*3.2. Vertiport Network in Support of Airports*

Establishing a vertiport network in the vicinity of airports and operating as first or last leg of a multi-modal trip to or from an airport may be convenient for the passenger and lucrative in terms of time-saving.

Placing a single vertiport of the network directly next to an airport requires the identification of constraints which might be locally different but since a lot of aerodromes are following (inter)national standards, they may be transferred and adjusted quickly. Based on the exemplary operating environment of *Cologne Bonn (Germany)* airport, ref. [91] developed a rating system considering passenger accessibility, obstacle clearance, noise impact on adjacent buildings, expandability, applicability and strategic availability in order to evaluate the potential of each identified vacant area adjacent to the airport. This included parking garages, parking fields and rooftops of an existing bus terminal and of a future hotel. Based on that rating system, ref. [91] prioritized the rooftop level of an adjacent parking garage which provided the best passenger accessibility and may enable an almost unhindered UAM operation. During this process, several requirements deemed crucial for successful integration including vertiport connection to existing transportation modes and the proximity to terminal buildings.

Similar but a more detailed analysis was conducted by [92] who used a 2019 LAX passenger survey as primary data set to determine the optimal vertiport location and network size based on the passengers' selected top ten origin destinations in the area of *Los Angeles (U.S.)*. Restricted airspace boundaries prohibiting overflying or restricting the placement of vertiports are taken into account. A mode choice model with varying assumptions for the in-vehicle travel time, additional shuttle time and the out-of-vehicle time was created to capture a traveler's mode-choice to and from the airport. The demanddriven vertiport placement methodology by [93] was used. As a result, a mixed logic model with different parameters such as travel time, travel cost and the value of time is created. Together with the Fuzzy C-means clustering method which places a certain number of clusters in a specific area, ref. [92] concluded with an optimally placed vertiport set of three network sizes: 50, 75 and 100 vertiports. Those vertiports located adjacent to LAX attract zero demand due to the short travel distance or airspace restriction, whereas the vertiport in LA downtown expected the highest demand.

Of contrast, for the 25 vertiport network in *Dallas Fort-Worth*, the vertiport adjacent to Dallas Fort-Worth airport shares 28% of the total UAM operation and resulted into the most demanded node [94]. Taken into account peak and off-peak demand distribution, an average vehicle load factor of 67 and by using a M/M/1 queuing model together with a target waiting time of four minutes, a 76% utilization factor for a FATO is proposed in order to be able to absorb operational deviations. A FATO count per peak, off-peak and average hours was calculated and concluded with a required number of 27.5 FATOs for the vertiport located at the airport in order to serve peak hours. Operating multiple pads will require sufficient separation on the ground (over 200 ft (61 m)) based on helicopter operations) and separate arrival and departure paths with individual obstacle-free protection surfaces.

The vertiport network in *Dallas Fort-Worth* assumed a 5% shift of long distance transportation, but still intra-city, into the air while considering early operations of UAM [94]. Ref. [92] derived a potential 3.6% market share of UAM operating mainly as an airport shuttle and providing trips from and to LAX. To achieve this, a vertiport network size of 75 vertiports is required. For comparison, ref. [95] predicted a 0.5% mode share for airport shuttle and air taxi operations in the whole U.S.

Since operating in airport environment often leads to operating in controlled airspace with multiple other airspace users, a safe separation has to be maintained throughout the entire operation. Ref. [96] investigates different route designs for VTOL aircraft operating as an airport shuttle in a non-segregated airspace inside the terminal radar approach control (TRACON) airspace of *Tampa (U.S.)*. By using a Rapidly Exploring Random Tree optimization algorithm, those trajectories with minimum design costs and sufficient distance to manned operations, obstacles and ground are being selected. A user-specified distance was set to 25 ft (7.6 m) which increased incrementally by 25 ft (7.6 m). Based on those selected routes, possible vertiport locations are determined. For the airport and TRACON airspace of Tampa (U.S.), three vertiport locations, two inside airport area and one outside, were found. The algorithm identified 100 ft (30.5 m) being the largest available distance for those two vertiports located inside which does not provide sufficient distance of terrain and manned aircraft. Therefore, "[. . . ] no acceptable airspace volumes could be found that would be permanently available for VTOL trajectories under current operating conditions" [96] for the selected airport (layout) in *Tampa (U.S.)*.

Adding environmental constraints, uncertainties and passenger interaction to the operation of individual vertiports located inside a UAM vertiport network, different vertiport layout and performance capabilities might be required to serve "nominal" demand [97,98]. An airport shuttle network in the *Washington D.C. (U.S.)* area was analyzed by [98] in regard to changing performances of vehicle speed, boarding time, vertiport operations times and arrival demand. A full set of requirements including historic travel demand, location constraints, capacity of vertiports, number of vehicles and charging limitations are considered. Additionally, the vertiport network "shall emit Day Night Average Sound Level (DNL) less than or equal to 65 dB", "[. . . ] shall limit vehicles arriving at vertiports from waiting more than 20 min for an available landing pad" and "[...] system shall provide passenger transportation with 95% flights being within 5 min of expected time" [98]. The deterministic simulation concluded with a five node vertiport network, two FATOs and two parking spaces each and 70 vehicles in total being able to serve the demand of high value travelers. Using normal distributions for vehicle speed, boarding time and vertiport operations time and a Poisson distribution depicting passenger arrivals, the required number of landing pads increased from two to three in order to achieve same orders of throughput. In contrast to [98], ref. [97] conducted a sensitivity analysis for several variables (e.g., arrival/departure service time at pad and stall) by applying a lognormal distribution in order to evaluate the impact on vertiport capacity and operational efficiency (for additional details see [99]).
