Analysis of an autonomous driving modular bus system

ScienceDirect Available online at www.sciencedirect.com Transportation Research Procedia 58 (2021) 181–188 2352-1465 © 2021 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering 10.1016/j.trpro.2021.11.025 10.1016/j.trpro.2021.11.025 2352-1465 © 2021 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering Available online at www.sciencedirect.com ScienceDirect Transportation Research Procedia 00 (2019) 000–000 www.elsevier.com/locate/procedia 2352-1465 © 2020 © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering 14th Conference on Transport Engineering: 6th – 8th July 2021 Analysis of an Autonomous Driving Modular Bus System Guillem Romeaa,*, Miquel Estradaa aCivil Engineering School of Barcelona. Universitat Politècnica de Catalunya – BarcelonaTECH, C. Jordi Girona 1-3, 08034 Barcelona, Spain Abstract One of the main complaints about the performance of the bus systems is the time lost at recurrent stops along the route, that results in a low bus speed service. Among the potential measures to speed up the bus service, the introduction of the Autonomous Driving Modular Bus, able to adjust the transport unit by coupling or decoupled pods, will represent a new paradigm in the provision of service. This concept is able to maintain high stable cruising speeds to the transport unit, decoupling only one pod with passengers that may alight at the next stop. This paper analyze the potentialities of this concept in terms of user travel times and operationg cost. This work also features the formulation and analytical models to calculate the performance and oeprational variables, implementing them into two existing bus lines in Barcelona. © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering Keywords: Bus, diesel, hybrid, electric, autonomous dirving modular, expenses 1. Introduction The current commercial speed of buses in crowded areas (8-15 km/h, Vuchic, 2013) is significantly lower than other public transportation modes in competition (15-40 km/h in tramways, 24-55 km/h in subways). There are several strategies to speed up buses along the routes based on segregated lanes, double lanes, multiple boarding platforms at stops, traffic light synchronization, etc. (see Kittelson and associates, 2013). Nevertheless, the travel time savings obtained are still marginal and require huge investments to deploy right of way measures or technological devices. In fact, these measures do not dramatically change the operation of buses: every transit vehicle, with passing passenger onboard, is still obliged to loose cruising speeds before and after a stop location, and is held at the stop facility during dwell time. The major challenge for bus operation would be whether the time lost performing these processes at intermediate stops can be removed from the bus motion. It means that bus services may resemble taxi systems, offering a direct door-to-door service from origins to destinations without intermediate stops. However, one of the innovation concepts that is gaining momentum and could meet this goal is the Autonomous Driving Modular Bus. With this new * Corresponding author. E-mail address: guille[email protected] Available online at www.sciencedirect.com ScienceDirect Transportation Research Procedia 00 (2019) 000–000 www.elsevier.com/locate/procedia 2352-1465 © 2020 © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering 14th Conference on Transport Engineering: 6th – 8th July 2021 Analysis of an Autonomous Driving Modular Bus System Guillem Romeaa,*, Miquel Estradaa aCivil Engineering School of Barcelona. Universitat Politècnica de Catalunya – BarcelonaTECH, C. Jordi Girona 1-3, 08034 Barcelona, Spain Abstract One of the main complaints about the performance of the bus systems is the time lost at recurrent stops along the route, that results in a low bus speed service. Among the potential measures to speed up the bus service, the introduction of the Autonomous Driving Modular Bus, able to adjust the transport unit by coupling or decoupled pods, will represent a new paradigm in the provision of service. This concept is able to maintain high stable cruising speeds to the transport unit, decoupling only one pod with passengers that may alight at the next stop. This paper analyze the potentialities of this concept in terms of user travel times and