**About the Editors**

#### **Giovanni Randazzo**

Prof. Dr. Giovanni Randazzo is an associate professor of coastal geomorphology and environmental geology at the University of Messina (Department of Mathematics, Computer Sciences, Physics and Earth Sciences, Via F. Stagno d'Alcontres, 31–98166 Messina, Italy). He holds a PhD in marine environments and resources obtained from the University of Messina (Italy). Since 1987, his research interest has focused on the study of coastal areas, their management and protection. In these last 30 years, he collaborated with the Smithsonian Institution of Washington D.C. in the study of the Nile delta; with the Thai Geological Service in the study of the east coast of the local peninsula; and with ENEA (the Italian National Agency for New Technologies, Energy and Sustainable Economic Development), he participated in the X Italian expedition in Antarctica. He collaborated in the environmental assessment impact of various public works (especially in coastal areas), as well as in the drafting of the Territorial Landscape Plan of the province of Messina (Sicily, Italy). In recent years, he actively participated in the debate on the emergence of waste, writing scientific articles, intervening in local press, and participating in various debates, where he presented a scheme of the management of emergency alternatives to those not acting for the Sicilian region. In 2013, he founded Geologis s.r.l., a branch of the University of Messina, active in the field of territory surveys using aerial and marine drones equipped with RGB cameras, LiDAR sensors, and thermal imaging cameras. On behalf of the European Union, he coordinated several projects at a national level and/or as a local unit related to coastal management and territorial security. Since December 2017, he has been the lead partner of the Pocket Beach Management & Remote Surveillance System - Program Interreg VA Italia Malta 2014–2020. He is the author of more than 120 scientific publications.

#### **Anselme Muzirafuti**

Dr. Anselme Muzirafuti, born in Rwanda, is an assistant professor at the University of Messina (Department of Mathematics, Computer Sciences, Physics and Earth Sciences, Via F. Stagno d'Alcontres, 31–98166 Messina, Italy). He holds a master's degree in applied geophysics and geology engineering obtained in 2015 at the University of Moulay Ismail, Meknes (Morocco), and a PhD in hydrogeophysics obtained in 2021 at the same university. Since 2009, he has received multi-excellence scholarships from different governments for his higher education studies, including the Government of Rwanda, the Government of the Kingdom of Morocco, and the European Union. In 2016 and 2018, he participated in major conferences on climate change and sustainability, namely, the 22nd conference of parties held in Marrakech (Morocco) and the 24th International Sustainable Development Research Society Conference (Action for a Sustainable World: from Theory to Practice), held in Messina (Italy). His research interest has focused on the use of structural geology, geomatics and geophysics for the sustainable management of territories. He worked on different projects in Morocco and Italy related to geomorphological mapping and surveys using images acquired with satellites and drones. He recently worked as an analysist of satellite images in the BESS project Pocket Beach Management & Remote Surveillance System - Program Interreg VA Italia Malta 2014–2020. The results of his works have been presented in international conferences and published in international journals.

#### **Dimitrios S. Paraforos**

Prof. Dr. Dimitrios S. Paraforos is serving as a deputy professor of agricultural engineering in the Special Crops at the Hochschule Geisenheim University (Department of Technology, Von-Lade-Str. 1, D-65366 Geisenheim, Germany). He holds an MSc from the University of Thessaly (Greece) and, in 2016, obtained a PhD from the University of Hohenheim, both in agricultural engineering. His research focuses on precision farming, digital technologies in agriculture, and, more generally, on control systems, robotics, and automation applied to agriculture; in addition, he has obtained industry experience working as an automation engineer in the food industry. He is the author of more than 53 papers indexed by Scopus, with an important contribution in the field of sensors and ISOBUS technologies for the enhancement of agricultural practices. Currently, he is coordinating the ERA-NET ICT-Agri II European project iFAROS on developing methods for increasing the efficiency and precision of site-specific fertilizer applications.

#### **Stefania Lanza**

Dr. Stefania Lanza is an administrator of Geologis s.r.l, an academic branch of the University of Messina (Via F. Stagno d'Alcontres, 31–98166 Messina, Italy). In 2007, she obtained a PhD in geology from the University of Messina (Italy) with a thesis on "The risk assessment of coastal areas: from planning to monitoring". She worked on different projects related to sedimentology and geomorphology mapping. In 2008, she took the final master's course exam with a thesis entitled: "Coastal monitoring of the coast of Badalona (Spain) contribution to the 2007–2008 survey campaign" supervised by Prof. Jordi Serra of the Autonomous University of Barcelona. In 2013, she co-founded Geologis s.r.l., a branch of the University of Messina, active in the field of territory surveys using aerial and marine drones equipped with RGB cameras, LiDAR sensors, and thermal imaging cameras. She recently worked as the coordinator of the geomorphological and sedimentological activities of the BESS project "Pocket Beach Management & Remote Surveillance System - Program Interreg VA Italia Malta 2014–2020". She is currently working as a coordinator of field activities for the "BIOBLU project - Robotic Bioremediation for Coastal Debris in Blue Flag Beach and in a Maritime Protected Area - Interreg V-A Italy - Malta 2014-2020 program".

## **Preface to "Future Transportation"**

This Special Issue on Future Transportation falls in the scope of the current effort to mitigate and adapt to a changing climate. It is launched with the aim of collecting and promoting recent scientific studies proposing and evaluating advances in technologies leading to the sustainable future transportation of people and goods. It mainly addresses the policy makers, entrepreneurs and academicians engaged in the fight against climate change by tackling the main contributors of greenhouse gas (GHG) emissions. A special thanks is addressed to the authors who submitted their manuscripts to contribute to this initiative. We acknowledge the funders who financially assisted the authors to conduct their research, and we also thank the reviewers and editors who contributed to the evaluation of the scientific quality of the submitted manuscripts.

#### **Giovanni Randazzo, Anselme Muzirafuti, Dimitrios S. Paraforos, and Stefania Lanza** *Editors*

## *Article* **Automobile Technological Transition Scenarios Based on Environmental Drivers**

**Julieth Stefany García 1, José D. Morcillo 2,\*, Johan Manuel Redondo <sup>3</sup> and Mauricio Becerra-Fernandez <sup>4</sup>**


**Abstract:** Different industrial sectors are assuming measures to mitigate their greenhouse gas emissions, facing the imminent materialization of climate change effects. In the transport sector, one of the measures involves the change in energy source of vehicles, leading to a transition from vehicles powered by fossil fuels (conventional) to electric. Nevertheless, electric vehicles have different drivers that promote their purchases. This work only considers the informed buyers' interest in making their decisions using environmental criteria. However, these technologies have a series of impacts, including the generation of hazardous waste such as used batteries, which leads consumers to question the environmental impacts generated by conventional and electric vehicles; consequently, it is uncertain which prospective scenarios will dominate in various nations and what will promote them. Therefore, the proposed model is studied as a dynamical system, with bifurcations of codimension 2, which means that it is possible to represent all possible prospective scenarios of this configuration through a bifurcation diagram. In this way, the analysis allows us to find that four families of technological transitions (trajectories that qualitatively can be identified as being of the same behavior class) emerge from the relationships established in the system, showing similarities to the different transition situations recognized on the planet. This model is an attractive tool to classify automobiles' technological transitions, despite having no other criteria. In fact, although decarbonization is an urgent quest in the transport sector, there are still too many challenges to guarantee environmentally friendly technologies.

