INTERNATIONAL European Friedrich-List-Prize Internationales Verkehrswesen (72) 3 | 2020 60 Development of electromobility services Electromobility, Charging infrastructure, Charging planning The widespread of environmentally friendly drivetrains and alternative fuels is expected in the upcoming decades. Therefore, this research was done to aid the alteration from a conventional car to electric cars to fit the extant transport system and electrical network. I have developed novel operational methods for electromobility services, including charging station locating and charging planning methods. The research was conducted from a system and process-oriented point of view. The results may contribute to facilitate and prepare the alteration of the transport system. Bálint Csonka N owadays, the environment and the emission gained huge attention during the planning and the operation of transport. Accordingly, the widespread of environmentally friendly drivetrains and alternative fuels is expected in the upcoming decades. Electrification is the most prominent option among the many alternative technologies because of its advantageous characteristic (e. g., durability); however, other concerns also arise (e.g., Pan-European transportation matters For the 16th consecutive time the European Platform of Transport Sciences - EPTS - awards the “European Friedrich-List-Prize”. This prize, dedicated to young transport researchers, is named to honour the extraordinary contributions of Friedrich List, the visionary of transport in Europe of the 19th century, being a distinguished economist and respected transport scientist committed to the European idea. The European Friedrich-List-Prize is awarded for out-standing scientific papers in each of the categories Doctorate paper and Diploma paper, addressing topics in the transport field within a European context. The award will be conferred during the 19th European Transport Congress at the University of Maribor, Slovenia, on 7 October 2021. The results will be introduced both in the “Internationales Verkehrswesen” November issue and online at www.international-transportation.com. In the following you find a random selection of this year’s submissions summarized in drafts. Source: University of Maribor European Friedrich-List-Prize INTERNATIONAL Internationales Verkehrswesen (72) 3 | 2020 61 environmental effect of batteries). In many countries, various actions have been taken to support electrification in road transport, namely the use of electric vehicles, the deployment of charging infrastructure, and the development of information and communication technologies. The operation characteristic differs from the habitual and may cause the aversion of new technologies [1], which may be mitigated by applying intelligent and personalised information systems. More and more charging units are needed to serve the growing energy demand of electric cars at the same service level. The locating problem of charging stations arises, especially in the early phase when the deployment and operation of charging stations cannot be done on a market basis. Therefore, the characteristic of the charging network (e.g., location, charging power) has a significant influence on the spread of electric car use. As well as an intelligent transportation system manages the mobility demand, the additional energy demand should be managed by novel information services. Therefore, a charging planning method was elaborated to mitigate the adverse effect of charging demand on the electrical network. In this paper, the main findings of my PhD research are summarised with a focus on three key areas: electric car user information service, charging station locating and charging planning. Integrated information service At first, the main differences between electrical car use and conventional car use were identified. The differences may cause inconvenience and aversion for users. The main differences were limited range, charging, the highest purchase price. The functions of the information service were derived from the differences to mitigate the adverse effects. The functions are as follows: •• Support for new vehicle selection: information on electric cars is provided before purchase. Furthermore, the information service may evaluate electric vehicles considering the operation cost and CO 2 emission based on the charging infrastructure, road network, and user’s travel demand. •• Journey planning: the personalised and vehiclised journey is determined. The novelty of the journey planner is that the characteristics of electric vehicles and charging networks may be considered, and the charging is included. •• Charging assistance: helps to start and finish the charging process and provides information on the status of charging during the session. Charging time is reduced by providing information on the charging process to help the user finish charging. The charging may be finished automatically as soon as the user’s demands are met. A detailed description of the integrated information service can be found in [2]. Charging station locating Intra-city and inter-city charging demand were distinguished. In the case of intra-city charging, the vehicles may be charged during parking. Therefore, the long charging time may not cause inconvenience if the charging station is close to the destination. In the case of intercity charging, the charging session interrupts the journey. Therefore, fast charging is utmost of importance. Because of the differences, separated