With tetrahedral (or any other volume elements) the thin layers would need a lot of elements (imagine just 4 elements per the FRP layer thickness and 4:1 aspect ratio!). I guess this modelling would not be feasible with today's computers...

A better option is to model the CFs as discrete bars at/near the concrete surface. About 2 bars per element (with the area corresponding to the total CF crossection area per the concrete element width) should be enough. However, it is not possible to capture debonding/delamination with this modelling.

Therefore, we recommend using shell elements for the wraps. Then, you can model the interface between the concrete and the wrap using contact (GAP) elements. You set the material to the properties of the matrix material (epoxy, glue; elastic modelling is often enough), and add the fibres as reinforcement layers (usually, a trilinear working diagram is enough to capture the behaviour of the fibres).

Please see the Shell elements description in the Manual. Here, I only summarize the most important settings:

1. For all macroelements that use shell/plate elements, the following has to be set:
1.1. "quadratic" in FE mesh - Generation
and
1.2. "brick" in FE mesh - Macroelements

2. For pure tension, a 3-4-layered shell is enough, for setups where bending of the shell has influence, usually 6-10-layered shells are needed to get accurate results.

3. Please pay attention to the wrapped edges - the best connection at edges is a 45 degrees (or a different corresponding angle if the thicknesses are not the same, see the figures in the manual).

It should not matter which surface is top and which bottom, the important thing is that the thickness is defined in the right direction  (the smallest one), and as long as the top and bottom surfaces are parallel, both give correct result. However, in corners where 2 FRP elements are connected, I would recommend that these have both the same orientation, i.e., both have "top" to the beam or both "bottom" to the beam.

Additionally, if one defines reinforcement layers in different depths of the shells, it makes sense to have consistent orientation such that the distances from top/bottom have the same meaning everywhere.

Do not hesitate to contact us when you have additional questions or run into any problems with the modelling.