**Keywords:** technological transitions; automobiles; system dynamics; dynamical systems; bifurcations

#### **1. Introduction**

In 2015, during the Paris Conference, governments agreed to limit global warming to levels below preindustrial levels [1–3]. To achieve this goal it will be necessary to keep emissions between 420 and 1200 GT of CO2 by 2100 [4,5], which led to 190 countries making commitments that define voluntary climate actions until 2030, such as Nationally Determined Contributions (NDCs) and action plans to achieve targets of the Sustainable Development Goals [6].

The energy sector is one of the largest contributors to emissions worldwide. For 2021, this sector emitted approximately 33 GT of CO2, which shows a growth of 4.8% compared to 2020 [7].

For its part, the transportation sector is responsible for 24% of direct emissions of CO2 from the use of fuels [8,9]. Likewise, road automobiles account for almost three-quarters of the CO2 emissions from transport and are also the fragment with the greatest opportunity for decarbonization [10]. For this reason, to fulfill the agreed commitments, the need

**Citation:** García, J.S.; Morcillo, J.D.; Redondo, J.M.; Becerra-Fernandez, M. Automobile Technological Transition Scenarios Based on Environmental Drivers. *Appl. Sci.* **2022**, *12*, 4593. https://doi.org/10.3390/ app12094593

Academic Editors: Giovanni Randazzo, Dimitrios S. Paraforos, Stefania Lanza, Anselme Muzirafuti and Daniel Villanueva Torres

Received: 25 February 2022 Accepted: 29 April 2022 Published: 1 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

to achieve an energy transition towards zero and low emission technologies have been identified to counteract the effects of climate change [11].

Rapid urbanization, the dynamism in the populated centers, and the dependence on transportation to improve the quality of life have increased the demand for automobiles, of which a large proportion run on fossil fuels that are strongly related to the emissions of greenhouse gases (GHG), leading to increased air pollution, increased respiratory illnesses, and impaired quality of life [12,13].

Among the alternatives established by nations to mitigate the emissions of CO2 from the transport sector are automobiles with alternative technologies that use zero and low emission fuels, such as hybrid vehicles (HEV), plug-in hybrid vehicles (PHEVs), and battery electric vehicles (BEVs). These alternatives are valuable for consumers who make their decisions based on the environmental impact generated by these technologies [14–18].

As a consequence, the growth in vehicle sales of these technologies has intensified. According to the International Energy Agency, 2.2 million electric cars were sold in 2019, representing 2.5% of global car sales. During 2020, the automobile market was contracted due to the effects of the pandemic, but sales of electric cars increased to 3 million, representing 4.1% of total automobile sales. During the year 2021, sales of electric vehicles doubled and reached a total of 6.6 million, which represents about 9% of the global car market [19]. According to the data, a total of 16 million electric vehicles travel on roads worldwide [19], with 90% of global sales of these technologies concentrated in China, Europe, and the USA [10].

A Tank to Wheel analysis conducted in [20] found that battery-powered electric cars reduce CO2 emissions by 32% compared to their conventional counterparts. Likewise, the transition towards these technologies reduces *GHG* emissions and contributes to the fulfillment of the commitments agreed by countries [9,13,21–24]. However, to achieve decarbonization in the transport sector, it is necessary to encourage generation with nonconventional sources of renewable energy, given that those nations in which fuels and coal have a high share may not perceive the benefits of the electrification of the sector [25–29]. For this reason, countries with high emissions of CO2 must diversify their energy generation matrix, take advantage of their electric vehicle capacity, and reduce CO2 emissions [30,31].

Electric vehicles have different drivers that promote their purchase; one of the main drivers is the interest of informed buyers who make their decisions using environmental criteria, such as reducing noise pollution and greenhouse gas emissions and improving energy efficiency, as well as improving the ecological image that the individuals and/or organizations have [32].

However, the zero and low emission vehicles sector still faces significant challenges in the transition towards electric technologies, such as increasing the charging infrastructure, increasing the energy efficiency, increasing the penetration of renewable energies in the energy matrix, and reducing the costs of batteries and charging stations, as well as improving the profitability of the cars [33]. Furthermore, it is necessary to establish incentives or policies that governments can promote to ensure the rapid and efficient adoption of environmentally friendly cars [34].

The transition to electric automobiles also has a series of negative significant environmental impacts related to mineral exploitation and final disposal of the electric and electronic parts used [35]. Examples of this are lithium, cadmium, and nickel, which show a balance between supply and demand in the short and medium-term but a scarcity in the long term due to the expected increase in demand for hybrid and electric vehicles (BEV, PHEV, and HEV) [36] and current recycling rates below 1% [36].

Regarding waste, the components of electric batteries are hazardous waste, causing significant health and environmental problems without proper management [37]. In this sense, it is necessary to establish recycling systems and carry out socialization actions to recover the lithium, nickel, and cadmium and be able to reintegrate them into the value chain, both for vehicle batteries and for other uses [37]. Furthermore, the battery's performance must be improved to reduce the intensity of exploitation of these minerals in the sector, while increasing other efforts to develop alternative options [36].

This work discusses the transition scenarios of automobile technologies towards zero and low emission alternatives, considering that current technologies generate other impacts related to the disposal of hazardous solid waste. The assumption is that the technological change decision has as its only determinants emissions and hazardous waste reduction; so, the buyers base their decision on environmental responsibility. As a result, the buyers fall into the paradox of choosing between the impact alternatives due to the benefits derived from automobile use.

Different authors have investigated issues in the transition towards zero and low emission technologies [32], and their approaches and methodologies are varied. Some methods to analyze transition are life cycle analysis [15], modeling with different techniques [31,35,38–43], system dynamics [35,44,45], cost analysis [23,29,46], literature reviews [47,48], linear regressions [49], LEAP software models [24], surveys and interviews [50–52], ANSWER MARKAL software models [53], and Well to Wheel analyses [26]. These authors have focused their studies on evaluating and analyzing how technological transitions occur in national transport sectors. However, they only address the current conditions and the incentives implemented in each country.

