Fast-Time Simulations to Study the Capacity of a Traffic Network Aimed at Urban Air Mobility

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

On page 5 (lines 168) authors state that they apply “…. a general horizontal separation of 0.5 NM…” On page 6 (lines 209-210) they explain that their model includes “…no elements that cause further delays once the aircraft enrouted.” What does it mean for a pair of drones following the same flightpath at the same altitude? Are they separated solely by the departure interval (1 min) and the pre-departure blocking? In general, a more detailed description on how the authors modelled separations could be useful.

On page 5 (line 181) authors state that they generated the traffic data “…from the trend of estimated passengers through the two airports on 26 June 2023”. As this is a past date: why did authors rely on an estimation instead of using real count data? Weren’t those available?

In lines 182-184 authors add the information that the analyzed traffic data “…were extracted from the Aeronautical Information Regulation and Control cycles provided by Eurocontrol.” Is this the right reference talking about movement or passenger count figures?

In line 184 authors use the term “constant” to characterize certain traffic patterns. Is this meant in the sense of continuous traffic or traffic at a certain constant level?

 

In Figure 4 at page 6 there are no data points at the x-value of 21 o'clock. Does this mean that the values at 20 o'clock refer to the number of PAX in between 20:00 and 21:00?

In Table 2 authors list numbers of hourly movements for the two vertiports. They might add some explanation on how these numbers are related to the numbers of passengers being serviced at the airports LIMC and LIML. In addition, authors might explain how they generated the baseline schedule for the drones and how they determined the scheduled departure times (if those are not random).

In line 210 authors specify the saturation trigger they used: “The maximum delay has been set at 10 minutes.” It might help the reader to have some more detail at this point (the rationale behind the choice, some theoretical background information on e. g. Level of Service concepts in general, maybe a literature reference on airport capacity assessment methodologies).

 

On page 7 (lines 219-220) authors state that they used a multi-run simulation setup with variants to “to generate randomness to the system.” Applying randomness should somehow be reflected in the simulation result analysis, but there are no other parameters discussed beyond mean values. Authors might add other statistical measures.

In Figure 5 authors refer to what they call a “Final Capacity Diagram”. Is this meant to be like what is depicted in Figure 19? If so, authors could include kind of a small version of it in this figure.

 

On page 13 (lines 322-323) authors describe Figure 19 stating that for this diagram data was filtered “…omitting all those high arrival-departure pairs with low multiplicity”. Which threshold (in terms of frequency of occurrence) did they use to exclude data from being depicted?

 

On page 14 (lines 333-342) authors discuss their results based on the values given in Table 3. In the text they e. g. sum up certain values of Table 3 referring to “provincial” and “city center” vertiports. Readers are probably forced to turn back pages to comprehend author's conclusions. It could help to rephrase the paragraph or to add some information to Table 3.

 

Author Response

On page 5 (lines 168) authors state that they apply “…. a general horizontal separation of 0.5 NM…” On page 6 (lines 209-210) they explain that their model includes “…no elements that cause further delays once the aircraft enrouted.” What does it mean for a pair of drones following the same flightpath at the same altitude? Are they separated solely by the departure interval (1 min) and the pre-departure blocking? In general, a more detailed description on how the authors modelled separations could be useful.

The drones are separated through the imposed take-off conditions (1 minute between one take-off and the other). Therefore, in flight, these separations are always respected. This leads to a zero-sequencing delay. The 0.5NM horizontal separation were not imposed at model level, but it was simply verified after the simulations that there was no infringement of this separation. We better write this assumption at lines 254-258: Two consecutive drone takeoffs have a time separation of 1 minute, which corresponds to 0.5NM spatial horizontal separation and it is assumed that this distance is always respected, during the flight. The maximum delay has been set at 10 minutes, less than the maximum delay considered at the airport level (15 minutes) to take into account the very short range and urban service of the flights.

On page 5 (line 181) authors state that they generated the traffic data “…from the trend of estimated passengers through the two airports on 26 June 2023”. As this is a past date: why did authors rely on an estimation instead of using real count data? Weren’t those available?