operationg cost. This work also features the formulation and analytical models to calculate the performance and oeprational variables, implementing them into two existing bus lines in Barcelona. © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 14th Conference on Transport Engineering Keywords: Bus, diesel, hybrid, electric, autonomous dirving modular, expenses 1. Introduction The current commercial speed of buses in crowded areas (8-15 km/h, Vuchic, 2013) is significantly lower than other public transportation modes in competition (15-40 km/h in tramways, 24-55 km/h in subways). There are several strategies to speed up buses along the routes based on segregated lanes, double lanes, multiple boarding platforms at stops, traffic light synchronization, etc. (see Kittelson and associates, 2013). Nevertheless, the travel time savings obtained are still marginal and require huge investments to deploy right of way measures or technological devices. In fact, these measures do not dramatically change the operation of buses: every transit vehicle, with passing passenger onboard, is still obliged to loose cruising speeds before and after a stop location, and is held at the stop facility during dwell time. The major challenge for bus operation would be whether the time lost performing these processes at intermediate stops can be removed from the bus motion. It means that bus services may resemble taxi systems, offering a direct door-to-door service from origins to destinations without intermediate stops. However, one of the innovation concepts that is gaining momentum and could meet this goal is the Autonomous Driving Modular Bus. With this new * Corresponding author. E-mail address: [email protected] 182 Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 2 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 concept, travel times could be reduced considerably while the service quality could be upgraded. This paper is aimed at analyzing the potentialities of this concept by means of an analytical model. This model consists of several compact formulas to estimate both the user performance (travel time) and operational cost of the service. The model developed is applied to low and high demanded routes in the current Barcelona Bus Network, so the implementation of this new system will be analyzed. The less demanded line corresponds to the V5 (Zona Franca-Avinguda Pearson), while the crowded one to the H10 (Plaça de Sants-Olímpic de Badalona). 2. Concept of Autonomous Driving Modular Bus Buses are still conceived as a classical configuration of one driver leading the bus and slowing at every stop that passengers request. In order to understand the meaning of this new technology, we must see beyond the concept of bus that we own. This new system is developed with the idea of not stopping at traffic lights neither at bus stops, with totally electric vehicles and, as its name suggest, driverless. First, no-stopping at traffic lights is not anything new. Many on-streets public transport, like the Barcelona, Zaragoza, or Granada’s tramways, rarely stop at them since activated traffic light priority measures are deployed (Estrada et al., 2009). This idea should be tested in buses with the aim to ease bus circulation in detriment of the private vehicle. Of course, this requests a complete restructuration of the current traffic lights configuration. Second, this technology changes drastically our conception of buses. Nowadays, the word “bus” makes us thing about a large and indivisible vehicle to convey people. But the “modular” bus suggests something different; the current bus will be transformed into a convoy made of n-pods that will be able to couple and uncouple freely, like a train with its wagons. The key idea in this conception is that pods will be detached from the convoy to give service at the bus stop, while the other part of the convoy continues its journey. A brief scheme of the concept is shown in Fig. 1. Fig. 1. Operational scheme. On the other side, there will always be a pod ready at the bus stop to depart when it detects that the convoy is approaching. This new pod will be attached to the front of the set, and the procedure will be repeated at the next bus stop. This ability of coupling and uncoupling gives passengers the chance of not stopping at every bus stop, as they will be able to move freely between pods when the convoy is running. In addition, it will be possible to couple and uncouple n-pods per stop depending on the demand that needs to be satisfied, and this ability will also ease the job when it comes to deploy more pods when the service requires it. Finally, these pods will be driverless and their traction will be based in electric batteries. Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 183 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 3 3. Model The model proposed has some simplifications to make the estimation of performance variable easier to reproduce and understand. Despite this, the results obtained are not far from the real ones (Daganzo, 2009), and can therefore be taken as valid. These simplifications are to assume a uniform demand through all the line, and an equidistance spacing between stops. The model presented aims to reduce the total cost of the system (ZT), as the sum of the agency’s (ZA) and user’s (ZU) costs (Eq. 1). 𝑍𝑍𝑍𝑍𝑇𝑇𝑇𝑇=𝑍𝑍𝑍𝑍𝐴𝐴𝐴𝐴+𝑍𝑍𝑍𝑍𝑈𝑈𝑈𝑈 (1) 3.1. Agency’s Cost The formula to determine all the cost incurred by the agency (Equation 2) is a combination of different factors that are introduced along the following subsections. 𝑍𝑍𝑍𝑍𝐴𝐴𝐴𝐴= €𝐿𝐿𝐿𝐿·𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇+ €𝑉𝑉𝑉𝑉·𝑉𝑉𝑉𝑉+ €𝑀𝑀𝑀𝑀·𝑀𝑀𝑀𝑀+ €𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 ·𝑀𝑀𝑀𝑀+ €𝐵𝐵𝐵𝐵·𝐶𝐶𝐶𝐶·𝑀𝑀𝑀𝑀 (2) 3.1.1. Infrastructure Cost (€L) and Route Length (LT) These firsts terms of the equation do not vary regardless of the bus type: diesel, hybrid or electrical. The proxy €L represents de infrastructure deterioration, expressed in [€/km-h], and the variable LT is the route length expressed in [km]. 3.1.2. Unit Vehicle Distance Cost (€V) and Distance Covered by the Fleet (V) The unit vehicle distance cost represents the energy cost per kilometer to run each pod plus its maintenance cost. It is expressed in [€/veh-km] and its values are (Estrada et al, 2021). Table 1. Estimation of the unit distance cost, €V [EUR/veh-km] Concept Value Energy Cost 0,036 Pod Maintenance 0,66 € V Total 0,696 It has been estimated that the pod’s energy consumption will fall between the consumption of an electric car and bus. Considering a Tesla Model S (5 meters long) with an average consumption of 0,20 kWh/km (Motorpasion, 2019), and a 1,00-1,40 kWh/km as an average current standard bus (12 meters) consumption (Estrada et al, 2021), a pod energy consumption of 0,30 kWh/km will be considered. Then, multiplying this number by the electricity price (0,1199 €/kWh Spain average in April 2020) the energy cost is obtained. Regarding the maintenance cost, the value has been taken directly from Estrada et al (2021). On the other side, the distance covered by the fleet per hour (V) [veh-km/h] is calculated with the following equation. 𝑉𝑉𝑉𝑉=𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇 𝐻𝐻𝐻𝐻· #𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝/𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐 (3) Where “H” represents the headway [hours]. 184 Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 4 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 3.1.3. Unit Vehicle Temporal Cost (€M) and Number Vehicles (M) The third term of the sum corresponds to the agency expenses related to temporal units, expressed in [€/veh-h], and the number of pods serving the line (M). These temporal expenses are determined in the following table. Table 2. Estimation of the unit temporal cost, €M [EUR/veh-h] Concept Value Driver Salary 0,00 Vehicle Amortization 1,587 Insurances, Technicians and Engineers 11,83 € M Total 13,417 The driver salary disappears because of the autonomous vehicles. Regarding the vehicle amortization, easymile (pod manufacturer) builds 4 meters long pods. So, considering the average acquisition price per meter of electric, hybrid and diesels buses according to Estrada et al (2021), an acquisition price of 100.000 €/vehicle will be approximated. If each pod runs for 15 years, 300 days/year and 14 hours/day, the value is obtained. The insurances and other salaries are also taken directly from Estrada et al (2021). On the other side, the number of pods needed per line are found with Eq. 4. 