The current high gas prices motivate end-users to replace their gas heating with electric heat pumps. This will likely cause frequent congestion issues in low-voltage (LV) distribution grids and slow down the heat pump adoption rate. To avoid or defer the expensive and complicated grid expansion, this study shares a solution approach of a Dutch Distribution System Operator (DSO) to enable the increasing adoption of heat pumps in existing dense housing areas. Data of the DSO and a local housing company have been combined to investigate the heat pump hosting capacity on a dense urban LV feeder, including realistic data of grid topology, load and heat dynamics, and practical operating characteristics of heat pumps. Our simulation compares two control strategies: (1) individual peak shaving and (2) central optimal power flow control. We show the central optimal power flow control with end-users' thermal comfort constraints and an objective function of minimizing losses can smoothen total grid loading and lead to flat voltage profiles. This allows the approach to be robust against baseload forecast errors, while the individual peak shaving is more prone to such errors. Moreover, by simulating the strategies on the worst-case scenarios where heat pumps are allocated to end-users at the end of the feeder, we determine the individual peak shaving strategy can slightly increase the heat pump hosting capacity from 49% where no control is imposed to 51%, while the central optimal power flow control allows 100% heat pump connections without causing grid congestion. Finally, recommendations to increase the heat pump hosting capacity are given based on simulation results.