In fact, due to the usually non-linear nature of systems, I somehow disagree with the DNV approach because the way it constructs the load cases is partly for non-linear systems and partly for linear systems. But I guess is not so important for DNV what I think.
Also I would rather apply the wind and the accelerations over the system at operating temperature instead of design temperature, because many (?) times the supports with gaps that are active in design condition are not in the normal operating condition.