Content uploaded by Frederick Peter Ortner
Author content
All content in this area was uploaded by Frederick Peter Ortner on Mar 23, 2021
Content may be subject to copyright.
Content uploaded by Frederick Peter Ortner
Author content
All content in this area was uploaded by Frederick Peter Ortner on Mar 23, 2021
Content may be subject to copyright.
MODELING UAM SCENARIOS FOR URBAN DESIGN
FREDERICK PETER ORTNER1and JEFFREY HUANG2
1Singapore University of Technology and Design
1peter_ortner@sutd.edu.sg
2École Polytechnique Fédérale de Lausanne
2jeffrey.huang@epfl.ch
Abstract. Recent developments in unmanned aerial vehicles (UAVs),
including drone delivery services and air taxis, are revolutionizing
urban transport, leading to a new field of research referred to as
Urban Air Mobility (UAM). While several contemporary efforts
to computationally model future scenarios for UAM exist, in this
paper we argue that these models tend to be narrowly conceived as
air-space design and management tools and provide little information
on ground-level impacts. This paper describes an ongoing effort to
create UAM modelling tools useful specifically to urban designers
as part of a push toward integration of urban airspace design with
ground-level master-planning. Current functions permit designers
to visualize drone-fleet origin-corridor-destination routes, generate a
strategic model of UAM noise, and compare tradeoffs between UAM
system efficiency and noise.
Keywords. Urban air mobility (UAM); urban design; data-driven
design; simulation; parametric design.
1. Introduction
1.1. URBAN AIR MOBILITY (UAM)
In the future many thousands of UAVs (unmanned aerial vehicles, like drones and
air taxis) will fly in the airspace above cities, not only making deliveries, but
also carrying out many as-of-yet unimagined tasks. A new field of research on
Urban Air Mobilities (UAM) is tackling questions of how this future airspace may
function and how it will change our cities. While air taxis remain on the distant
horizon, many technology companies are working today to make drone delivery
commonplace, including Amazon and Google spin-off Wing Aviation (Curlander
et al. 2017). Though today’s drone deliveries mainly occur in rural or peri-urban
areas of low population density, this is likely to change in the near future as UAV
delivery companies pursue larger urban markets (Cervantes and Herrera 2019).
1.2. AIRSPACE DESIGN AND URBAN DESIGN
Current literature indicates that UAM traffic in urban areas is unlikely to follow
direct, ‘as the crow flies,’ origin-destination routes. Jiang et al. write, for example,
RE: Anthropocene, Proceedings of the 25th International Conference of the Association for Computer-Aided
Architectural Design Research in Asia (CAADRIA) 2020, Volume 2, 71-80. © 2020 and published by the
Association for Computer-Aided Architectural Design Research in Asia (CAADRIA), Hong Kong.
72 F.P. ORTNER AND J. HUANG
that a future UAM airspace would include, ‘elements of airspace design, corridors,
dynamic geo-fencing,’ among other restrictions which would limit the paths UAVs
could take through airspace (Jiang et al. 2017). The elevation at which different
types of UAVs will operate is also an open regulatory question. In 2015 Amazon
proposed a widely-cited two-tiered airspace design for UAVs with ‘low-speed
localized traffic,’ operating up to an elevation of 61m and a ‘high-speed transit’
zone from 61 to 122m for, ‘highly automated vehicles operating beyond line of
sight’ (Amazon Prime Air 2015). Airspace design will have an important impact
both on the efficiency of any UAM system, as well as its impacts on ground-level
activities.
General guidelines, however, are not enough to design airspace for UAM:
detailed computational models and flight control tools are needed. Several UAM
simulations and flight control tools are in the first stages of being developed and
tested. AirMap in the United States is currently testing a UAS Traffic Management
(UTM) Platform on that nation’s first 50-mile, ‘drone corridor,’ in upper New York
State (New York State Governor’s Office 2019). In Singapore the NTU Air Traffic
Management Research Institute is developing Traffic Management for Unmanned
Aircraft Systems (Salleh et al. 2018). They have modeled urban route network
scenarios for an urban town in Singapore with an emphasis on simulating system
capacities over different urban terrains (Ibid).
