From where or how does the figure of 50 m sec protection switching calculated for MPLS/ ERPS?
The 50 ms maximum delay is a figure derived from practical use. Many real-time protocols can suffer a packet loss for that period without too much impact. That limit is a trade-off between reliability and cost.
It's "offical" in some IETF RFCs, probably starting with RFC 6378.
I also found this source stating that 50 ms originated from Automated Protection Switching (APS).