After the print of Fig9
one sees using ?U1 and FullDefinition[U1] that U1 is still well defined (seemingly). So lets test the U1 on the diagonal
In[76]:= U1[a, a]
Out[76]= 1.81085*10^-13 - 2.16846*10^-13 I
In[77]:= U1[4 a, 4 a]
Out[77]= 3.1515*10^-14 - 1.61214*10^-14 I
these small imaginary parts prevent the graphics from appearing (Graphics3D wants real values) and bring you to the idea to type
Fig10 = Plot3D[Re[U1[x, y]] + U2[x, y], {x, a, 4 a}, {y, a, 4 a}, PlotRange -> Automatic]
to see U1 - its real part actually - in the sum again:
It lies in your responsibility to check whether U1 had imaginary parts already right after its definition and if so, why the first picture Fig7 did print.
HTH
