Keeping drones safe in urban airspaces

Researchers from NTU tap on high performance computing to identify hazardous airspaces through urban weather simulations in order to facilitate effective route planning and management.
"A block size simulation domain requires a large amount of memory for mesh generation, and a large number of CPUs are needed to run the parallel solvers, even for stead-state simulations. The computational time requirement for unsteady simulations could be substantially greater. We used NSCC's HPC resources extensively for CFD simulations with OpenFOAM while the visualisation server allowed us to view the flow solution as a whole instead of in smaller chunks on a local workstation due to memory and bandwidth limits."
John Wang
Senior Research Fellow / Assistant Programme Director
School of Mechanical & Aerospace Engineering
Air Traffic Management Research Institute (ATMRI)
Nanyang Technological University (NTU)

An important aspect of urban airspace utilisation that is often overlooked is the effect of building wake on the usability of the airspace. Given the relatively low crosswind and turbulence tolerance of a multirotor at cruise velocity, the crossing of an intersection with perpendicular funnelled flow could easily send the multirotor crashing into buildings or into the ground.

 

A team of researchers at Nanyang Technological University’s School of Mechanical & Aerospace Engineering, Air Traffic Management Research Institute, are leveraging on NSCC’s supercomputing resources to determine the fluid dynamic metrics that could be used to plot out hazardous airspaces to avoid based on prevailing surface wind conditions.

 

Urban wind field simulation was performed over a variety of terrain and buildings in order to identify very low-level airspace under or immediately above the building heights where funnelled, shear, or recirculating flow would pose a threat to multirotor stability or its track-keeping ability. Due to limitations with computational power and resolution of building models, only steady-state solutions could be used to construct the wind-field database.

"A block size simulation domain requires a large amount of memory for mesh generation, and a large number of CPUs are needed to run the parallel solvers, even for stead-state simulations. The computational time requirement for unsteady simulations could be substantially greater. We used NSCC's HPC resources extensively for CFD simulations with OpenFOAM while the visualisation server allowed us to view the flow solution as a whole instead of in smaller chunks on a local workstation due to memory and bandwidth limits."
John Wang
Senior Research Fellow / Assistant Programme Director
School of Mechanical & Aerospace Engineering
Air Traffic Management Research Institute (ATMRI)
Nanyang Technological University (NTU)

The teams aims for the simulation studies to lead the way towards enabling Unmanned Aircraft System Traffic Management (UTM) service providers to identify hazardous airspaces to be geo-fenced out from usage and to enable route-replanning without encountering additional hazard.

To find out more about how NSCC’s HPC resources can help you, please contact [email protected].

 

NSCC NewsBytes September 2023

Other Case Studies