Conversely, the approach presented in this work aims to be systemic, by identifying the trend behavior derived from consumers whose only criterion for automobile choice is environmental responsibility. In this sense, we present a first approximation that allows identifying different categories of national transition, called prospective transition scenarios, which emerge from an analysis based on bifurcation theory. This approach is novel in the literature, only a first approximation for analyzing capacity scenarios in electric markets was proposed in [54,55].

In this way, this work recognizes and describes the technological transition scenarios involving conventional and electric automobiles based on the environmental drivers of emissions and disposal of hazardous waste.

In summary, the emissions generated by the transport sector are decisive when consumers take action. Among the alternatives to decarbonize the transport sector is the transition from conventional to less polluting technologies. In particular, electric vehicles are an alternative that has gained strength, which has led them to increase their market share. This technology has become more popular among shoppers who make environmentally responsible purchasing decisions. However, EVs generate hazardous waste from their batteries. This causes the consumers to question whether to prioritize the environmental impacts generated by fuel vehicles or EVs. This will impact the growth or reduction in the fleet of both of these technologies. As a consequence, it is uncertain which prospective scenarios will have significant participation and what will promote them. In this research work, only the informed decision of consumers with an environmental commitment was considered a driver.

The document is organized as follows. This introductory section has shown the problem and purpose of this work, along with the state of the art. Section 2 presents the mathematical modeling carried out from system dynamics to obtain the dynamical system from which the results come. This section explains a dynamical system, a bifurcation analysis, how the system equilibrium points were calculated, and how their stabilities were established. Section 3 presents the phase portraits, time series, and the bifurcation diagram obtained for the dynamical system under study. In Section 4, the implications of the results are presented and future work is proposed. Finally, Section 5 presents the conclusions of this research.

#### **2. Materials and Methods**

The starting point of the analysis presented in this document is the mathematical model of Equation (4). This system of first-order ordinary differential equations has been obtained from system dynamics methodology (see for example [56]).

The rules of causality are simple: (1) conventional automobiles generate emissions of CO2 [22,23,57], (2) electric automobiles produce hazardous waste [36,58], (3) emissions and hazardous waste generate nonconformity in vehicle consumers taking the option to change technology to reduce its environmental impact [32], (4) the enthusiasm to acquire one or another technology leads to the entry of automobiles into the system. The hazardous waste of conventional vehicles is out of the scope of this research work, since it is not as important a selection criterion for buyers as the hazardous waste of electric automobiles.

From these causality rules, we built the Forrester diagram shown in Figure 1. Conventional and electric automobiles are the level variables of the system. The equations presented below are an interpretation of the Forrester diagram. The rules used were [56]: (1) the change over time in level variables, represented in boxes, is the difference between the inflow and outflow variables, symbolized as valves; (2) the flow variables are the products between the variables connected with arrows, and (3) the auxiliary variables (represented as circles), which are functions of the variables connected through arrows, are defined according to the relationship between them.

**Figure 1.** Stocks and Flows diagram of the automobile technology transition from environmental drivers.

The conventional automobiles *x* change over time (represented by a point on the variable), given in [*automobiles*], is the difference between automobiles entering the market *CAI* and those leaving *CAO*, both given in [*automobiles*/*year*]. The electric automobiles *y* change over time, given in [*automobiles*], is the difference between the automobiles that enter the market *EAI* and those leaving *EAO*, both given in [*automobiles*/*year*]. So, the level equations are defined as follows:

$$\begin{aligned} \dot{\mathbf{x}} &= \mathbf{C}AI - \mathbf{C}AO \\\\ \dot{\mathbf{y}} &= \mathbf{E}AI - \mathbf{E}AO \end{aligned} \tag{1}$$

The change ratio equations presented in Equation (1) show that the increase in the quantities of each automobile category depends on the disagreement that users have with the other category, and the decrease depends on an average depreciation, as follows:

$$\begin{aligned} \text{CAI} &= a \cdot \mathbf{x} \cdot (1 - \text{CEA}) & \quad \text{CAO} &= b \cdot \mathbf{x} \\\\ \text{EAI} &= c \cdot y \cdot (1 - \text{CCA}) & \quad \text{EAO} &= d \cdot y \end{aligned} \tag{2}$$

where *a*, *b*, *c*, and *d* are exchange rates in the range [0, 1], given in [*year*−1], and the conformities with each category depend on the closeness of the emissions *E*, given in [*TonCO*2/*year*], or hazardous waste disposal *W*, given in [*Ton*/*year*], with an allowable limit value (emissions limit *N*, given in [*TonCO*2/*year*], and hazardous waste limit *W*, given in [*Ton*/*year*]) that could be agreed through national or international environmental commitments. The equations for emissions and hazardous waste are defined as follows:

$$\begin{array}{ll}\mathsf{CEA}=\mathsf{W}/L & \mathsf{CCA}=\mathrm{E}/N\\\\ \mathrm{E}=\mathsf{e}\cdot\mathrm{x} & \mathrm{W}=\rho\cdot y\end{array} \tag{3}$$

where *-* ≥ 0 and *ρ* ≥ 0 are exchange rates given in [*TonCO*2/(*automobile* · *year*)] and [*Ton*/(*automobile* · *year*)], respectively.

Finally, by replacing the equations, and considering *α* = *a* − *b*, *β* = *c* − *d*, *δ* = *aρ*/*L*, and *λ* = *c-*/*N*, we obtain the following mathematical model:

$$\begin{aligned} \dot{\mathfrak{x}} &= \mathfrak{x}(\mathfrak{a} - \delta \mathfrak{y}) \\\\ \dot{\mathfrak{y}} &= \mathfrak{y}(\beta - \lambda \mathfrak{x}) \end{aligned} \tag{4}$$

Note that if the system of Equation (4) is studied under the variation of *α* and *β*, different scenarios of exchange rates will be considered.

The model validation was performed using the tests reported in [59]. In particular, we conducted the empirical structure confirmation test and empirical parameters confirmation test. The theoretical structure confirmation test was performed through the literature review mentioned above to explain the causal rules construction. The theoretical parameters confirmation was carried out during the construction of the model by defining the application ranges, as shown above. The dimensional consistency test appears together with the equations explanation, verifying the units proposed. We also carried out extreme value tests allowing us to recognize the limits and scope of the model. When possible, the bifurcation analysis is better than the system sensitivity tests because the bifurcation analysis can identify all the probable system behaviors, while the sensitivity analysis only allows visualizing the transitory state under the parameter variation in a limited range, using the Monte Carlo method. For this reason, our bifurcation analysis shown below supports the validity of our proposed model. Finally, we carried out phase relationship tests to compare the behavior between the state variables and verify the behavior consistency through phase portraits, as shown in Section 3. We did not conduct behavior validation with real time series, because it is still too early to have enough national data that could be comparable with the behaviors shown by our prospective mathematical model.