The authors thank the Reviewer. We made a mistake in writing. We considered the data recorded on 26 June 2023. We corrected the sentence as follows (lines 224-225): This traffic was generated from the monitored value of passengers through the two airports on 26 June 2023, which showed an above-average number of movements and limited delays.

In lines 182-184 authors add the information that the analyzed traffic data “…were extracted from the Aeronautical Information Regulation and Control cycles provided by Eurocontrol.” Is this the right reference talking about movement or passenger count figures?

Yes, it is. We used the data provided by Eurocontrol (line 222).

In line 184 authors use the term “constant” to characterize certain traffic patterns. Is this meant in the sense of continuous traffic or traffic at a certain constant level?

It means at a certain constant level. We corrected the sentence at lines 222-223: In particular, five time slots with a constant level of traffic between 07:00 and 21:00 were considered.

In Figure 4 at page 6 there are no data points at the x-value of 21 o'clock. Does this mean that the values at 20 o'clock refer to the number of PAX in between 20:00 and 21:00?

It is assumed that the values at 20 o'clock refer to the number of PAX in between 20:00 and 21:00. We explained in lines 223-225 adding the following sentence: Figure 4 shows the values of each hourly range, so the number of passengers monitored between 20:00 and 21:00 is included in the x-value of 20.

In Table 2 authors list numbers of hourly movements for the two vertiports. They might add some explanation on how these numbers are related to the numbers of passengers being serviced at the airports LIMC and LIML. In addition, authors might explain how they generated the baseline schedule for the drones and how they determined the scheduled departure times (if those are not random).

We added the following sentence at lines 228-231: It has been assumed that only a very limited percentage of airport passengers will use UAM (up to 2%), being a mode of transport that can be chosen by business and first-class passengers. Starting from these hypotheses, the flights have been scheduled considering a trend similar to that of the movements recorded in the airports.

In line 210 authors specify the saturation trigger they used: “The maximum delay has been set at 10 minutes.” It might help the reader to have some more detail at this point (the rationale behind the choice, some theoretical background information on e. g. Level of Service concepts in general, maybe a literature reference on airport capacity assessment methodologies).

At the airport operations level, a flight is considered late when it departs 15 minutes late. We chose 10 minutes since this is a very short-haul and urban service and the threshold equal 15 minutes would have been excessive. We added the following sentence at lines 256-258: The maximum delay has been set at 10 minutes, less than the maximum delay considered at the airport level (15 minutes) to take into account the very short range and urban service of the flights.

On page 7 (lines 219-220) authors state that they used a multi-run simulation setup with variants to “to generate randomness to the system.” Applying randomness should somehow be reflected in the simulation result analysis, but there are no other parameters discussed beyond mean values. Authors might add other statistical measures.

We added the following sentence to explain the sentence, at lines 268-270: Randomness is automatically applied by AirTOp varying by a few minutes (before or after), the scheduled flights generated on the base scenario by the software itself.

In Figure 5 authors refer to what they call a “Final Capacity Diagram”. Is this meant to be like what is depicted in Figure 19? If so, authors could include kind of a small version of it in this figure.

Thank you for the suggestion. We changed Figure 5.

On page 13 (lines 322-323) authors describe Figure 19 stating that for this diagram data was filtered “…omitting all those high arrival-departure pairs with low multiplicity”. Which threshold (in terms of frequency of occurrence) did they use to exclude data from being depicted?

Simply we excluded the outlier, indeed we excluded all the data for which the capacity was not balanced: we changed the sentence at lines 372-373: The data were filtered by omitting all those points for which the capacity was not balanced were excluded.

On page 14 (lines 333-342) authors discuss their results based on the values given in Table 3. In the text they e. g. sum up certain values of Table 3 referring to “provincial” and “city center” vertiports. Readers are probably forced to turn back pages to comprehend author's conclusions. It could help to rephrase the paragraph or to add some information to Table 3.

Thank you for this suggestion: we added the second column of Table 3.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this paper, certain viable solutions are investigated to implement an Urban Air Mobility network in Milan, Italy, and authors analyze its influence on the airspace capacity with the method of simulation. Results seem to be interesting and this paper can be received if some revisions are done.

