Why would a compressible flow solver give negative entropy ?

Where could one look for possible error sources?

Thanks in advance

> Why would a compressible flow solver give negative entropy ?

Because entropy is not a numerically conserved quantity in the particular scheme. Violating physical bounds of quantities that are not numerically conserved is not necessarily an indication of an error in the code.

> Where could one look for possible error sources?

If the scheme is consistent, refining the grid will reduce the numerical errors and hence reduce whatever you are looking at. If it does not then something is broken and then you can look for errors.

In my multi-fluid cases, I experienced such a thing in computing ghost-fluid treatments.

To avoid that, I nondimensionalized the variables.

If it is not violated any rule, you can nondimensionalize variables and you may avoid any error related with entropy.