[Q]Steady state gain matching with controller reduction

Dear Skogestad and Postlethwaite,

I have adopted a 2dof Hinf control to my system successfully and it
shows good tracking performance.
The controller has a little bit high state order so that I reduced the
states of the controller using optimal Hankel norm approximation. In
that case, reduced controller can't follow the reference command in
SIMULINK simulation. I guess that there needs some modification for
steady state gain matching.
The scaling constant matrix Wi given by (remark3 :eq(9.89), pp.388)
doesn't match.
How can I modify the scaling matrix Wi ?

Best Regards,

Ouiserg,Kim