While the examples cited above demonstrate that unmanned aircraft system
traffic management (UTM) is developing rapidly, little effort has been focused on
modelling the impacts of Urban Air Mobility for city planning, urban design and
architecture. This oversight seems particularly concerning when we consider the
potential ground-level impacts of large-scale UAM systems.
1.3. GROUND-LEVEL EFFECTS OF UAM
The primary effects of UAM on urban design will be through changes in mobility
patterns. In the short term some ground-based delivery services will be replaced
by UAM services. In the longer term, air-taxis like Volocopter may change
the itineraries people take through the city. These primary effects are already
partially anticipated and built into products being developed by Amazon Prime
Air, Alphabet’s Project Wing, Uber Elevate, Matternet in Switzerland, Flytrex
in Israel, and Volocopter in Germany. Urban designers and architects like
Norman Foster, Jonathan Ledgard, Liam Young and Keith Kaseman have begun
to anticipate these changes in their designs (Taub 2019).
The externalities, or unintended consequences, of scaled-up UAM, however,
remain relatively under-studied. We know anecdotally that city residents may
experience inconvenience from drone noise, privacy intrusion or drone crash
(Jensen 2016, Garrett and Anderson 2018). However, there is a knowledge gap on
how these potentially adverse secondary effects will evolve as UAM scales up, and
in particular we are missing quantitative and spatial data for responding to these
possible impacts as urban designers. Considering how the introduction of modern
airports disrupted 20th century cities, it would seem prudent to anticipate the
potential externalities of UAM in our future cities before problems occur (Ortner
2017). Urban designers and planners will need the capability to weigh in on where
MODELING UAM SCENARIOS FOR URBAN DESIGN 73
UAM traffic should be concentrated over pre-existing areas of the city, and for new
areas of the city to integrate airspace design with urban design.
1.4. A PRELIMINARY UAM MODELLING TOOL FOR URBAN DESIGN
To address the knowledge gap on the ground-level effects and externalities of
UAM today’s urban designers need tools to help them consider airspace design
and ground-level design side-by-side and compare tradeoffs between the two. This
integrated approach recognizes that we now live in a truly three-dimensional city,
and that planning and design for our urban future requires an integration of airspace
and ground-space planning. Future tools for urban design will need to support
designers as they advocate for airspace designs which maximize benefits to the
city below while minimizing negative externalities.
As a preliminary step, on the way to a more general UAM model, we developed
a model to study a corridor-based urban drone delivery system and its noise
impacts at ground level. This decision was based on a few key learnings from our
literature review; namely that drone delivery in urban areas is likely in the near
future, that drones in urban areas will not fly direct origin-destination routes but
will likely be confined to geofenced areas or corridors, and finally that aggregate
ground level impacts like noise are only anecdotally addressed in current literature.
In a methods section below we describe the key variables, outputs and
functions of our model. In a subsequent results section, we describe how the tool
was used in an academic urban design project for the city of Glasgow, allowing
designers to anticipate the impacts of airspace design scenarios for a district
masterplan. Finally, in a conclusions section we describe limitations of the current
model and future plans to improve and expand it.
2. Methods
2.1. DEFINING KEY OUTPUTS: UAM EFFICIENCY METRICS VS.
GROUND-LEVEL NOISE IMPACTS
Our goal in producing this tool was to assist urban designers to consider future
UAM airspace scenarios as they develop masterplans. Specifically, we aimed
to support understanding tradeoffs between the design of urban airspace and the
design of ground-level urban masterplans. These tradeoffs could be understood
in terms of airspace benefits vs. groundspace benefits but also in terms of
airspace efficiency vs. unintended ground level consequences like noise. In a first
step toward a more comprehensive UAM modeler for urban designers/planners
we chose in this preliminary tool to consider trade-offs between UAM system
efficiency and ground-level UAM noise distribution.