charging station locating methods were elaborated for urban areas and national roads. Urban areas A two-level weighted sum model had been elaborated to determine the areas where the willingness to use an electric car is high and public locations where users would frequently charge (figure 1). On the macroscopic level, the distribution of charging stations among territory segments in the investigated territory considering the potential of electric car use was performed. The potential was determined based on the number of registered electric cars, average income, tourism importance, and other influencer effects such as subsidies on parking and charging. On the microscopic level, the territory segments are divided into hexagons. The hexagon size was based on the willingness to walk between the charging station and the destination. The installation potential was calculated for each hexagon. The installation potential was determined based on the general parking demand at location types, willingness to walk, residential area type, the number of points of interest and the population in the hexagon. Various scenarios were determined using different weights. A detailed method can be found in [3]. National roads A weighted multicriteria location optimisation method with ranking and selection was elaborated. The selection of proposed fast-charging station sites from the candidate sites was performed on several layers. It was desired and assumed that the inter-city traffic is concentrated on the main national roads. Hence, road categories were put in focus and assigned to layers. E.g., motorways and main roads may be considered on different layers. Unlike in several other studies [4], O-D flows were not considered to provide an applicable method if the origin-destination (O-D) data are unavailable. Thus, the focus was put on the spatial coverage and not on origin-destination flows. Figure 1: Territory segments on macroscopic and hexagons on microscopic level - the subjects of evaluation Own work INTERNATIONAL European Friedrich-List-Prize Internationales Verkehrswesen (72) 3 | 2020 62 The candidate site with the highest installation potential was selected. The installation potential is calculated considering the traffic volume, the total population of nearby settlements, service level at the site, and the location of existing fast-charging stations. Already selected candidate sites are considered as existing fastcharging stations. The method was verified using O-D data for several electric car range scenarios. It was found that a high share of served traffic volume can be achieved even at a low electric car range by emphasising the importance of traffic volume. A detailed method can be found in [5]. Charging planning In the literature, centralised [6] and decentralised [7] charging planning methods can be found. In the case of centralised planning, the calculation is performed at a centre based on the characteristics of supply and demand. The aim is to achieve a global optimum. In the case of decentralised control, the aim is to achieve local (EV user) optimum. Namely, the optimisation of charging sessions is performed separated. Thus, only one electric vehicle’s charging demand is considered at once. I have elaborated a decentralised charging planning method to minimise the user’s charging cost. The charging plan includes when and where to charge. The charging plan was determined based on the supply and demand (figure 2). A two-way energy stream between the vehicle and the electrical network was considered. The efficiency of the method was tested for several scenarios considering various charging strategies and dynamic tariff systems. It was found that the elaborated charging planning method is an efficient tool to minimise the charging cost and decrease the fluctuation of the electricity demand. Depending on the dynamic rates, the method may significantly reduce charging costs. A detailed method can be found in [3]. Conclusion The elaborated concept of the integrated information system is the basis of the implementation, and it provides a framework for electromobility services that support electric car use. The novel functions, the necessary input data groups, the components, and the relationship among them had been revealed, which are beneficial for the operators of future electromobility services. The elaborated charging station locating methods support the deployment along national roads and in urban areas in the early phase, while the deployment cannot be done on a market basis, and the locations have a substantial influence on electric vehicle use. The location selection was made in consideration of the estimated charging demand. The e-Mobi Elektromobilitás Ltd applied the methods. The precondition of the elaborated charging planning method is the different electricity rates at charging stations. Thus, the user may benefit from the use of charging optimisation. Reducing the grid load fluctuation by decentralised charging optimisation a dynamic electricity rate based on the free capacity is necessary. However, such a dynamic electricity rate is not resolved yet; there is a huge potential in the application of charging optimisation. ■ REFERENCES [1] Büscher, M.; Coulton, P.; Efstratiou, C.; Gellersen, H.; Hemment, D.; Mehmood, R.; Sangiorgi, D. (2009): Intelligent mobility systems: some socio-technical challenges and opportunities. In: International Conference on Communications Infrastructure. Systems and Applications in Europe, pp. 