The mathematical model of Equation (4) is a continuous dynamical system. For this reason, we analyzed the equilibrium points and their stability and determined that the system has bifurcations in codimension two.

A dynamical system (see [60,61]) is a triple (*X*, *T*, *ϕ*), where *X* is the state space, *T* is an ordered set representing time, and

$$\begin{array}{c} \varphi: T \times X \longrightarrow X \\ (t, \mathfrak{x}) \longmapsto \varphi\_{\mathfrak{f}}(\mathfrak{x}) \end{array} \tag{5}$$

is a parameterized family of evolution operators that satisfies the following:

• The system state does not change spontaneously

$$
\varphi\_0(\mathbf{x}) = \mathbf{x} \quad \text{for all } \mathbf{x} \in \mathbf{X} \tag{6}
$$

• The evolution law of the system does not change in time

$$
\varphi\_{l+s}(\mathbf{x}) = (\varphi\_l \circ \varphi\_s)(\mathbf{x}) = \varphi\_l(\varphi\_s(\mathbf{x})) \quad \text{for all } \mathbf{x} \in \mathcal{X} \text{, and all } \mathbf{t}, \mathbf{s} > \mathbf{0}. \tag{7}
$$

In the model of Equation (4), the state space is defined by the state variables of conventional automobiles *<sup>x</sup>* and electric automobiles *<sup>y</sup>*; that is, *<sup>X</sup>* <sup>=</sup> {(*x*, *<sup>y</sup>*) <sup>∈</sup> <sup>R</sup>}. The set of time is a subset of the real numbers, so the dynamical system will be said to be continuous; that is, *<sup>T</sup>* <sup>=</sup> {*<sup>t</sup>* <sup>∈</sup> <sup>R</sup> : *<sup>t</sup>* <sup>≥</sup> <sup>0</sup>}. Finally, the evolution law is given by the solution of the differential system of Equation (4).

The mathematical perspective from dynamical systems is relevant in a trend behavior analysis because it allows characterizing the qualitative structure of the system's behavior beyond just understanding the behavior for a specific time, which could be misleading. Roughly speaking, the dynamical systems identify invariant sets and characterize their stability. Invariant sets are states that do not change as the system evolves, i.e., state sets that remain in time. Examples include equilibrium points, periodic orbits, and strange attractors. The stability can characterize the system trajectories as attracting, repelling, fixing (centers), or combining these three options.

Since the equilibrium points occur when the system does not change, it must be satisfied that *x*˙ and *y*˙ are zero [60,61], making the differential problem an algebraic issue, as follows:

$$\begin{aligned} \dot{\mathfrak{x}} &= 0 \\ & \implies \\ \dot{y} &= 0 \end{aligned} \implies \begin{aligned} \mathfrak{x}(\mathfrak{a} - \delta y) &= 0 \\ y(\beta - \lambda \mathfrak{x}) &= 0 \end{aligned} \tag{8}$$

The stability of the equilibrium points is calculated with the eigenvalues of the evolution matrix of the differential system around the equilibrium points [60,61], i.e., consider Equation (4) around the equilibrium points in the form *X*˙ = *JX*, where *X* = (*x*, *y*)*<sup>T</sup>* is the state vector, and *J* is the Jacobian matrix defined as:

$$f(\mathbf{x}, y) = \begin{pmatrix} \frac{\partial \boldsymbol{x}}{\partial \mathbf{x}} & \frac{\partial \boldsymbol{x}}{\partial y} \\ \frac{\partial \boldsymbol{y}}{\partial \mathbf{x}} & \frac{\partial \boldsymbol{y}}{\partial y} \end{pmatrix} = \begin{pmatrix} \boldsymbol{\alpha} - \delta \boldsymbol{y} & -\delta \boldsymbol{y} \\ -\lambda \boldsymbol{y} & \beta - \lambda \boldsymbol{x} \end{pmatrix} \tag{9}$$

By the Hartman-Grobman theorem [60,61], the nonlinear behavior so close to an equilibrium point is the linearized behavior given by the Jacobian matrix in the equilibrium point. So, the evolution matrix around an equilibrium point (*x*¯, *y*¯) is *J*(*x*¯, *y*¯).

The eigenvalues *χ* are defined as the roots of the characteristic polynomial *<sup>χ</sup>*<sup>2</sup> − *tr*(*J*)*<sup>χ</sup>* + *det*(*J*), where *tr*(*J*) is the trace of the Jacobian matrix *<sup>J</sup>* obtained by summing the diagonal entries of *J*, and *det*(*J*) is the determinant of the Jacobian matrix.

A more robust analysis, called bifurcation analysis, also seeks to identify how the system parameters' variation can give rise to the appearance/disappearance of invariant sets or change their stability. This bifurcation analysis consists in finding system perturbations that transform its phase portraits (system trajectory structure) [60,61]. These perturbations are represented by the variation of the system parameters in such a way that if the variation of *n* parameters qualitatively changes the phase portraits, it is said to be a bifurcation of codimension *n* [60,61].

#### **3. Results**

The results presented in this section are based on the dynamical system analysis of Equation (4). These ordinary differential equations were programmed and solved using the *ODE*45 in the MATLAB® software. Then, and in order to keep the sense of the application, four transition families in the nonnegative space of the state variables were found. To clarify, from a qualitative perspective, a transitions family is a set of similar behavior

curves (topologically equivalent trajectories in the states space). For example, two orbits that converge to the same equilibrium point showing a sigmoidal behavior are said to be topologically equivalent, although the route of their trajectories is different (quantitatively different). Usually, these behaviors are easily visualized in the phase portrait, which is a representation of all possible solutions or trajectories of a system of differential equations for any initial condition and a specific set of parameter values. As shown in Figures 2 and 3, each of these transition families is displayed as a *different* phase portrait.

By solving Equation (8), it is found that Equation (4) has the equilibrium *Eq*1(0, 0) and *Eq*2(*β*/*λ*, *α*/*δ*), each of which has a stability that depends on the values of the *α* and *β* parameters. For the equilibrium *Eq*1, the eigenvalues are *α* and *β*; while for the equilibrium *Eq*2, the eigenvalues are <sup>±</sup>*αβ*. This leads to four cases called transition families; each one is presented in Figures 2 and 3.

The first family transition corresponds to two unstable nodes (see the phase portrait in Figure 2a). The first is located at the origin (source type), while the second is in the nonnegative states space (saddle type). Because it is possible to have different trajectories in the phase portrait depending on the initial condition, this family transition has four distinct trend behaviors (which is why we call the transition scenarios families).