Efficiency of UAM systems is indicated by two metrics in our model; 1) total
km travelled (for a given number of origin-destination-corridor itineraries); as well
as 2) number of out of range destinations (for a given set of itineraries.) These two
metrics provide a simple means of understanding the feasibility of an airspace
scenario and of comparing efficiency between different airspace scenarios with
identical origins-destinations. These two metrics and their use is further explained
in section 2.2.
74 F.P. ORTNER AND J. HUANG
Ground level noise levels in our model are indicated by time-averaged noise
pressure levels measured in decibels. These noise levels are simulated for a grid
of sample points defined by the user. Explanation of the function for generating
ground level noise values is detailed in section 2.3.
We anticipated that comparing UAM efficiency against a noise-level heat
map would permit urban designers to discriminate between airspace/masterplan
designs in search for variations that would maximize efficiency of airspace use and
minimize noise-based disruption to sensitive urban areas. This is not presented as
a comprehensive UAM model, but rather as a preliminary step toward supporting
future integration of urban land-space planning and air-space design. Further
discussion of how the key output metrics were used in an urban design test case
are detailed in section 3.
2.2. KEY FUNCTIONS: VISUALIZING ORIGIN-CORRIDOR-DESTINATION
SCENARIOS
The first key function of our drone airspace visualization tool allows designers to
combine origin and destination data with possible corridors and compare resulting
UAM airspace scenarios. In this first function we assume that regulations will
require drones to begin their flights by ascending directly to a pre-determined
corridor height, then proceeding along the shortest path to the drone corridor
(figure 1). Beginning flight on a vertical vector minimizes nuisance to surrounding
areas and would be a probable feature of future urban drone regulation according
to contemporary literature (Ippolito et al. 2019).
Figure 1. Drone routing sequence diagram (right), flight corridor elevations on site (left).
The corridor-based UAM flight-routing function used in our tool is derived
from similar procedures identified during our literature review. It reflects, for
example, a simplified version of the point-to-point package delivery use-case
presented by NASA’s Safe50 reference design study in 2019 (Ippolito et al. 2019).
The origin-corridor-destination function requires four inputs: (1) a list of origin
points PO (i.e. droneports), (2) a list of destination points PD, (3) a list of corridors
(as polylines), and (4) a corridor height value (figure 1). From a given destination
PD the tool works backwards, first finding the closest corridor and then the closest
origin point to that corridor. The sub-segment of the corridor lying between the
closest points to origin and destination is then identified. A polyline is constructed
linking origin (PO), a point at corridor height above the origin (P1), the corridor
sub-segment (P2 - P3), a point at corridor-height above the destination (P4) and
the destination point (PD) (figure 1).
MODELING UAM SCENARIOS FOR URBAN DESIGN 75
We anticipate that delivery drones would need to return to their point of
origin (the droneport) after delivery. To reflect this our function permits the
initial origin-destination polyline to be linked to an additional return polyline
which replicates (in reverse) the process described above. In figures 2 and 3
we have placed this return path below the origin-destination polyline to facilitate
visualization.
Figure 2. Screenshot of UAM flight path generator within Grasshopper for Rhino.
Figure 3. Corridor scenarios: direct routes (left), corridor above river (right).
The tool facilitates the comparison of different origin-corridor-destination
scenarios by offering a measure of the total distance flown by all drones in the
scenario (figures 2,3). This offers a means of comparing efficiencies when testing
different corridors (all other variables being held equal). As contemporary drones
have significant range limits, we also include a function which sorts out routes that
exceed a user-defined range limit. Routes that exceed the range limit are assigned
a different color in the visualization, shown here as magenta (figures 2,3).
To visualize the real-time flights of drones (and to later produce a simulation of
cumulative drone noise) we created an add-on script that animates drones ‘flying’
sequentially along their designated routes (figure 5). The drones are assigned an
average speed by the user. This add-on script incorporates a timer to release the
‘drones’ at regular intervals and to mark their advancement along their routes.