𝑀𝑀𝑀𝑀=�𝑉𝑉𝑉𝑉 𝑐𝑐𝑐𝑐𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 +𝑛𝑛𝑛𝑛·𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇 𝑝𝑝𝑝𝑝�+ (4) Where “n” represents the number of pods stopped at each bus stop, 𝑣𝑣𝑣𝑣𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 stands for the commercial speed [km/h], and “s” is the separation between two bus stops [km]. Through all this model, “n” will be considered 1. Please, note that the second quotient represents the total number of stops thanks to the simplification of equidistance spacing between stops. 3.1.4. Unit Vehicle Charging Facility Cost (€ER) The charging of pods will be performed during the 10 hours per day that they do not run. Because obtaining a reference value is quite complex, it has been taken the one from Estrada et al (2021) for electric buses with a slight modification. Now, it will be considered that each charger can supply energy to two pods simultaneously, and its lifespan should be 30 years, 300 days/year and 10 hours/day. This parameter is expressed in [€/veh-h]. Table 3. Expenses of the Charging Facility. Concept Value Facility Amortization (€/day) 9,60 Facility Maintenance (€/day) 10,00 Pods per Charger 2 € ER Total (€/veh-h) 0,98 3.1.5. Unit Temporal Battery Cost (€B) and Battery Capacity (C) This new technology must satisfy that the battery endures for all day run without charging. Of course, a different charging scheme must be implemented with charging points along the road, but to simplify the model, it will be considered that pods can only be recharged at depots. Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 185 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 5 First, the term €B represents the battery cost expressed in [€/kWh-h] considering a battery lifespan of 4 years (Estrada et al, 2021). What is more, Estrada et al (2021) also suggests recommends using a battery price of 400$/kWh. So, if the pod runs 300 days/year and 14 hours/day, “€B” stands for a value of 0,021 €/kWh-h. The Battery Capacity, expressed in [kWh], is something to be determined with the results shown at the end, as it is totally related to the headway (H), stop spacing (s) and the number of pods per convoy. However, the pod’s energy consumption per day is found in Eq. 5, and it satisfy the Battery Capacity (C) requirement. 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑛𝑛𝑛𝑛𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑛𝑛𝑛𝑛=𝑣𝑣𝑣𝑣𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁�𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 ℎ�·𝑓𝑓𝑓𝑓𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑁𝑁𝑁𝑁𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐�𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘ℎ 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘�·14 ℎ 𝑝𝑝𝑝𝑝𝑑𝑑𝑑𝑑𝑐𝑐𝑐𝑐· (1 + 𝑆𝑆𝑆𝑆𝐹𝐹𝐹𝐹𝑑𝑑𝑑𝑑𝑐𝑐𝑐𝑐𝑁𝑁𝑁𝑁𝑝𝑝𝑝𝑝𝐹𝐹𝐹𝐹) < 𝐶𝐶𝐶𝐶 (5) The consumption factor is 0,30 kWh/km, the pod runs 14 hours/day, and a 20% safety factor has been taken, which is enough to cover extra consumptions like accelerations, air cooling and slopes. Finally, the commercial speed (or net speed) is approximated with the following equation. 𝑣𝑣𝑣𝑣𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 ≈𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇 �𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇 𝑣𝑣𝑣𝑣𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶�+(𝐻𝐻𝐻𝐻+𝜏𝜏𝜏𝜏)·�𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇 𝑠𝑠𝑠𝑠·1 #𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑠𝑠𝑠𝑠/𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐𝑣𝑣𝑣𝑣𝑝𝑝𝑝𝑝𝑐𝑐𝑐𝑐+1�− (6) Where 𝑣𝑣𝑣𝑣𝐶𝐶𝐶𝐶𝐹𝐹𝐹𝐹 represents de cruising speed of the pod [km/h], and “τ” the time lost by each pod for breaking and accelerating at a bus stop in comparison with a non-stopping one [hours]. 3.2. User’s Cost When users travel, they spend time and money. The money side is fully covered with the ticket price, but one way to transform this time into monetary units is using the so-called Value of Time, which takes into account the user salary. The formula used to determine the user’s cost is the following one, where Λ represents the hourly demand expressed in [pax/h], β the value of time and the term into brackets is the expected door-to-door travel time of a single passenger. 