140-152. www.doi.org/ 10.1007/ 978-3-642-11284-3_15 [2] Csonka, B.; Csiszár, C. (2019): Integrated Information Service for Plug-In Electric Vehicle Users Including Smart Grid Functions. In: Transport 34(1), pp. 135-145. https: / / doi. org/ 10.3846/ transport.2019.8548 [3] Csiszár, C.; Csonka, B.; Földes, D.; Wirth, E.; Lovas, T. (2019): Urban public charging station locating method for electric vehicles based on land use approach. In: Journal of Transport Geography 74, pp. 173-180. https: / / doi.org/ 10.1016/ j.jtrangeo.2018.11.016 [4] Capar, I.; Kuby, M.; Leon, V. J.; Tsai, Y-j. (2013): An arc cover-path-cover formulation and strategic analysis of alternative-fuel station locations. In: European Journal of Operational Research 227, pp. 142-151. https: / / doi.org/ 10.1016/ j.ejor.2012.11.033 [5] Csiszár, C.; Csonka, B.; Földes, D.; Wirth, E.; Lovas, T. (2020): Location optimisation method for fast-charging stations along national roads. In: Journal of Transport Geography 88, 102833. https: / / doi.org/ 10.1016/ j.jtrangeo.2020.102833 [6] Klaimi, J.,; Rahim-Amoud, R.; Merghem-Boulahia, L.; Jrad. A. (2018): A novel loss-based energy management approach for smart grids using multi-agent systems and intelligent storage systems. In: Sustainable Cities and Society 39, pp. 344-357. https: / / doi. org/ 10.1016/ j.scs.2018.02.038 [7] Mal, S.; Chattopadhyay, A.; Yang, A.; Gadh. R. (2012): Electric vehicle smart charging and vehicle-to-grid operation. In: International Journal of Parallel, Emergent and Distributed Systems 28(3), pp. 249-265. https: / / doi.org/ 10.1080/ 17445760.2012.663757 Bálint Csonka, PhD Research associate, Department of Transport Technology and Economics, Faculty of Transportation Engineering and Vehicle Engineering, Budapest University of Technology and Economics, Budapest (HU) csonka.balint@kjk.bme.hu Figure 2: Charging planning method input and output Own work Changing the Game in Transportation With „International Transportation - Collection 2021” we present the complete range of this year’s English contributions - plus additional articles in the sections Strategies • Best Practice • Products & Solutions • Science & Research “International Transportation - Collection 2021” is an e-journal - ready for desktop computers as well as mobile devices. Publishing date will be 4 October 2021. Further information: www.international-transportation.com European Friedrich-List-Prize INTERNATIONAL Internationales Verkehrswesen (72) 3 | 2020 63 Estimation of turning rates in roundabouts, applying state-space estimation methods Traffic estimation, Roundabout, Turning rate, Traffic count, Kalman Filter, Constrained Kalman Filter The aim of this paper is the examination and comparison of different estimation methods used for determining turning rates (OD-matrix) in roundabouts. A traditional iteration-based approach as well as state-space estimators are validated on real-world traffic data. For the estimation procedures, the traffic flows (measured at each leg of the intersection) are the input. In this way, the manual origin-destination traffic count at an intersection can be substituted by automated traffic detection at the cross-sections together with an adequately implemented estimation process. Mánuel Gressai R oad traffic infrastructure planning or development is initiated based on reliable traffic modeling. The input of the modeling is the vehicular flows on road links and turning rates at intersections. Traffic volumes at cross-sections can be straightforwardly measured manually or with help of a wide variety of traffic sensors. At the same time, turning flows or turning rates can be collected by human resources solely, which is quite costly. Therefore, if turning flows are collected, typically more than one person is needed to perceive all movements. The more, observing turnings in roundabouts is extremely problematic due to the special geometry and size of this type of junction. Figure 1 demonstrates the possible turning movements at a roundabout for vehicles arriving at Entrance 1. V 1j is the turning traffic flow from Entrance 1 to exit j, whereas V 1,in and V 1,out are the total traffic volumes entering and exiting at the corresponding junction leg. Using the volumes in figure 1, turning rates can be defined as follows: (1) where n D is the number of exits. Counting traffic on the legs of a roundabout and adequately estimating turning rates based on the collected data has the potential to substitute labor-intensive turning flow counts. This could reduce the cost of determining turning rates at an intersection significantly. In this paper, cross-sectional counts are used as a basis to estimate turning rates at a roundabout. This proposes a possible solution to overcome the obstacles posed by turning movement observation. This paper is divided into 5 sections. A description of the examined estimation methods follows the introduction. Next, the testing of different estimation methods is presented. The methods are then compared using different error metrics. Finally, a case study is established to determine whether the tuning of the best performing estimator depends on the traffic conditions, or it can be considered robust under different circumstances. Recommendations and future directions for the research are stated, and conclusions are drawn the last section. Estimation methods This section covers different methods used for turning rate estimation. Biproportional procedure is discussed as a traditional iterative