Is the modification a localised modification over a finite distance or along the entire length of the column?  Are the welds on both sides of the 2.5” x 0.25” plates?

If the modification is localised then the designer must ensure the adequate transfer of the flange forces.  Provided the details are suitable at the change from “proper” flange to the “replacement” flange, then my feeling is that you have close to the original section capacity.

We have modified a lot of columns, mainly using flange doubler plates or beam tees to make a cruciform section, and these have been assumed to contribute to the overall section.  

If the existing column is in situ at the time of the modification, then the "tee" section will have some existing stress state.  Perhaps the additional flanges (or other additional stiffening elements) don’t contribute to resisting this stress state, however, as the column become more heavily loaded and approaches the ultimate limit state my feeling is that the combined section will be effective.

What is your technical rationale for assuming the replacement flange does not contribute to axial capacity or bending capacity?

If you take off the flange and then replace it 1" further towards the center of the web, then you have a different section with different properties.  Instead of a W10, now you have a W9.

ya... but buckling was likely about the weak axis... <G>

Bearing different too

A little more information on this discussion.  The flange is removed from the bottom 1/3 of an in-situ column.  The web of the column has also been reduced by approximately 1/3 of the original depth.  In analyzing the column, it will be fine if a flange is added at the edge of the web.  However, given the connection of plate to the side of the web with fillet weld only that are not full penetration, I believe that this would only be a T section with longitudinal stiffeners.  The column would be over designed for weak axis bending, but would be subject to failure when combined axial and bending about the major axis is applied.  I have no justification for my argument other than the fact that no one has yet to convince me otherwise.  If you have a simple example that can prove or disprove my theory, pleaqse share it.

In the simplified model of an I-beam, the flanges carry all the bending moment and none of the shear while the web carries all of the shear and none of the bending moment.

Calculate the shear flow at the web/flange interface as follows:
v = V x Q / I
   v = longitudinal shear flow
   V = shear force
   Q = first moment of area of flange about neutral axis
   I = second moment of area of section

Thus the flange, whether connected continuously by fillet weld or discretely by bolts or rivets, can be effective if the connecting elements can adequately transfer the shear flow.  Consider the design of plate girders as an example.

Get out your university textbooks or search the Internet if you need to read more about shear flow.