 

1. when using AirTOP software for simulation, it can be appropriate to show the simulation process with some figures;

2. For LIMI LIMC, please specify the full name of them in advance;

3. Whether only runway delay is enough to analyze the operation characteristics, and discuss the system capacity based on this;

4. For line180-181, it is recommended to explain the data characteristics in the simulation scenario, such as passenger  data, and the estimation method.

5. How to determine the accuracy of the simulation?

Author Response

In this paper, certain viable solutions are investigated to implement an Urban Air Mobility network in Milan, Italy, and authors analyze its influence on the airspace capacity with the method of simulation. Results seem to be interesting and this paper can be received if some revisions are done.

1. when using AirTOP software for simulation, it can be appropriate to show the simulation process with some figures;

We can’t provide any image of AirTOP, because the software owner doesn’t give permission to do so. Therefore, we have described better the software at line 136-140:

The simulation models developed for this study were created with the AirTOp simulator version 5.0.0 P2. The AirTOp simulation platform is an advanced gate-to-gate simulation tool, created for the design, modeling, and simulation of air traffic, both for the evaluation of traffic management in the airport environment and en route and approach.

And at the lines 152-167:

A simulation scenario is a set of elements necessary to represent the operational environment and/or the infrastructure being studied. The base scenario, for a simulation, reproduces the environment being measured in “standard conditions” such as:

• ICAO International Standard Atmosphere;
• No Wind;
• Visibility Conditions 1;
• Military areas not in use;
• Correct functioning of all systems;
• Air traffic control management rules and procedures agreed with the operational contact point and the client.

The verification of the performance of an operational scenario is the iterative process through which the ATM system, an infrastructure or a new operational concept can be evaluated. The verification process used is consistent with the Eurocontrol document [40] which provides for the definition of objectives; preparation of the validation plan; definition of simulation exercises; analysis of results; and development and distribution of conclusions.

2. For LIMI LIMC, please specify the full name of them in advance;

We specified the name at line 119: (i.e., Milan Linate, LIML, and Milan Malpensa, LIMC).

3. Whether only runway delay is enough to analyze the operation characteristics, and discuss the system capacity based on this;

We specify at line 252-258: In this study, only the runway delay has been considered because it was assumed a simplified land side with infinite capacity parking and there are no elements that cause further delays once the aircraft enrouted. Two consecutive drone takeoffs have a time separation of 1 minute, which corresponds to 0.5NM spatial horizontal separation and it is assumed that this distance is always respected, during the flight. The maximum delay has been set at 10 minutes, less than the maximum delay considered at the airport level (15 minutes) to take into account the very short range and urban service of the flights.

4. For line180-181, it is recommended to explain the data characteristics in the simulation scenario, such as passenger data, and the estimation method.

We considered the data recorded on 26 June 2023, which showed an above-average number of movements and limited delays and this data was considered for the baseline scenario. Then the traffic was increased until the saturation. We modified lines 217-225: A capacity study using simulation models involves the generation of baseline traffic, which increases until the system saturation. This traffic was generated from the monitored value of passengers through the two airports on 26 June 2023, which showed an above-average number of movements and limited delays. The traffic data analyzed were extracted from the Aeronautical Information Regulation and Control cycles provided by Eurocontrol. In particular, five time slots with a constant level of traffic between 07:00 and 21:00 were considered. Figure 4 shows the values of each hourly range, so the number of passengers monitored between 20:00 and 21:00 is included in the x-value of 20.

5. How to determine the accuracy of the simulation?

AirTOp provides theoretical output data, a theoretical capacity, based exclusively on the input data entered. The reliability of the results therefore depends on the validity and representativeness of the operational reality of the input data. The extreme accuracy of a tool such as AirTOp allows the transfer of the methodology to any urban reality. We added this comment in the Conclusions (lines 435-439).