algorithm, then the Kalman Filter and its extension with constraint handling are introduced as well as the Moving Horizon Estimation. The biproportional procedure (BP) is an iterative algorithm [1], where the variation of two coefficients causes the variation of turning flows in each iteration. The BP procedure aims to estimate the elements of the current OD-matrix based on the current flows on each leg and the prior OD-matrix. State-space based estimators such as the Kalman Filter include a model of the system and noises [2]. Some procedures are apt to manage constraints concerning- the- estimated values (e.g., for each turning rate to be- non-negative). Moreover, these methods estimate Figure 1: Possible turning movements at a four-legged roundabout from one direction Own work INTERNATIONAL European Friedrich-List-Prize Internationales Verkehrswesen (72) 3 | 2020 64 the- mean and standard deviation for all states in each interval. The Kalman Filter (KF) uses the estimated states (turning rates) in the previous step and the measurements in the current step to estimate the current state. The constrained Kalman Filter (cKF) contains an optimization after each step for the estimated values to satisfy previously defined constraints. The constraint handling section contains a weighing matrix which can be set to an identity matrix (I) or the error covariance matrix (P) in each step [3, 4]. Based on these options, cKF-I and cKF-P methods were defined. The MHE not only contains an optimization problem to minimize the state and measurement noises, but it can also consider the estimated values of more than one previous steps [5]. The horizon length ranged from 1 to 4 during the tests, and these cases were named MHE1, MHE2, MHE3, and MHE4, respectively. Applying the estimation methods Real turning movement volumes were counted at two different roundabouts for the research. In this way, after the calculation of turning rates, real data was available for validation and comparison of the proposed estimation algorithms. The 30-minute counts were divided into 1, 2, and 5-minute intervals, so that the estimation methods could be tested for different data input frequencies. The estimation algorithms introduced in the paper are based on the counted number of vehicles expressed in PCE (Passenger Car Equivalent [6]). The input of the estimators is the traffic flows, but the real turning rates are also known. This makes it possible to compare the estimated turning rates with the real data using error metrics. Four different error metrics [7] have been applied during the evaluation of estimation procedures: •• mean absolute error (MAE), •• root mean square error (RMSE), •• mean absolute percentage error (MAPE), and •• symmetric mean absolute percentage error (SMAPE). These metrics were also used during the tuning of the state-space estimators. The weighing between the state noise and measurement noise covariance matrices determines the attributes of the estimation. The tuning was carried out by searching for the minimum of errors. Based on the results, it can be stated that the longer the interval, the more accurate the estimation. The 5-minute interval led to the smallest errors in the case of every examined method. This clearly means that on longer time intervals, the algorithms can better smooth their estimations. Based on this, 1or 2-minute sampling intervals are not suggested to be applied in this practical problem. Table 1 lists the average error measures for all examined estimation methods in the case of 5-minute interval sizes. A ranking in the MAE values is also assigned to the procedures. In this case, the cKF and the MHE outperform the BP method and the unconstrained Kalman Filter. The comparison revealed that, although the performance of the MHE is generally slightly higher than that of the constrained Kalman Filter, tuning the MHE is often problematic. Therefore, the constrained Kalman Filtering is suggested as the best estimation procedure, taking the tuning circumstances and the performance into account. Simulation-based case study A case study was also carried out, the basis of which is another real-world traffic count. The examination was then extended to a simulation environment created in PTV Vissim modeling software. In this case, the 4-hour traffic measurements consisted of 15-minute intervals. Every method was tuned to the available traffic data sets. In case of the 15-minute intervals, the constrained Kalman Filter (cKF-P) turned out to estimate with the smallest errors. After the GEH validation [8, 9], a simulation model was established, and different traffic circumstances were created in the modeled environment. The cKF-P proved to be the most effective estimator; therefore, its performance was tested in the different scenarios, without changing the tuning. Three traffic parameters were altered to create the scenarios: •• traffic volumes, •• the proportions of the main road and side road volumes (traffic ratio), •• the position of the main road (opposite or adjacent legs). The created scenarios are as follows: •• S0: control scenario, the average of morning and afternoon traffic counts; •• S1: S0 scenario’s OD matrix, multiplied by 1.3; •• S2: 1: 2 traffic ratio, main road is opposite legs (2, 4); •• S3: 1: 6 traffic ratio, main road is opposite legs (2, 4); •• S4: 1: 2 traffic ratio, main road is adjacent legs (2, 3); •• S5: 1: 6 traffic ratio, main road is adjacent legs (2, 3). Comparing the errors of the scenarios with the