The combination of heatwaves with the urban heat island (UHI) effect can lead to very strong heat stress levels. A multiscale numerical approach is developed to model the urban microclimate, downscaling from city to local scale. First at mesoscale, cities and their surroundings are modelled with a domain size of around 50 km, using the nested meteorological weather prediction model WRF (Weather Research and Forecast Model). Second at microscale, urban neighbourhoods are modelled with a domain size around 1-2 km. This is performed using a solver developed by the authors, urbanMicroclimateFoam (2024), based on OpenFOAM libraries. The solver employs consecutive calculations of air flow with steady Reynolds-averaged Navier-Stokes (RANS), coupled with an unsteady heat and moisture (HAM) transport model in urban porous materials. HAM considers the dynamic heat and moisture storage and provides the surface temperature and humidity values as boundary conditions for the computational fluid dynamics (CFD) model. The long-wave and reflected short-wave radiative exchanges are calculated with view factors using a radiosity approach with multiple reflections. Direct solar irradiation is modelled with ray casting to capture shading. Foliage of trees is modelled as porous zones in the CFD domain using a leaf area density (LAD), which provides shading and modifies wind flow. Source/sink terms are determined within these porous zones for momentum, heat, moisture. Heat balance at the leaf surface of sensible, radiative and latent heat fluxes is solved for determining the leaf surface temperature. Grass-covered surfaces are modelled with a leaf area index (LAI) and radiative exchanges between grass and underlying soil are considered. Local boundary conditions obtained from mesoscale simulations are applied to drive the flow in the microscale simulation using blending zones. As examples, this approach is used to simulate urban thermal comfort and assess potential heat mitigation measures for several case studies.