Upon returning from the destination to the droneport, the ‘drone’ is removed from
the list of active drones. We next used this simulation of urban drone flight to
create a simple UAM noise simulation.
76 F.P. ORTNER AND J. HUANG
2.3. KEY FUNCTIONS: ESTIMATING CUMULATIVE DRONE NOISE
Making strategic urban design decisions about the future of drone delivery means
not only considering the efficiency of drone airspace, but also thinking about its
impact on life on the ground. In a first step toward understanding drone delivery’s
impact on urban areas we used our drone delivery simulation described in 2.2 to
create a cumulative drone delivery noise simulation.
This simulator works using a model of outdoor noise propagation known as
Spherical Free Field Sound Propagation which calculates sound pressure level at
a variable distance given a known pressure/distance reading (figure 4).
Figure 4. Spherical free field sound propagation diagram and formula.
This equation provides the often-quoted rule of thumb that noise will decrease
by 6dB with each doubling of distance from the source (Beranek 1954). This
method of noise level calculation does not yet account for some key urban acoustic
phenomena like reflections. We hope to account for these functions in the future
and reserve further discussion for section 3 below.
To build a simulation from the Spherical Free Field Sound Propagation
equation we first ask the user to input a drone noise level in decibels. These
measurements are publicly available for some models of drones (Tinney 2018).
A second input required from the user is a grid of sample points at which we will
measure the cumulative noise level of the drones.
Figure 5. Screenshot of UAM noise simulator within Grasshopper for Rhino.
To simulate noise for a single drone we would calculate the noise level based
on the distance of the drone from each point in the sample grid. To produce a
MODELING UAM SCENARIOS FOR URBAN DESIGN 77
time-averaged noise simulation we would then measure the noise level at each time
state as the drone flies by, then sum these noise levels together using logarithmic
summation and average them based on the number of time samples.
However, our simulation requires estimating the urban noise from many drones
flying through the sky simultaneously. This requires measuring the noise level
contribution from all flying drones to all points on the sample grid, then running
the logarithmic sum of these noise levels. The cumulative noise of all the drones
at each sample point is then time averaged following the same method described
above.
Figure 6. UAM noise simulation: direct routes (left), vs. a multi-corridor scenario (right).
To visualize the numeric estimate of the noise created bya fleet of drones flying
over a flat area the tool has two outputs which are continuously written to each
sample point: (1) time-averaged decibel level for each sample point, and (2) an
RGB value based on a gradient from magenta to cyan. In figure 6 we have assigned
the color magenta to values ≥ 70dB and the color cyan to values ≤ 20dB. The
color gradient assigned to each sample point creates a ‘heat map,’ that gives users
an immediate overview of where drone noise is concentrated for a given airspace
scenario. An average of all the decibel levels simulated for each sample point is
also returned to the user (figure 6).
Our tool shows, not unexpectedly, that concentrating UAM flights at single
droneports or along corridors produces distinct noise ‘hotspots’ in the urban
plan. In figure 6 we compare noise heatmaps for two contrasting airspace
scenarios: in the image on the left, drones are allowed to fly directly to
their destinations after reaching cruise elevation; on the right, drones follow
an origin-corridor-destination path as described in 2.2, with a choice between
a corridor above the river or above an adjacent elevated railway. While the
multi-corridor scenario shown on the left produces a more distinct/concentrated
noise hotspot, we can see however that it has a lower average decibel level. This
lower average decibel level is due to the fact that the corridors have effectively
pulled drone traffic to the periphery of the study area. The rectangular study area
shown in figure 6 reflects an approximation of the masterplan area used in our
design studies. As discussed in greater detail in the section 3, the presence of
infrastructural sites on two sides of our site provided us with open space where
drone traffic could be diverted, minimizing impacts to noise-sensitive areas.
The spatial evaluation of urban drone noise maps based on intuitive readings
of heatmaps along with key airspace efficiency metrics supported a more precise
discussion of future UAM scenarios in our design tests. In the next section we
78 F.P. ORTNER AND J. HUANG
discuss the impact on design method and output of the tools we have presented
thus far in the context of an academic urban design studio.
