@srestha, mam, in the question they mentioned that
that all processes are long running tasks.
it means the CPU Bound process have more than the TQ, By taking first cycle completion, we are calculating the utilization of CPU.
But there is a problem with considering the context switches at front ( Before starting the schedule, we choose one process then we have to load it, some people call it as also context switch ) and at end (May it enter into next cycle due to loading the process which is already preempted. )
But we can say it is approximately, $\frac{20}{21.1}$ or $\frac{20}{21}$