genetic algorithm Fitnesss and mean fitness convergence

Im using GA tool box

my fitness function is

f=f1+f2

where f1 and f2 are the outputs of a function called obj [f1,f2]=obj(x)

I have a constraint on f1 that it has to be f1>10

and x (design variable vector) is a thickness vector

I also have a mass const..

So i created a non linear constrain code for the mass and f1

however for the code to run fast i set a condition inside obj(x),so that i dont have to go through the code when the solution is already infeasible

[f1,f2]=obj(x)

if mass>massmax

f1=100;

f2=0;%

now my code runs fast but the best fitness and mean fitness curves are not reaching same values at the end of the stall generations

what did i do wrong and how do i fix this?