thermal equilibrium. 

The following diagram shows (exagerates) the defects related to these particular cases: 

 

Unfortunately, the result of the wire kerf is going to change with the block sweep. The higher the sweep angle, the more important will be these parasitic effects and the  differences between tip and root. The following diagram explains this particular case:

• Menu:

The [CNC:Wire temperature] menu allows to define default conditions for wire temperature and glowing effects. The software uses a database with all types of blocks and foam types wich is stored in the file Foam.bd . This is a text file, and may be modified with any text editor (preferably while the CNC software is *NOT* running). The file format follows:

1st line :     V1

For all type of block/foam, one line containing: 
Name: Cutting_Speed_V  Wire_glow_effect_at_V Wire_glow_effect_at_V/2

One blanck character between each data value. The wire glow effect is the diameter of the melted foam in millimeters.

Note that it is possible to manually enter another speed value. In this case, this value remains valid until another type of block is selected.
 

Standart values for wire kerf:

During tests, some values have been determined with currently used foams. These values seems to give correct results. You can see the results hereafter.

• Steps:

Définition of new wire kerf values:

 
1. If you want to use a new type of material or if you are trying to adjust the default values, the use the Cut[Test Cut] menu

In order to avoid distorsion at block ends:

1. Define a block slightly wider, with a trailing end margin (around 2.5 cm). This is used as an entry point into the foam and allows the wire to reach thermal equilibrium prior to effectively begin the real airfoil section cut.

In order to avoid distortion at the beginning of the cut:

These margins are located at both block ends (left and right), but also at leading and trailing edges. When cutting the trailing edge, if the cutting speeds are different at root and tip, the wire temperature and the glow are also different, so the trailing edge is misaligned. This remains true even with an instantaneous adjustment of the wire temperature.
In fact, to have a perfect match and smooth section evolution over the wing, you need to have the wire cut beginning at the same instant for the trailing edge, from root to tip. Of course, you should also end the cut simultaneously at the leading edge (For those beginning the cut with the LA, same thing with reverse order). Imagine a trapezoidal wing, with 200mm and 150mm chords, and 10 mm margins added at both leading and trailing edges. Then as your margins are not proportionnal to the chords, as the wire moves at constant speed (as Gilles asserts us), and displacement at both wing ends are calculated to penetrate and get out of the foam at the same time, the wire will actually reach the root trailing edge whilst still in the margin at tip. At the leading edge the wire will reach the margin at root prior to end the airfoil cut at the tip.
The simplest workaround is to add at leading and trailing edges margins which are proportionnal to the choords.
So:

Cut your block - and your wing - wider than necessary (2 cm margin at both root and tip)
After this first shaping, remove the distorted margins with a knife
With everything setup and adjusted correctly, the end precision of a wing pannel is really good.
 

A method to evaluate and adjust the wire kerf compensation:

    Evaluation, Why ? :

To allow the Cnc software to compensate for the wire kerf when cutting a wing (straight or trapezoidal), and get foam cores of the desired thickness.

    How to evaluate :

A test cut requires to cut horizontal slices in a foam block of known thickness. The wire kerf is equal to the thickness difference before and after cuts, divided by the number of slices made in the block. This test should be done with the best wire temperature for a given speed V, and with the same temperature at half speed V/2 (This mimicks the case of a wing with a tip chord dimension half of the root chord).

    Practising:

For all tests, prepare two foam blocks, identical to those used for wire temperature adjustment. As foam blocks are not plane, they have to be shaped on one side with the 'Test cut' menu. You can proceed with all blocks in a single operation if you want to save time. Then measure the thickness of the blocks. With one block, at V speed and ideal wire temperature, cut as many slices as possible, with a minimal thickness for a slice around 10 mm (If slices are too thin, this will corrupt the result). Working conditions are the same than for wire temperature determination, same displacements at both ends, so the wire should begin and end cut simultaneously over the whole span. Then measure the resulting thickness of the remaining block (slices in fact). Preferably in the middle and without too much pressure. Then calculate the ray value for the nominal speed. Then repeat the process with the second block, with the same wire temperature adjustment but a a speed devide by two. This will give you the ray value at V/2.

The values can then be entered in the "Wire Temperature" menu and will be stored then into the foam.bd file.

For best compensation by the CNC software, this evaluation should be done over the whole speed range you can use for foam cuts. For instance from 0.5 mm/s up to 3.5 mm/s with 0.5 mm/s steps. All the values are to be entered in the "Wire Temperature", or directly into the foam.bd file.

Findings:

If you proceed with this test over the whole usable speed range of your foam cutting table, you are going to realize that for a given foam type, at the ideal wire temperature adjustment, the wire kerf values does not change. For instance, with expanded polystyren KNAUF 400 EM and a 0.35mm wire, ray at V = 0.8mm and ray at V/2 = 1.5mm.
Another finding is the fact that the ideal temperature adjustment is nearly proportional to speed. The evaluation and adjustment of wire temperature is so simplified. Rather than performing ideal temperature adjustments over the whole speed range, it may be determined at lowest and highest speeds and then interpolated between those two measure points. 
 

 

As ray values at V and V/2 are identical for all speed values, the wire kerf determination may be performed at both maximal and minimal speed and then averaged at V and V/2. These averaged values will be set for all intermediate speeds in the wire temperature menu (without further tests).

Future : (NB already implemented in recent software versions)

With an electronic control board which supports the wire temperature control and allows the software to read back the values (BB20XX) you will not have to adjust the wire temperature any longer as this will be performed by the CNC software.