tips.md

@@ -6,5 +6,8 @@ The domain can be as large as requested. However, placing the farfield boundarie
 * Mesh  
 To attain results that are grid independent, surface mesh sizes of about 1/200 of the local chord need to be considered, alongside a stretching factor for volume elements of 1.1. However, these parameters will produce large and dense meshes not suited for fast preliminary design, especially in 3D. Engineering accuracy can usually be achieved by using surface mesh sizes of 1/100 of the local chord combined with a stretching factor of 1.2.  Additionally, the mesh at the trailing edge of wingtips should be coarsened to avoid the local singularity caused by the infinite speed vortex (see [Known issues](bugs)).
 
+* Kutta condition  
+In order to obtain results as accurate as possible, the original nonlinear formulation of the Kutta condition should be used. However, in cases where convergence is difficult (typically in high-speed and high-lift conditions), the linearized Kutta condition should help.
+
 * Linear solver  
-For best performance, PARDISO or GMRES should be used. If memory is not a concern, PARDISO is the recommended choice. If GMRES is used to treat cases where convergence is difficult, using a fill-in factor of 3 instead of 2 for the ILU preconditioner may help.
+For best performance, PARDISO or GMRES should be used. If memory is not a concern, PARDISO is the recommended choice. If GMRES is used to treat cases where convergence is difficult, using a fill-in factor of 3 instead of 2 for the ILU preconditioner, as well as increasing the number of restart may help.
\ No newline at end of file