For the first case, see Figure 2a,b, a behavior is visualized in which the two technologies grow over time until they reach a particular point, in which electric automobiles tend to disappear, while the conventional vehicles become dominant. For the second case, see Figure 2c,d, the behavior is similar, but the disappearance now occurs for conventional automobiles, while electric vehicles become dominant. In the third case, see Figure 2e,f, there is a drop in the technologies used that, after a specific time, ends and causes electric automobiles to recover and become the dominant technology. Finally, in the fourth case, see Figure 2g,h, there is again an abandonment of the two technologies, but the conventional technology recovers in the long term.

The second family transition corresponds to an unstable node at the origin (saddle type), see the phase portrait in Figure 3a. Another equilibrium point is outside the nonnegative states space (center type), but its analysis is irrelevant for the application. This second family transition is the one that looks like the most expected transition: conventional automobiles diminish until they disappear, while electric automobiles gradually increase their dominance of the market, until they are the only type of vehicle (see Figure 3b).

The third family transition corresponds to an unstable node at the origin (saddle type), see the phase portrait in Figure 3c. Another equilibrium point is outside the nonnegative states space (center type), but its analysis is irrelevant for the application. This scenario appears as a theoretical possibility, and does not represent the interests of those who talk about the transition of vehicle technologies; it shows the strengthening of conventional automobiles, while electric automobiles disappear (see Figure 3b).

The last family transition corresponds to a stable node at the origin (sink type), see Figure 3e. Another equilibrium point is outside the nonnegative states space (saddle type), but its analysis is irrelevant for the application. In this scenario, the absence of interest of consumers in the two types of technologies leads to their disappearance as transport alternatives (see Figure 3f).

**Figure 2.** Case 1 simulations. This case occurs when vehicle entry rates are higher than exit rates in both categories. In (**a**,**b**), both technologies grow; however, finally, the conventional one dominates, while in (**c**,**d**), the opposite occurs. In (**e**,**f**), both technologies decrease, but the electric one manages to recover; while in (**g**,**h**), the opposite occurs.

**Figure 3.** Simulations for cases 2, 3 and 4. This figure corresponds to another case of entrance rates growth being higher than exit rates in both automobile categories. In (**a**,**b**), the scrapping of conventional automobiles is greater than the entry into the fleet (sales), while in (**c**,**d**), sales of conventional automobiles are greater than their operating output (scrap), which increases the number of automobiles in the fleet. Finally, in (**e**,**f**), the lack of interest in the two technologies can be seen, causing their disappearance.

#### *Two-Dimensional Bifurcation Analysis*

In principle, one-dimensional bifurcation diagrams are computed when it is necessary to study the stability or behavior of a dynamical system under the variation of any of its parameters. On the other hand, when it is necessary to study the stability or behavior of a dynamical system under the simultaneous variation of two important parameters, then a two-dimensional bifurcation diagram is required [60]. In essence, two-dimensional bifurcation diagrams show different colors, each of which represents a precise combination of the two parameters leading to a specific system behavior.

To obtain these bifurcation diagrams, it is necessary to program a behavior detector, so that when a specific behavior is detected, the program stores the parameters leading to such dynamics, and an identification number will represent a color in the diagram.

Since the equilibrium points' stability of the dynamical system in Equation (4) depends on the bifurcation parameters *α* and *β*, it is possible to obtain a two-dimensional bifurcation diagram, as depicted in Figure 4. This diagram summarizes the behaviors displayed by the system for the different values of the two bifurcation parameters. Notably, the parameter array with the lowest probability of occurrence is for the first family of transitions (in the figure, this family appears as case 1 for graphical simplicity). The parameter arrays to obtain transition families 2, 3, and 4 have been denoted as case 2, case 3, and case 4, respectively. The probability areas of these last cases are broader than case 1; so, in theory, their occurrence chances must be greater than case 1.

**Figure 4.** Two-dimensional bifurcation diagram varying *α* and *β*. Every color represents the system behavior for each case previously analyzed.

#### **4. Discussion**

Below is a discussion of the behaviors found in each transition family of vehicle technology from environmental drivers. Each of these transition families would allow the recognition of a different type of transition worldwide, allowing their classification, as will be shown next. Furthermore, we present the possible future work of this research.

#### *4.1. Family Transition 1*

From the application perspective, transition 1 is interesting, because the stable manifold and the unstable manifold of the saddle node define four key regions whose behaviors are different.

The trend behaviors presented in Figure 2b,d can be identified with the current behavior, in which the number of conventional and electric automobiles continues growing globally. The long-term outcome is the one that can change due to the final dominance of one of the two technologies. Many nations would be classified in this state today. However, in developed countries, the behavior could be similar to the one shown in Figure 2d, due to the greater investment capacity and awareness of climate change that its inhabitants have; while in underdeveloped countries, it is easier to interpret their trend behavior as that shown in Figure 2b, due to the survival situation of their inhabitants, in addition to the massive confluence of the population in urban centers, where restrictive transport measures and the inability to purchase an electric vehicle due to other criteria than environmental ones (such as the high price of these technologies) determine the purchase of a second conventional vehicle.

On the other hand, the trend behaviors presented in Figure 2f,h would correspond to the assumption that the number of conventional and electric automobiles is so high that there is a general rejection of these automobiles use because of their environmental impacts over climate change and land health (related with hazardous waste). This leads in the long term to the disappearance of one of the two technologies, while the other one becomes dominant. This behavior can be considered in nations in which vehicle use is abandoned in favor of migrating towards public transport and active mobility.

Despite the fact that the behavior shown in this transitional family appears to be the least probable in the bifurcation diagram of Figure 4, the values for which it is satisfied are not utopian. Nor it is utopian to think that the engine of this de-escalation of conventional and electric automobiles could occur due to the impacts they generate on the environment.

#### *4.2. Family Transition 2*

The trending behavior presented in Figure 3b can be pointed out as the most natural and expected within the transitions. The trajectories show a smooth transition from conventional to electric technologies, which is how a handover would occur.

It is the most desired transition, led by environmental awareness about climate change and its consequences on global socioecological systems. In this transition, the issue of hazardous waste is an issue on which we can still wait.

#### *4.3. Family Transition 3*

The trending behavior presented in Figure 3b shows that automobiles with electric technologies are not convincing due to their effects on the environment related to the management of their hazardous waste.

It should be remembered that the model does not consider the mining activities and their social and ecological implications in the nations where the raw materials are obtained. This should be included in future work for analyzing these transitions. It is intuited that its inclusion in the model could generate new transition behaviors, but it must be tested.

The question would be: how could this scenario be viable when fossil fuel reserves will be scarce so soon?

#### *4.4. Family Transition 4*

Finally, the trending behavior presented in Figure 3f shows the lack of interest in the two technologies, causing their disappearance. In this case, no negative environmental effects are prioritized, showing losing interest in the technologies based on conventional or electric mobility and migrating to alternatives such as active mobility or the use of public transport.