control case, the following observations can be made: •• the alteration of traffic parameters did not affect the performance of the constrained Kalman Filter substantially; •• the increase of traffic volumes did not cause anomalies in the estimation; •• concerning the main and side road volume ratios, the cKF-P was more accurate when the side road volume proportions were lower; •• adjacent main road legs resulted in larger errors; •• in case of opposite main road legs, estimation performance was affected more by the traffic ratios. Intervals: 5 min Method MAE RMSE MAPE SMAPE Rank (MAE) BP 0.0670 0.1050 27.89% 11.53% 5 KF 0.0742 0.1118 32.51% 11.95% 7 cKF-I 0.0692 0.1048 27.84% 11.66% 6 cKF-P 0.0608 0.0945 24.72% 10.29% 3 MHE1 0.0599 0.0928 26.84% 10.92% 1 MHE2 0.0602 0.0942 25.81% 10.66% 2 MHE3 0.0633 0.1027 31.36% 11.74% 4 MHE4 0.0745 0.1054 32.39% 13.68% 8 Table 1: MAE rank for the examined methods (5-min intervals) European Friedrich-List-Prize INTERNATIONAL Internationales Verkehrswesen (72) 3 | 2020 65 Conclusions The main contribution of the paper is the validated comparison of different methods on real-world data sensed by drone and then counted manually. Analyzing the results, the following conclusions can be drawn: •• in general, longer intervals result in more accurate estimations; •• managing constraints improves the accuracy of the state-space estimators significantly; •• the adequately tuned constrained Kalman Filter and MHE outperform the unconstrained Kalman Filter and the traditional iterative procedure. The comparison was followed by a simulation-based case study. Tuning the state-space estimators determined that, for the examined roundabout and circumstances, the constrained Kalman Filter (cKF-P) was the most adequate method. After building the model and validating it, different scenarios were created as an input for the cKF-P. The case study led to these conclusions: •• in general, different traffic situations did not affect the performance of the estimation significantly; •• the tuning of the cKF-P can be considered robust; •• increasing traffic volumes did not cause anomalies in the estimation performance; •• the estimation is more accurate, when the side road traffic volume proportions are lower; •• the estimation is more accurate, when the main road runs on opposite legs. The continuation of this research is twofold. On the one hand, managing different road vehicle categories is an important research aim, as road planning companies generally require traffic counts with four different vehicle classes. On the other hand, a more sophisticated tuning can be developed for the state-space estimators. This can involve determining different parameters for the main road and the side road, or even traffic responsive tuning. As the result of the research, it can be stated, that state-space estimation methods combined with automatic cross-section traffic counts provide a real alternative to paper-based, manual intersectional traffic counts in case of roundabouts. ■ REFERENCES [1] Ben-Akiva, M.; Macke, P. P.; . Hsu, P. (1985): Alternative methods to estimate route-level trip tables and expand on-board surveys. In: Transportation Research Record. [2] Kalman, R. E. (1960): A new approach to linear filtering and prediction problems. In: Journal of Basic Engineering (ASME). [3] Gupta, N.; Hauser, R. (2007): Kalman filtering with equality and inequality state constraints. Oxford University Computing Laboratory. [4] Simon, D. (2010): . Kalman filtering with state constraints: a survey of linear and nonlinear algorithms. In: IET Control Theory & Applications. [5] Haugen, F. A. (2018): A brief introduction to optimization methods. [6] Lay, M. (2009): Handbook of Road Technology. Spon Press. Abingdon, UK. [7] Chen, C.; Twycross, J.; Garibaldi. J. M. (2017): A new accuracy measure based on bounded relative error for time series forecasting. In: PloS one. [8] Feldman, O. (2012): The GEH measure and quality of the highway assignment models. In: Association for European Transport and Contributors, pp. 1-18. [9] TfL (Transport for London): Traffic Modelling Guidelines: In: TfL Traffic Manager and Network Performance Best Practice Version 3.0. 2010. Mánuel Gressai Research associate, Department of Control for Transportation and Vehicle Systems, Faculty of Transportation Engineering and Vehicle Engineering, Budapest University of Technology and Economics, Budapest (HU) gressai.manuel@edu.bme.hu Road transport price Correlation of rates for road transport services in domestic transport in-Poland Economic transport, Data analyst, Road transport, Freight forwarding, Transport management The work focuses on the study of the correlation between the GDP of geographical regions and transport rates. Analyzed what influences the price of transport the thesis was put forward that transports whose loads are located in economically developed regions cost more than those whose loads are in less developed regions. The research used data on transports carried out in Poland. The results were compared with the GDP ratio given for each NUTS3 sub-region. A correlation was found and the thesis made at the beginning was confirmed. Artur Budzyński T he main aim of this article is to investigate how the GDP of the NUTS geographical sub-regions affects the price of a transport service. Thesis put forward is that there is a correlation between the transport rates and the areas of loading these shipments. It is assumed that transport from more economically developed