Many Azure Search pricing tiers support changes in the number of replicas and/or partitions. Do operations increasing or decreasing the number of replicas or partitions affect the performance of queries?
i.e. would it interrupt any queries in progress, prevent new queries from being sent, or otherwise affect performance while these scaling operations are ongoing?
Increasing or decreasing the number of replicas or partitions consums computing resources in order perform the operation. Hence, it might affect the performance (Splitting or copy data across resources will affect query, suggest or even index operations), As an evidence, if you scale while reaching your limit during a load test you might get throttled.
I'm not saying you shouldn't scale during ongoing operations but as a heads up it will affect these operations.
On the other side, it's recommended to have 2 replicas for high availability of read-only workloads (queries), 3 or more replicas for high availability of read-write workloads (queries and indexing) on production as mentioned here.