TV
T. Vargiemezis
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
1 records found
1
Climate change is intensifying urban heat stress, and microclimate simulation can support climate-responsive planning. Computational fluid dynamics (CFD) is the dominant approach at the microscale, but urban CFD studies are predominantly isothermal, neglecting buoyancy effects. Yet these effects become non-negligible under the low-wind, thermally active conditions that climate change is making more frequent. Coupled solvers such as urbanMicroclimateFoam (uMF), built on OpenFOAM, resolve airflow, heat and moisture transport, radiation, and vegetation interactions that wind-only formulations omit. However, it remains unclear under which conditions a wind-only simulation is sufficient and when the additional complexity of a coupled solver is justified. Addressing this question is hindered by three further gaps: published applications of uMF to realistic urban sites with complex terrain are scarce, its computational costs are not well characterised, and rigorous validation against field measurements is limited.
To address these gaps, this thesis asks three questions: whether uMF can be feasibly applied to realistic urban models with non-flat terrain; how its mesh requirements, solver parameters, and computational demands compare to those of the wind-only solver simpleFoam (SF) on the same case; and how the two solvers’ wind-speed predictions compare against street-level field measurements during a representative heatwave hour.
Applied to the Carnegie Mellon University campus in Pittsburgh during one hour of a heatwave on 27 August 2024, with a 3D model reconstructed in City4CFD from LiDAR and building footprints and validated against four MORICHI street-level stations, three findings emerge. First, applying uMF to realistic terrain is feasible but requires time-consuming, case-specific geometric adaptations compared to the geometry-robust SF. Second, on identical geometry and 20 cores, uMF took 3.6× longer than SF (6 h 28 min versus 1 h 49 min), which does not represent a barrier to the use of the solver. Third, uMF improved wind-speed agreement across every aggregate indicator: FB moved from +0.65 to −0.11, MG from 1.89 to 0.83, NMSE was reduced by 65%, and FAC2 rose from 0.50 to 0.75.
These results indicate that uMF merits adoption when buoyancy effects are non-negligible and the analyst can absorb the case-setup overhead. The findings are based on a single one-hour window under one inflow direction, with the observed wind speeds at three of the four stations falling below the anemometer accuracy threshold, and vegetation modelled as a porous medium. Addressing these limitations is identified as a priority for future work. ...
To address these gaps, this thesis asks three questions: whether uMF can be feasibly applied to realistic urban models with non-flat terrain; how its mesh requirements, solver parameters, and computational demands compare to those of the wind-only solver simpleFoam (SF) on the same case; and how the two solvers’ wind-speed predictions compare against street-level field measurements during a representative heatwave hour.
Applied to the Carnegie Mellon University campus in Pittsburgh during one hour of a heatwave on 27 August 2024, with a 3D model reconstructed in City4CFD from LiDAR and building footprints and validated against four MORICHI street-level stations, three findings emerge. First, applying uMF to realistic terrain is feasible but requires time-consuming, case-specific geometric adaptations compared to the geometry-robust SF. Second, on identical geometry and 20 cores, uMF took 3.6× longer than SF (6 h 28 min versus 1 h 49 min), which does not represent a barrier to the use of the solver. Third, uMF improved wind-speed agreement across every aggregate indicator: FB moved from +0.65 to −0.11, MG from 1.89 to 0.83, NMSE was reduced by 65%, and FAC2 rose from 0.50 to 0.75.
These results indicate that uMF merits adoption when buoyancy effects are non-negligible and the analyst can absorb the case-setup overhead. The findings are based on a single one-hour window under one inflow direction, with the observed wind speeds at three of the four stations falling below the anemometer accuracy threshold, and vegetation modelled as a porous medium. Addressing these limitations is identified as a priority for future work. ...
Climate change is intensifying urban heat stress, and microclimate simulation can support climate-responsive planning. Computational fluid dynamics (CFD) is the dominant approach at the microscale, but urban CFD studies are predominantly isothermal, neglecting buoyancy effects. Yet these effects become non-negligible under the low-wind, thermally active conditions that climate change is making more frequent. Coupled solvers such as urbanMicroclimateFoam (uMF), built on OpenFOAM, resolve airflow, heat and moisture transport, radiation, and vegetation interactions that wind-only formulations omit. However, it remains unclear under which conditions a wind-only simulation is sufficient and when the additional complexity of a coupled solver is justified. Addressing this question is hindered by three further gaps: published applications of uMF to realistic urban sites with complex terrain are scarce, its computational costs are not well characterised, and rigorous validation against field measurements is limited.
To address these gaps, this thesis asks three questions: whether uMF can be feasibly applied to realistic urban models with non-flat terrain; how its mesh requirements, solver parameters, and computational demands compare to those of the wind-only solver simpleFoam (SF) on the same case; and how the two solvers’ wind-speed predictions compare against street-level field measurements during a representative heatwave hour.
Applied to the Carnegie Mellon University campus in Pittsburgh during one hour of a heatwave on 27 August 2024, with a 3D model reconstructed in City4CFD from LiDAR and building footprints and validated against four MORICHI street-level stations, three findings emerge. First, applying uMF to realistic terrain is feasible but requires time-consuming, case-specific geometric adaptations compared to the geometry-robust SF. Second, on identical geometry and 20 cores, uMF took 3.6× longer than SF (6 h 28 min versus 1 h 49 min), which does not represent a barrier to the use of the solver. Third, uMF improved wind-speed agreement across every aggregate indicator: FB moved from +0.65 to −0.11, MG from 1.89 to 0.83, NMSE was reduced by 65%, and FAC2 rose from 0.50 to 0.75.
These results indicate that uMF merits adoption when buoyancy effects are non-negligible and the analyst can absorb the case-setup overhead. The findings are based on a single one-hour window under one inflow direction, with the observed wind speeds at three of the four stations falling below the anemometer accuracy threshold, and vegetation modelled as a porous medium. Addressing these limitations is identified as a priority for future work.
To address these gaps, this thesis asks three questions: whether uMF can be feasibly applied to realistic urban models with non-flat terrain; how its mesh requirements, solver parameters, and computational demands compare to those of the wind-only solver simpleFoam (SF) on the same case; and how the two solvers’ wind-speed predictions compare against street-level field measurements during a representative heatwave hour.
Applied to the Carnegie Mellon University campus in Pittsburgh during one hour of a heatwave on 27 August 2024, with a 3D model reconstructed in City4CFD from LiDAR and building footprints and validated against four MORICHI street-level stations, three findings emerge. First, applying uMF to realistic terrain is feasible but requires time-consuming, case-specific geometric adaptations compared to the geometry-robust SF. Second, on identical geometry and 20 cores, uMF took 3.6× longer than SF (6 h 28 min versus 1 h 49 min), which does not represent a barrier to the use of the solver. Third, uMF improved wind-speed agreement across every aggregate indicator: FB moved from +0.65 to −0.11, MG from 1.89 to 0.83, NMSE was reduced by 65%, and FAC2 rose from 0.50 to 0.75.
These results indicate that uMF merits adoption when buoyancy effects are non-negligible and the analyst can absorb the case-setup overhead. The findings are based on a single one-hour window under one inflow direction, with the observed wind speeds at three of the four stations falling below the anemometer accuracy threshold, and vegetation modelled as a porous medium. Addressing these limitations is identified as a priority for future work.