Explanation

Step 1: Create a virtual network...
We recommended that for test failover, you choose a network that's isolated from the production recovery site network specific in the Compute and Network settings for each VM. By default, when you create an Azure virtual network, it is isolated from other networks. The test network should mimic your production network:
The test network should have same number of subnets as your production network. Subnets should have the same names.
The test network should use the same IP address range.
Step 2: Add a public IP address...
Because Site Recovery does not replicate the cloud witness, we recommend that you deploy the cloud witness in the disaster recovery region.
Step 3: Shut down production virtual machines
Make sure that the primary VM is shut down when you run the test failover. Otherwise there will be two VMs with the same identity, running in the same network at the same time. This can lead to unexpected consequences.
Step 4: Select Test failover from the Recovery Plans blade
References:
https://docs.microsoft.com/en-us/azure/site-recovery/site-recovery-test-failover-to-azure