That method introduces more problems than it solves. There have been many discussions about how to properly offset the surfaces from the mean line.In this airfoils families, Ordinates of the cambered lying sections are obtained by laying off the thickness distributions "perpendicular to the mean lines".

__Here's one__, take note of the little spike in the pressure distribution at the leading edge of the NACA airfoil section, that is caused by the NACA's shortcut to draw in the leading edge radius, there's also a link to the Xfoil archive in post 24 of that thread in which Dr. Drela explains why he didn't implement the rolling circle method in Xfoil.

Yep, if you want a thicker version of an existing airfoil the best way to do it is find an acurate thicknnes form of the same family and add the mean line from the airfoil you started with so that you preserve both the camber and thickness distribution. Just multiplying the Y ordinates by 1.n will change the mean line and blunt the leading edge. By blunt I mean that, if you start with an airfoil that has a simple arc segment leading edge, after scaling the Y values the leading edge will be an ellipse with its major axis perpendicular to the chord.Because this, scale in one direction, plot mean line by midpoint "Y" ordinates and anything else of this type is only misinterpretation about airfoil section generation and the plot that you obtain is not any related airfoil.