𝑍𝑍𝑍𝑍𝑈𝑈𝑈𝑈=𝛬𝛬𝛬𝛬·𝛽𝛽𝛽𝛽·(𝐴𝐴𝐴𝐴+𝑊𝑊𝑊𝑊+𝐼𝐼𝐼𝐼𝑉𝑉𝑉𝑉𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼) (7) 3.2.1. Value of Time (β) This first parameter of the equation is crucial because it translates the time spend on public transport into monetary terms. Its units and value are [€/h] and 12,50 €/h, respectively. This average value comes from considering a salary of 2.000 €/month and working 40 hours per week. 3.2.2. Access and Exit Time (A) The first and last part of every journey by public transport corresponds to the walking time to the bus stop and to our destination once we get off the transport. Because one of the simplifications is to consider an evenly distributed demand, the expected access and exit time are the same, so the term “A”, [hours], represents the sum of both. 𝐴𝐴𝐴𝐴=𝑠𝑠𝑠𝑠 4 𝑐𝑐𝑐𝑐𝑤𝑤𝑤𝑤+𝑠𝑠𝑠𝑠 4 𝑐𝑐𝑐𝑐𝑤𝑤𝑤𝑤=𝑝𝑝𝑝𝑝 2·𝑐𝑐𝑐𝑐𝑤𝑤𝑤𝑤 (8) Where “vw” is the walking speed of the pedestrian [km/h]. 186 Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 6 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 3.2.3. Waiting Time (W) The waiting time includes factors like service regularity, traffic jams, and even smart-phones applications that tells the remaining time for the vehicle to come. Nevertheless, to simplify the calculous, considering a perfect regularity and no use of mobile phones, and with the aid of the same simplification used in 3.2.2., the expected waiting time is half the headway. 𝑊𝑊𝑊𝑊=𝐻𝐻𝐻𝐻 2 (9) 3.2.4. In-Vehicle Travel Time (IVTT) This is usually the bigger time when travelling by public transport. So, if an improvement of the service offered needs to be done, it is imperative to reduce the time spend by the user inside the bus. In this particular case of the autonomous and modular bus, the user will only perceive an acceleration, travelling at constant speed, and breaking time. So, the equation used is shown hereunder. 𝐼𝐼𝐼𝐼𝑉𝑉𝑉𝑉𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼=�𝑙𝑙𝑙𝑙 𝑐𝑐𝑐𝑐𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶�+(𝜏𝜏𝜏𝜏) (10) Where “l” represents the length travelled by the user inside the pod (which has been considered as LTotal/4 to be conservative) in [km], and “τ” the time lost by a pod to accelerate and brake in comparison with a non-stopping one. 4. Model Application in V5 and H10 Lines The Autonomous Driving Modular Bus has not been implemented yet in any city as a conventional mode of transport. Thus, these chapter will not study any real configuration but to give the optimal parameters to obtain the best configuration and minimize the total cost of the system (ZT). These “key” parameters are the headway (H), the stop spacing (s), and number of pods per convoy. In addition, it will be also possible to determine the battery capacity (C). Besides the battery capacity requirement explained in 3.1.5., any public transport service must also satisfy the occupancy (O) one. It means that the service provided must be enough to cover the whole demand, otherwise there would be users that will not be able to go aboard. The occupancy is found with Eq. 11. 𝑂𝑂𝑂𝑂=𝑙𝑙𝑙𝑙 𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇·𝛬𝛬𝛬𝛬·𝐻𝐻𝐻𝐻 ≤𝐶𝐶𝐶𝐶𝑝𝑝𝑝𝑝 (11) “Cp” represents de pod capacity [pax/veh], with a value of 15 pax/pod according to easymile. Finally, and before getting into detail with the results, it must be stated that a minimum and maximum threshold on those key parameters have been adopted to simulate a real solution, and not letting the optimization fall into an unfeasible one. Table 4. Optimization thresholds. Concept V5 Line H10 Line Headway (minutes) 5-15 1-7 Stop spacing (km) 0,30-0,75 0,30-0,75 Minimum pods/convoy 2 3 Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 187 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 7 4.1. Input parameters The set of parameters (Guida et al, 2018; Estrada et al, 2021) used are presented below. Table 5. Input parameters. Concept V5 Line H10 Line Line Length (L Total ) km 15,20 23,70 Demand (Λ) pax/h 396 6.247 Length covered by Users (l) km 3,80 5,93 Free Speed (v Cr ) km/h 40,00 Maximum Speed (v Max ) km/h 50,00 Acceleration (a) m/s2 1,30 Value of Time (β) €/h 12,50 Pedestrian Walking Speed (v walk ) km/h 4,00 Infrastructure Cost (€ L ) €/km-h 7,61 Unit Vehicle Distance Cost (€ V ) €/veh-km 0,696 Unit Vehicle Temporal Cost (€ M ) €/veh-h 13,417 Unit Vehicle Charging Facility Cost (€ ER ) €/veh-h 0,98 Unit Temporal Battery Cost (€ B ) €/kWh-h 0,021 4.2. Results After combining all the equations and numbers seen up to now, and doing an optimization process with the main aim to reduce the total cost of the system (Eq. 1), the following optimal parameters are found. Table 6. Optimization results. Concept V5 H10 Headway (H) min 8,00 1,50 Stop spacing (s) km 0,73 0,30 Battery Capacity (C) kWh 63,96 118,03 Pods/Convoy - 2 4 Total Pods (M) veh 39 243 Occupancy (O) pax/veh 13,20 39,04 Agency’s Costs (ZA) €/h 888,40 6.922,44 Users’ Costs (ZU) €/h 1.266,63 15.702,83 Travel Time per Trip min 15,35 12,07 It can be clearly seen that the occupancy is always below the convoy capacity. What is more, all convoys have one extra pod in case any user wants to get off at the next bus stop. 4.3. Comparison with Current Configuration If the optimal results obtained are compared with the current bus configuration in those lines (run by diesel, hybrid, and electric buses), the user’s and total costs suffers a huge downfall. It must be clarified that the V5 is served with standard buses (12 meters), while H10 with articulated ones (18 meters). 188 Guillem Romea et al. / Transportation Research Procedia 58 (2021) 181–188 8 Romea, G., Estrada, M. / Transportation Research Procedia 00 (2019) 000–000 Figure 2.1. User’s cost per type of vehicle. Figure2.2. Relative Total Cost to Diesel vehicles. As reflected in both figures, user and total costs are reduced when technology evolves to greener vehicles. In addition, one must notice the tremendous reduction the Autonomous Driving Modular Bus could bring to the system if it is implemented. However, it has been seen that the agency cost would increase considerably. In this sense, the current configuration for the H10 line run by electric buses, produces expenses up to 2.228 €/h, while the optimal one for the autonomous driving modular bus provides an agency cost of 6.922 €/h. This occurs due to the large number of pods needed in comparison with the number of buses deployed today. 5. Conclusions It is inevitable to express the necessity for developing a more efficient and sustainable public transport. The results express it themself; as technology upgrades, the reduction in users’ and total costs is significant. And not only that, but in environmental and health terms too. The results also demonstrate that low-demanded lines do not perceive that change in the technology as much as the high-demanded ones. This means that for the low ones, the improvement in technology is not a key factor when it comes to determine the optimal type of bus (diesel, hybrid…). Nevertheless, the high-demanded lines experience a contrary effect; the need for a better and sustainable bus type rises in effigy while the total costs are more than halved. Thus, a first conclusion is the good behavior this new technology would have in high-demanded lines. Although the results displayed demonstrated the high efficiency it presents, the model has been applied to a very particular case. So, it needs further study in other environments and situations, as well as an exhaust research in different operational schemas and charging itineraries within the lines, which will come in handy to reduce costs even more. Finally, this further study also needs to focus on the elongation of these lines through the suburbs, which will improve the territorial cohesion with the peripherical neighborhoods and will connect them in a fast and efficient way. References Daganzo, C. F., 2009. Structure of competitive transit networks. Transportation Research Part B: 434-446. Motorpasión, 2019. Probamos el Tesla Model S 100D, el coche eléctrico que soñaba con ser un deportivo. Accessed 4/12/2021. Retrieved from https://www.motorpasion.com/pruebas-de-coches/tesla-model-s-100d-prueba Estrada, M., Mensión, J., Salicrú, M., Badia, H., 202x. Charging operations in battery electric bus systems considering fleet size variability along the service. Transportation Research Part C. Emerging Technologies. Under review. Tarifa Luz, 2020. Tarifa de Luz a Tiempo Real en España. Accessed 4/2020. Retrieved from https://tarifaluzhora.es/ Easymile. Accessed 5/2020. Retrieved from https://easymile.com/ Easymile. Accessed 5/2020. Retrieved from https://easymile.com/solutions-easymile/ez10-autonomous-shuttle-easymile/ Transport Metropolitans de Barcelona (TMB), 2018. Pressupostos 2018 de TMB. Accessed 5/2020. Retrieved from https://www.tmb.cat/documents/20182/89894/Pressupostos+2018_CA/a6b7ef8e-3428-488e-9254-9d5fa95038e8 Guida, U., Estrada, M., Amat, C., Cuevas, V., 2018. Effect of electric buses on the network operations. Deliverable D53.5. Projected funded by the European Comission.