Ok, plotting the integrand shows that there are vertical asymptotes in N*Pi, where N is an integer. Integrating from 0 to Pi/2 means the result would be half the area of the spikes in the plot. My maths knowledge is relatively modest but I remember something about improper integrals and maybe the solution - if there's a solution - could be something like the directional right limit of the integral when one of the endpoints tends to zero, i.e. limit(int(ss,t=a..Pi/2),a=0,right);, where ss is the integrand but, again, had no luck. I'm going to consider this a dead end and try something else. Anyway, thank you for your time. Pedro