While I have not performed this analysis, I would recommend setting the two phenomena as distinct chemical reactions, wherein, the surface erosion reaction occurs only at the water-pipe boundary and then the volume erosion reaction occurs in the volume of the pipe.  The reaction rate of the volume could be governed by the concentration of the oxidizing agent which could be allowed to "diffusion" into the pipe material.  Furthermore, you could setup a moving mesh which reduced the volume or increased the porosity of the pipe as the two phenomena occurred.  This would me a little more complicated, but probably more realistic.  I would definitely build two separate models, one for volume and one for surface, to validate they are working correctly and then integrate them.