Blockchain is a distributed ledger technology that ensures secure and transparent management of transactions. It is increasingly being adopted across diverse fields such as finance, logistics, and healthcare. A key performance metric in blockchain systems is block generation time, which affects transaction confirmation speed and overall system responsiveness. In Proof-of-Stake (PoS) blockchains like Cosmos, block generation time was previously assumed to be nearly constant. However, recent studies have revealed non-negligible variability. To explain this, prior research modeled the block generation process in Cosmos using a simplified framework, showing that the resulting distribution aligned reasonably well with empirical data. These models, though informative, did not account for the network topology connecting validators. Since the block generation process in PoS blockchains involves broadcasting verification requests to all validators across the network, the underlying network topology is expected to impact block propagation latency. In this paper, we investigate the relationship between network topology and block generation time by simulating information propagation over networks generated by various network generation models. We evaluate how accurately the block generation time distributions can be approximated by statistical models. As a result, we found that when assuming communication delays between validators follow an exponential or log-normal distribution, the Gumbel distribution provides the best fit regardless of the network topology.