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors,

in this paper you described and investigated the possibilities to implement a UAM network and to analyze its impact on the airspace capacity. Overall, the paper is very well structured, but I have some comments to be considered:

- even though you are using Milan, Italy as a test corridor, in the Abstract and Introduction (and in overall consideration), the study should be generalized (as the title suggests), with the projection of the idea on the Milan airspace;

- Figure 1 should be described more precisely;

- the structure of the paper at the end of the Introduction is missing;

- In Figure 2, general location points for readers unfamiliar with the Milan area should be added;

- also, noise reduction is noted in the paper, but there are only a few words regarding the limitations of autonomy and its influence on the proposed concept;

- after such a detailed presentation of the results, the proper discussion with a general comparison of the results, and their impact on the implementation of such a proposed concept is missing;

- Conclusions should be improved and expanded.

Author Response

in this paper you described and investigated the possibilities to implement a UAM network and to analyze its impact on the airspace capacity. Overall, the paper is very well structured, but I have some comments to be considered:

- even though you are using Milan, Italy as a test corridor, in the Abstract and Introduction (and in overall consideration), the study should be generalized (as the title suggests), with the projection of the idea on the Milan airspace;

We stressed this concept in the abstract at lines 18-24: However, the results are ideal due to the strong idealism of the system, which overlooks several factors, and they should be considered as the maximum limit that can be obtained. Despite this, the method presented in this article can also be adapted for other urban areas with high population density. In addition, the use of a simulation tool of this type allows, in addition to a numerical analysis, also a qualitative analysis of the network behavior in terms of traffic, thus highlighting the criticalities of the proposed systems.

We explained the concept in the Introduction (lines 126-127): The method described in the article can be adopted in any other metropolitan area, thanks to the extreme flexibility of the simulation methods.

and at lines 435-439:

AirTOp provides theoretical output data, a theoretical capacity, based exclusively on the input data entered. The reliability of the results therefore depends on the validity and representativeness of the operational reality of the input data. The extreme accuracy of a tool such as AirTOp allows the transfer of the methodology to any urban reality.

- Figure 1 should be described more precisely;

We added the following description at lines 80-87: Figure 1 shows the concept developed by the FAA that highlights the relationship between UAM, ATM, and UTM in different classes of airspace. Different corridors can be used depending on the type of operation and the type of aircraft, which therefore have specific access conditions. The FAA also states that a separation service within the corridors is not necessary since the latter is guaranteed by the operational characteristics of the corridors themselves. This article adopted the FAA concept of U-Space because it provides pre-established flight corridors based on the operation type by varying access conditions [29].

- the structure of the paper at the end of the Introduction is missing;

At the end of the Introduction, lines 128-133, we described the structure of the paper: Apart from this introductory paragraph, the rest of the document is organized as follows. Section 2 describes the simulation method used to evaluate the capacity of the air-space when drones are included. In this session, the characteristics of the studied area are also described. Section 3 describes the results. Section 4 discusses the results and their impact on the implementation of such a proposed concept with a general comparison of data. Section 5 presents conclusions and closing remarks.

- In Figure 2, general location points for readers unfamiliar with the Milan area should be added;

Thank you for your suggestion. We changed Figure 2.

- also, noise reduction is noted in the paper, but there are only a few words regarding the limitations of autonomy and its influence on the proposed concept;

Yes, it is true, but the topic is not in the aim of the paper. However, the authors know this aspect cannot be overlooked.

- after such a detailed presentation of the results, the proper discussion with a general comparison of the results, and their impact on the implementation of such a proposed concept is missing;

We added the Discussion section: The traffic network analyzed in this study was designed with two distinct configurations of vertiport infrastructure. The first configuration featured vertiports equipped with a single FATO (Final Approach and Takeoff Area), while the second configuration incorpo-rated a double FATO setup. This difference in vertiport design had a direct impact on the modeled take-off and landing procedures for drones, which were used to assess overall network capacity. Concerning the first configuration, Legnano and Busto Arsizio are the first two vertiports to reach saturation. While, in the second configuration, only Legnano vertiport is critical. The saturation traffic is about directly proportional to the number of FATOs, and the capacity of the vertiports has more than doubled from the first to the sec-ond configuration. A similar capacity is in provincial vertiports (i.e., Rho, Lainate, Legnano, and Busto Arsizio). Indeed, they have the same capacity, as the two vertiports in the city center (i.e., Porta Romana and City Life). This depends on how basic traffic flights were initially distributed among all vertiports. Therefore, the sum of the capacities of all the provincial vertiports and the two vertiports in the city center is equal to the sum of the capacities of the two airport vertiports. Future research should explore ground infrastructure simulations that are informed by a more detailed analysis of the selected vertiport sites.