3. Results and Conclusions
3.1. RESULTS: DESIGN STUDIO FOR UAM ON POST-INDUSTRIAL SITE
We tested our UAM simulation tools in an urban design studio at EPFL that
addressed more broadly the potential impact of the automated economy on a
largely vacant post-industrial district in Glasgow. We asked our designers to
develop a district master plan in parallel with an airspace design for drone
delivery supported by the tools presented in this paper. We observed two distinct
use-patterns associated with the tool.
A first use-pattern we would call, ‘speculative,’ or airspace-driven. In this use
pattern, designers first attempted to design an efficient UAM corridor system, and
then used this to organize their district plan, for example moving sensitive program
away from noisier areas.
A second use-pattern we would call, ‘assertive,’ or groundspace-driven. In
this use pattern the designer developed a general sense of the masterplan they felt
would work best on the site, and then established ‘quiet zones,’ where they asserted
that there should be minimal UAM noise impact. They then iteratively tested
corridor options till they were able to achieve a result adequate to their intentions.
An example of a design that asserted desired quiet zones as a pre-condition
for UAM airspace design is shown in figure 7. In this academic example the
drone noise simulation has been merged with a district noise map obtained from
an open-data source (SEPA 2016). While this design increases the total distance
flown by drones in our model, it minimizes noise impacts to central areas where
the designer hoped to introduce residential and educational programs. The design
shown in figure 7 has positioned the drone port and drone traffic near a highway
and rail corridor, where drone noise is masked by pre-existing traffic noise. This
design used an add-on tool we developed to merge pre-existing municipal noise
maps with our simulated UAM noise maps. This function is still in development
and exceeds the scope of this paper.
Figure 7. Design tests established ground-level quiet zones (cyan) using the UAM
model.(student work: Jean-Baptiste Clochet, Nicola Mahon, EPFL Media x Design Lab).
MODELING UAM SCENARIOS FOR URBAN DESIGN 79
We felt that the ability of the designers to take an assertive stance toward UAM,
by pre-emptively establishing quiet zones and iterating their designs to achieve
desired noise levels, was a validation of the potential of our preliminary UAM
model. In the future we believe that an expanded version of our tool will help
urban planners/designers in non-academic applications to come to the table with
aviation authorities and private companies and argue for airspace designs that best
anticipate the concerns of urban residents. This pre-emptive ability could facilitate
the design of UAM airspace in a manner that minimizes community dissatisfaction
before it occurs.
3.2. RESULTS: LIMITATIONS OF PRELIMINARY UAM MODEL AND
ON-GOING WORK
Future UAM airspace, according to our literature review, will likely be
volumetrically defined, with individual UAV checking in and out of zones as they
move through urban airspace. The corridor-based model as described in section
2.2, thus presents a simplification of the way future airspace will likely work.
While we feel that our corridor-based simulation is sufficient to preliminary urban
design studies, and that it is has provided valuable input in our design explorations,
we recognize that even for urban design applications a polygonal corridor model
can be over-restrictive. In our on-going work we are moving toward a volumetric
air-space model that would nonetheless still provide a simplified and global
airspace overview useful to urban designers and planners.
The noise simulation we have presented in this paper also has limitations
that we have recently explored in conversation with an acoustician. Spherical
propagation provides a strategically sufficient model of UAV noise when the
UAV is at cruising altitude and well-above elements of the built environment.
An effort at benchmarking is currently underway. However, when UAV are in
closer proximity to ground or architectural surfaces, reflection would significantly
increase noise from certain vantage points.
As the study area for our master plan design study was low-rise and largely
vacant, the impact of the ‘urban canyon’ noise effect was less important for our
design studies. Looking forward, as we would hope to produce a tool with wider
applicability, taking into account noise reflections would likely be a justified
next step. Additionally, the ability to recalculate UAM noise levels based on
the presence of proposed future buildings would be a valuable function for urban
designers.
