several mode shapes of a building

Absolute summation of all the results for using load patterns proportional to the modes is too conservative. You may try combining them with the SRSS rule, but a better approach in my opinion is to use adaptive pushover method.

You can't just sum them up. Even the use modal combination rule is questionable due to the present nonlinearity, though there has been some papers published proposing combination rules. I would be cautious to do before applying any of those methods. Are you using adaptive modal pushover analysis that Chopra proposed or any other?