#### *4.5. Future Research*

The research group recognizes that the only transition drivers are not those related to a full awareness of the environmental implications of each of the vehicle technologies presented. We can also list those corresponding to the price of the vehicle ([50,62]), the charging infrastructure [47,63], the energy value [64], the incentives offered by the government [18,44,51], and the useful life of the technology [38,65].

In this way, the future work must include other criteria in the modeling, with the corresponding analysis of the trend behavior that each new criterion defines on the system behavior.

However, the related issue that makes this inclusion nontrivial lies in the bifurcation analysis, through which all possible prospective scenarios for the system configuration could be visualized [66].

We also expect to obtain management recommendations that at least guarantee adequate management of the related environmental issues to each family of transitions.

#### **5. Conclusions**

This work allows the classification of the different types of technological transitions that are taking place around the world, from conventional automobiles to electric automobiles, motivated by a single decision criterion for consumers based on environmental awareness.

The transition scenario definition has not been defined a priori but has emerged from systemic modeling and its corresponding bifurcation analysis.

Due to the bifurcation analysis, it can be assured that the trend behaviors for the configuration and assumptions made to the system are completely bounded to the four families of transitions listed in this work. In this way, specific recommendations for the four transition families can be generated rather than general ones that end up ignoring the complexities of each of the configurations.

Promoting environmental awareness and disseminating the environmental benefits of electric vehicles can contribute to the growth of the fleet, increasing its popularity among a segment of buyers whose decision criterion is protecting the environment. For this popularity not to diminish, it is necessary to establish actions and programs that allow the creation of a relevant value chain for the waste generated by the use of EVs, such as lithium or nickel. This will allow the reduction in mining exploitation. These are public policy actions that could promote the exit of the circulation of conventional technologies and contribute to the circular economy commitments of the countries.

On the other hand, if the two technologies are not of interest to users, it is relevant to strengthen public transport and active mobility in populated centers, guaranteeing road infrastructure, safety, quality, and equitable access. This will imply the development of public policies and government investments.

Finally, the energy transition of the transport sector is a priority step in the decarbonization of anthropogenic activities to tackle climate change. However, this work shows that there are still many challenges for reducing or eliminating the impacts that energy systems have on nature. This raises questions about energy sources, the materials used, processes' efficiency, etc. In this way, this work reinforces the evidence that the present energy solutions are short-term and present a challenge to physicists and engineers to propose solutions that satisfy long-term multidimensional needs.

**Author Contributions:** Conceptualization, J.S.G., J.D.M., J.M.R. and M.B.-F.; methodology, J.M.R. and J.D.M.; investigation, J.S.G. and M.B.-F.; validation, J.M.R. and J.D.M.; formal analysis, J.M.R. and J.D.M.; writing—original draft preparation, J.S.G., J.D.M. and J.M.R.; writing—review and editing, J.S.G., J.D.M. and J.M.R.; visualization, J.D.M. and J.M.R.; project administration, J.S.G.; funding acquisition, J.D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Universidad de Monterrey and Universidad Católica de Colombia under the project CON0000497 "Dynamic model for renewable energy supply in Colombia".

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors want to acknowledge Universidad de Monterrey and Universidad Católica de Colombia for their support of this research work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article* **COVID-19***-* **s Pandemic Effects on Bike Sharing Systems: A New Reality for Urban Mobility?**

**Efstathios Bouhouras 1, Socrates Basbas 1,\*, Stamatia Ftergioti 2, Evangelos Paschalidis <sup>3</sup> and Harris Siakantaris <sup>4</sup>**


**Abstract:** On 11 March 2020, the World Health Organization made the assessment that a new disease (COVID-19) caused by a novel coronavirus (SARS-CoV-2) could be characterized as a pandemic. From that point, a chain reaction of events and difficult decisions requiring action was launched. National governments all over the world announced partial or total quarantine lockdowns in an effort to control the virus' spreading in order to save as many lives as possible. The effects of the pandemic were multifaceted and transport was not excluded. The current paper examines data regarding the level of usage (provided by the administrator) of bike-sharing systems in three case studies/medium-sized Greek cities (Igoumenitsa, Chania, and Rhodes) and through a statistical analysis identifies if there is a correlation between the implemented measures and the modal choice of the residents. The main results and conclusions of this analysis reveal that the level of usage of these specific bike-sharing systems was significantly increased during the lockdown period compared to the situation before the lockdown and the pandemic in general.

**Keywords:** COVID-19; bike sharing system; urban mobility; regression analysis

#### **1. Introduction**

The first case of COVID-19 in Greece was recorded on 26 February 2020. From that day and in less than a month, the Greek government began the implementation of restrictions which led to a general quarantine lockdown, affecting people's mobility, thus creating a new reality for urban mobility (modal shift, mobility restrictions, etc.). It is rather clear that COVID-19 has affected, and continues to affect, people's transport mode choice. According to surveys "*consumers are already showing an increased cautiousness about health, meaning that safety is likely to be a more important factor for consumers when deciding if, how and when to move*" [1].

Three major European cities during the first months of the pandemic announced measures that allowed bikers and pedestrians to comply with social requirements. Specifically, according to the press, authorities in Milan, Italy, decided to reallocate street space from private cars to cycling and walking, in response to COVID-19. Over 35 km of streets were to be transformed over the summer of 2020, through a rapid experimental citywide expansion of cycling and walking space in order to protect residents. The Strade Aperte plan states: "*included low-cost temporary cycle lanes, new and widened pavements, 30kph speed limits and pedestrian and cyclist priority streets*" [2].

Another European city adopting measures aiming to promote cycling and walking is Paris, France. According to the press, an investment of over 300 million EUR is underway in the Ile-de-France area for the construction of an extensive network of separated

**Citation:** Bouhouras, E.; Basbas, S.; Ftergioti, S.; Paschalidis, E.; Siakantaris, H. COVID-19 s Pandemic Effects on Bike Sharing Systems: A New Reality for Urban Mobility? *Appl. Sci.* **2022**, *12*, 1230. https://doi.org/10.3390/app 12031230

Academic Editors: Jianbo Gao, Giovanni Randazzo, Anselme Muzirafuti and Dimitrios S. Paraforos

Received: 30 November 2021 Accepted: 17 January 2022 Published: 25 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

broad cycle paths and highways. The local authorities highlighted the fact that cycling is one of those transport means that allows users to maintain necessary and appropriate social distancing and therefore should be supported in order to play its role, not only during the COVID-19 crisis but also after its end [3]. In addition, the World Health Organization has urged cities to make full use of cycling and walking. Specifically, on a brochure published by the WHO entitled "Moving around during the COVID-19 outbreak", it is mentioned that whenever feasible, cycling and walking should be selected since they provide physical distancing while helping people meet the minimum requirement for daily physical activity [4].