Finally, it is possible to estimate the share of passengers who could use the UAM ser-vice:

• Milan Linate: the maximum capacity is 28 movements in configuration 1 and 60 in configuration 2, consequently the number of passengers is 56 and 120, respectively. Given that the maximum number of hourly passengers on the reference day is approximately 3,000, the UAM service can cover, as a minimum, 1.8% and 4% of the total traffic for the two configurations, respectively;
• Milan Malpensa: the maximum capacity is 42 movements in configuration 1 and 90 in configuration 2, consequently the number of passengers is 84 and 180, respectively. Given that the maximum number of hourly passengers on the reference day is approximately 8,000, the UAM service could cover, at a minimum, 1% and 2.5% of the total traffic for the two configurations, respectively.

Therefore, considering the maximum number of passengers per hour at the two air-ports as approximately 11,000, the hourly capacity of the entire network is equal to the sum of the capacities of the two airport vertiports (i.e., 70 and 150 respectively for the first and second configurations, respectively), and the whole network could handle, at a minimum, 1.26% and 2.72% of the total passenger for the two configurations, respectively. Concerning the total number of passengers on the reference day (i.e., approximately 113,000), the network at maximum capacity throughout the day can handle 1.72% (about 1,950) and 3.7% (about 4,200) of the total passengers.

- Conclusions should be improved and expanded.

We modified the Conclusions section: This study evaluated the feasibility of implementing a U-Space within a city context, such as Milan, to connect two airports with the city center and the most populated provinces. This context, given the considerable number of vertiports, is a futuristic scenario implementable under some conditions: 1) the availability of all U-Space services will be available and 100% integrated with the current services for the management of unmanned traffic, 2) all the aspects related to the regulation of the ecosystem will be mature and available, and 3) the social acceptance will be reached and the actual usefulness of the UAM understood.

The simulation in the AirTOp environment can reproduce the behavior of highly complex dynamic systems, such as an airport and the ATS network. Unlike analytical models, in this case, the functioning of the system is not mathematically modeled; in-stead, the system is replicated by modeling each actor involved using specific tools. Therefore, the simulation model is a simplified and virtual replica of a real system, capable of reflecting a set of characteristics deemed relevant to the study's objectives. AirTOp provides theoretical output data, a theoretical capacity, based exclusively on the input data entered. The reliability of the results therefore depends on the validity and representativeness of the operational reality of the input data. The extreme accuracy of a tool such as AirTOp allows the transfer of the methodology to any urban reality.

The traffic network designed and studied was proposed in two configurations of vertiport infrastructure. In particular, the first configuration featured vertiports equipped with a single FATO, while the second one with a double FATO. This difference affected the drone take-off and landing procedures modeled to assess the network capacity. The other relevant factors were the applicable separations, the traffic distribution during the day, and above all the lack of an effective land-side infrastructure in terms of parking stands and special procedures applicable during the flight (i.e., vectoring, route changes, and altitude, and waiting points, circuits, or hover points). Therefore, this study overlooked safety, security, obstacles, and spaces available to construct vertiports. All these factors could change the location of the vertiports, routes, procedures, and the final network capacity.

The key findings indicated that doubling the number of FATOs can effectively double the capacity of each vertiport, and by extension, the overall network capacity, allowing for a greater volume of flights and passengers. However, a dual FATO configuration also significantly increases associated costs. A larger area is required to accommodate additional parking stands, expanded passenger terminals, and enhanced facilities, all of which drive up construction, operational, and maintenance expenses. Moreover, traffic management becomes more complex, with implications for service providers that must be taken into account. Future work could explore simulations of ground infrastructure, incorporating a more detailed analysis of the selected vertiport sites, which might also necessitate revisions to air and ground procedures.

Author Response File: Author Response.pdf