CFD Online Discussion Forums - Solution breaks on highly stretched grid

I am solving 2d Reimann problem test cases with a WENO5 2D solver I wrote. It works fine when I use a uniform Cartesian mesh but when I add a transformation to have exponential compression in the y-direction, for large values of the exponent, the solution starts giving negative pressure.

If $\zeta$ and $\eta$ forms my computational domain (unit uniform cartesian grid),
I am defining

$x=\zeta$

$y=\frac{e^{k*\eta}-1}{e^k-1}$ ,

where k is a factor i control.

For k=1, the solution is fine but when i increase k to 3 (highly stretched out in y-direction), the solution starts breaking down (negative pressure).

Does grid transformation actually cause problems or is there some problem with my code?

Yes i have made the appropriate changes in the conservation equations (Multiplying the necessary Jacobians).