Finally, authorities in the city of Brussels, Belgium, decided that the area of Pentagon (city center), the district within the Kleine Ring, will become a residential area at least during the COVID-19 crisis. The measure endured for 3 months, from 11 May 2020, and during that period (a) pedestrians were allowed to use the full width of the public road and not only the pedestrian paths but (b) vehicles were also allowed inside that area with a maximum speed of 20 km/h to ensure pedestrians' priority and cyclists' safety and (c) existing pedestrian zones retained their status [5].

Another press release regarding the performance of bike-sharing systems in the United States of America during the first months of the COVID-19 pandemic, highlights the fact that the bike-sharing systems in Boston, Chicago, and New York (all administrated by the same company) were free for healthcare or public transportation workers and first responders. Moreover, the bike-sharing system in Washington D.C. was free of charge to essential workers while the respective system in Minnesota was free for healthcare workers. The Houston bike-sharing system saw increased usage, even after COVID-19 precautions closed some stations to prevent gatherings in parks. Many systems—including Kansas City, Detroit, and Memphis—were temporarily offering unlimited free rides for all users. Another aspect of the COVID-19 pandemic effects on the bike-sharing systems in the United States concerns the fact that "*some systems had developed partnerships with local restaurants, allowing them to use their bikes for food delivery. Of the more than 189 cities with bike sharing systems in the U.S., only a few had shut down during the time of the pandemic*". However, there were many cases of bike-sharing systems that presented a significant decrease in their level of usage due to the fear of contagion among other reasons (Santa Monica, Central Los Angeles, West Los Angeles, and North Hollywood) [6].

A questionnaire-based survey conducted by the Laboratory of Renewable and Sustainable Energy Systems of the Environmental Engineering Department of Technical University of Crete, Greece, under the framework of the Horizon 2020 CIVITAS DESTINATIONS research project, addressed to the residents of the cities of Chania and Rethemno, Crete, during the period 16–22 March 2020, revealed that the percentage of choosing bicycles as their transport mode during the previously mentioned period compared to the period of January–February 2020, increased by almost 18% [7], embossing the dynamic of the bicycle as a primary transport mode.

The effect of the COVID-19 pandemic on bike-sharing systems is still being monitored and recorded by researchers worldwide. Surveys have been implemented focusing on collecting information regarding individuals' immediate responses to the travel restrictions during the pandemic in specific areas [8,9] while others have covered larger areas [10,11]. Researchers are focused on four major topics: (a) environmental quality [12–14], (b) socioeconomic impacts [15], (c) management and governance, and d) transportation and urban design [16–19] without excluding urban road freight transport [20]. Many of the abovementioned research activities are focused on the impacts of implemented restriction measures concerning urban mobility in an effort to control the COVID-19 outbreak [8,21–23] by introducing, for example, terms such as "responsible transport" [24] or by applying Big Data from mobile spatial statistics attempting to estimate population density patterns in order to customize applied measures to the characteristics of urban areas [25].

Based on the information and knowledge gained through a literature review on this topic (COVID-19 pandemic and mobility restriction measures), it was decided to develop

a paper regarding BS systems in Greece. The present paper examines the performance of the bike-sharing system in three Greek cities (Igoumenitsa, Rhodes, and Chania) comparing the periods before, during, and after the general lockdown, based on data provided by the operator (Cyclopolis) of these bike-sharing systems. An attempt has been made to identify changes in travel behavior related to the usage of BS bicycles, aiming to assist policymakers and policy takers to better design and implement appropriate and efficient customized policies. Moreover, the authors' intention was not to develop a forecasting model or to explain horizontally how restriction measures affected the performance of bike-sharing systems, but through the findings of this paper to provide a small amount of knowledge to the researchers in the specific field and help them understand and compare travel behaviors and impacts in other cities/countries.

Analysis of the provided data was performed by developing a regression model with aggregated data and thus conclusions could be extracted concerning the performance of the bike-sharing systems across three Greek cities. Based on the first level analysis and its findings, it was decided to develop a regression model using aggregated data in an effort to identify correlations among the parameters describing the BS systems' performance. Since such data were not available at the time the research took place, we were unable to deal with the modal shift in the framework of the present paper. The second section (Materials and Methods) describes the methodology followed for the data collection and analysis while in the third section (Results), the key findings of the descriptive and indepth statistical analysis are described. Finally, in the Discussion and Conclusions section, the limitations of the study are presented, along with the main conclusions and future research.

#### **2. Materials and Methods**

The methodology followed in the current paper was defined mostly by the availability of the collected (by the operator of the selected BS systems) relevant data (number of trips, number of registered users, time period of recorded data, etc.) as well as the examination of the available literature (similar cases, methodologies developed for evaluating the BS systems' performance during the pandemic). Concerning the available data, the contribution of the operator (Cyclopolis) of several BS systems in Greek cities was crucial for the implementation of the research. The examination of the relevant literature/references revealed that the majority of the information came from press publications and that there were not many scientific publications on the issue during the first six months of the pandemic in Europe.

The methodology followed for the development of the current paper can be summarized in the following steps:


The first concerned the lockdown period and the second regarded the periods before and after the lockdown. During these periods, independent and dependent variables were set and then examined in order to identify which was the effect of the pandemic.

To evaluate the effect of general quarantine lockdown on the performance of bikesharing systems in Greek cities, we used data provided by the operator of the bike-sharing systems (Cyclopolis). It must be mentioned that in the cities of Rhodes and Chania, public transport is operational throughout the entire year while in the city of Igoumenitsa it is operational only during summertime (June–August) and for specific routes (mainly connecting the urban areas with the beaches located near the city). Daily data were collected for three Greek cities (Chania, Igoumenitsa, and Rhodes) for a rather long time. Specifically, for the cities of Chania and Rhodes, collected data were available from May 2019 to May 2020 while for the city of Igoumenitsa data were available for the time period July 2019–July 2020. However, in order to analyze the data in the same time period, the number of trips, avoided cars (the number of private cars that will not be used due to the usage of BS bicycles), and avoided CO2 emissions were examined for the period May 2019–May 2020 by using simple equations. To proxy the performance of the bikesharing systems, we used the number of rides per week in a given city (a ride refers to the process of an individual renting a bicycle and following a route of their choice before returning it to a rental station, aka a trip). Thus, our dependent variable was the weekly number of rides (Rides) recorded in the city's bike-sharing system.

