The correct behavior aligns with Scenario 2. After the initial scaling action, the system waits for the cooldown period to end before starting a new duration evaluation. Therefore, an instance is added every 15 minutes if the CPU usage consistently exceeds 70%.
For more information:
-https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-understanding-settings
-https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-best-practices
If above links are not of any help I would suggest raising a support request directly to Azure for further assistance as sometimes scaling can be tricky in terms of behaviour.