3.3. CONCLUSIONS
Future UAM airspace designs will be produced by many actors, from public
institutions to private corporations, to citizen groups and designers. While no
single actor should dominate the debate over UAM organization, we feel that
designers of the urban realm should have an equal ability to advance their concerns
relative to aviation industry and aviation authorities. A UAM model created
specifically for urban designers would enable them to actively shape discussion
over airspace design instead of a taking on more limited reactive role to proposals
80 F.P. ORTNER AND J. HUANG
put forward by the aviation industry or aviation authorities.
The preliminary tool we have presented this paper lays out some first steps
toward producing a UAM model for urban planners/designers and provides a road
map for future efforts in this direction. Comparing tradeoffs between airspace
efficiency and ground-level noise provides an important quantitative feedback on
UAM design. As work continues on the tool, we hope to study ground-level
privacy impacts of UAM airspace designs as well as mobility network impacts.
While we designed our UAM tool for urban design applications, it will also
assist urban planners to reserve space for future UAM infrastructural needs, and
aviation authorities to route growing UAM traffic over urban terrain. We hope the
tool will support design of UAM systems that are complimentary to ground level
activities without causing excessive nuisance.
Acknowledgements
The authors would like to thank EPFL Media x Design Lab students and staff who
tested our tool.
References
“Scotland’s Noise, Scottish Environment Protection Agency (SEPA)” : 2016. Available from <
https://noise.environment.gov.scot/.> (accessed 30th April 2019).
Amazon Prime Air, : 2015, Revising the Airspace Model for the Safe Integration of Small
Unmanned Aircraft Systems, NASA.
Beranek, L.: 1954, Acoustics, McGraw-Hill, New York.
Cervantes, A. and Herrera, S.: 2019, The Drones Are Coming! How Amazon, Alphabet and
Uber Are Taking to the Skies, Wall Street Journal,25 October 2019(Business), na.
Curlander, J.C., Gilboa-Amir, A., Kisser, L.M., Koch, R.A. and Amazon Technologies, I.N.C.:
2017, Multi-level fulfillment center for unmanned aerial vehicles, U.S. Patent 9,777,502.
Garrett, B. and Anderson, K.: 2018, Drone methodologies: Taking flight in human and physical
geography, Transactions of the Institute of British Geographers,43(3), 341-359.
Ippolito, C., Krishnakumar, K., Stepanyan, V., Chakrabarty, A. and Baculi, J.: 2019, SAFE50
Reference Design Study for Large-Scale High-Density Low-Altitude UAS Operations in
Urban Areas, AIAA Scitech Forum, San Diego, California.
Jensen, O. B.: 2016, Drone City - power, design and aerial mobility in the age of “smart cities”,
Geographica Helvetica,71, 67-75.
Jiang, T. and Geller, J.: 2016, Unmanned Aircraft System Traffic Management: Concept of
Operation and System Architecture, Unmanned Aerial Vehicles and Remote Sensing,5(3),
123–35.
New York State Governor’s Office, : 2019, First-In-The-Nation 50-Mile Drone Corridor
Created, The Lansing Star,15 November 2019 edition, na.
Ortner, F. P. 2017, Metric as Mediator: Data-driven negotiation over Heathrow’s Noise
Landscape, in B. Boucsein (ed.), The Noise Landscape: A Spatial Exploration of Airports
and Cities, nai010 publishers, Rotterdam, 215-226.
Salleh, M., Chi, W. and Wang, Z.: 2018, Preliminary Concept of Adaptive Urban Airspace
Management for Unmanned Aircraft Operations, AIAA Information Systems-AIAA Infotech
@ Aerospace.
Taub, B.: 2019, Ideas in the Sky: Jonathan Ledgard Believes Imagination Could Save the World,
The New Yorker,September 23, 2019, 36-41.
Tinney, C. and Sirohi, J.: 2018, Multirotor Drone Noise at Static Thrust, AIAA Journal,56(7),
2816-2826.