Due to limitations in data availability, we used a reduced number of explanatory variables. The main variable of interest was a dummy variable (Lockdown 23 March–4 May 2020) which captures the general quarantine lockdown and all the restrictions imposed due to the COVID-19 pandemic in Greece (it should be noted that on 4 May 2020 the Greek government announced the gradual de-escalation of restrictions that lasted until July 2020, including the partial lifting of movement restrictions and not the total lifting of restrictions). This variable takes the value of 1 during the quarantine lockdown period, i.e., the weeks between 23 March and 4 May 2020 and zero otherwise. Additionally, we included an alternative dummy variable (Lockdown 23 March–forward), which takes the value of 1 during the quarantine lockdown period and beyond, i.e., all the weeks after 23 March 2020 even after the easing of restrictions on 4 May, and zero all the weeks before that date. We assumed that quarantine lockdown was positively related to the performance of bike-sharing systems, increasing the number of rides during the quarantine period and beyond. Therefore, we expected a positive sign for both variables, implying that the quarantine lockdown led to an increased use of rental bicycles. Through this approach, an increase in the BS system would be proven to be a conscious choice of the users and not a random incident.

In order to examine the trips' duration, we include the average duration of the rides (Duration) per week, expressed in minutes. Although the trips' duration could be associated with the number of rides, we do not have a priori expectation on the sign of this variable. Finally, taking into account the form of bike-sharing systems and renting infrastructure within each city, the number of rental stations (Stations) was included in the set of regressors. The automatic bike-sharing system of the municipality of Igoumenitsa consists of three rental stations, the municipality of Rhodes has seven rental stations while the municipality of Chania had four rental stations until January 2020 which were reduced to three from February 2020 onwards. Since BSS bicycles were rented and returned by the users to a rental station, we expected a positive effect of the number of available rental stations on the use of BSS bicycles.

Our analysis consists of panel data that contains observations collected at a regular frequency across similar units (i.e., cities in Greece). The use of one panel data model instead of three separate time series models has the advantage of containing more information, variability, and more efficiency than pure time series data. This occurs because panel data models allow for heterogeneity across units (i.e., cities), namely, allowing for any or all the

model parameters to vary across cities whereas time series models assume that the model parameters are common across cities.

Furthermore, the idea of developing three discrete (each for a city examined in the framework of this paper) models was examined but not followed as the restriction measures applied were the same for all of Greece and for the same time period and concerned similar BS systems under the same administrator; thus, it was decided that one model should be developed covering all three cities. Table 1 presents the descriptive statistics and specifically the values of the observations as provided by the administrator. These values were examined and the marginal values were excluded by the model's developing process, although they represent real cases (for example, the minimum value regarding the average duration in minutes is recorded in one of the three BS systems and refers to a case in which the system failed to properly record the duration of the user's trip).

**Table 1.** Descriptive statistics.


The final step of the followed methodology concerns the development of the paper and presents the results of the data analysis by most of the effects of the pandemic on the performance of the BS systems based on the users' choices made before, during, and after the lockdown period in the above-mentioned Greek cities.

The bike-sharing system in Igoumenitsa was initially installed and operated in the framework of the ADRIMOB (Sustainable coast MOBility in the ADRIatic area) Project by the Regional Unit of Thesprotia. The project was implemented under the European Territorial Cooperation Programme IPA ADRIATIC 2007–2013 [26]. This is considered the first phase of the bike-sharing system in Igoumenitsa, during which the system was free of charge. The system allowed access to a bike in an automated manner by calculating the actual time of the use of the bicycles. It could accommodate people who were subscribers and made use of the special electronic card EasyBike. The bike-sharing system was operative 24/7 but the working hours of the two contact points were Monday to Friday, 08:00–15:00. The number of registered users was rather small. In 2015, 22 EasyBike cards were subscribed, while in 2016 the number was increased to 25 cards and in 2017 only 1 new card was subscribed. There were two rental stations available equipped with 10 bicycles in total. In the city of Igoumenitsa, there are two bicycle roads constructed and operating. The first one (bicycle and pedestrian road simultaneously, mixed usage) begins near the facilities of the Technological Institute of Epirus at the 28th Oktovriou Str. The bicycle–pedestrian road was constructed in 1998 (total length = 2.8 km and width = 4 m, two-way road) and ends at the road heading to Drepano Beach, which is located 1.5 km from the end of the bicycle–pedestrian road. The users, if they want to continue their trip to Drepano Beach, must use the road constructed for all traffic, which is very dangerous, especially during summertime, when the traffic volumes from and to Drepano Beach are high. The second bicycle road is an urban bicycle road along Leoforos 49 Martiron (location: Ladohori–Igoumenitsa). The total length is 1.1 km, the width of the surface used by the

bicycles is approximately 2 m with pavement on both sides. The road crosses roundabouts and the existing horizontal and vertical signs are considered to be adequate [27].

The second phase of the bike-sharing system in Igoumenitsa concerned the time period during which the SUMPORT project was implemented. Aiming to further promote sustainable mobility, a new rental station was installed near the facilities of the Technological Educational Institute of Epirus. The new station of the new and currently active BS system in Igoumenitsa is presented in Figure 1 [28]. Through this map, the user is informed not only about the location of the rental stations, but by choosing any of the stations, is informed in real time about the capacity of the station (each station can host up to 6 bikes) as well as the number of available bikes and the number of free docking slots.

**Figure 1.** Interactive and real time informative map of the new bike-sharing system in the city of Igoumenitsa [29].

A comparison of the previous and current forms of the bike-sharing system in the city of Igoumenitsa is presented in Table 2 regarding their basic characteristics. The specific comparison highlights that the current system in the city of Igoumenitsa is more organized and user friendly than the previous, encouraging the residents and visitors to use it in an easier way and thus avoiding significant bureaucratic problems. The current system became fully operational just a few months before the COVID-19 pandemic.

Upon securing the assent of the Intermediate Managing Authority of the South Aegean Region for the tendering of the project titled 'Procurement and installation of bicycle stations in the city of Rhodes' within the context of the 'Crete and the Aegean Islands 2017–2013' Operational Programme, the Municipality of Rhodes conducted an open tender for the procurement and installation of public-use bicycle stations (60 bicycles, 5 bicycling stations, and an operating system) [30]. Subsequently, a relevant contract was concluded between the Municipality of Rhodes and the company Cyclopolis Ltd., London, UK, concerning seven renting stations in total (Aquarium, Marina, Mitropoli, Ag. Nikolaos Square, Eleftheria Square, Symis Square, Ag. Athanasios—Ag. Fragkiskos Gateways). According to the bike-sharing system operating rules, BSS is defined as the automated bike-sharing system

implemented by the municipality of Rhodes, permitting the short-term rental of bicycles. The operation and management of the BSS will be undertaken by the competent services of the municipality and include maintenance of the bicycles and consumables, redistribution of bicycles among stations, promotion of the system, commercial exploitation of the system, and the general operation and management of the system.

**Table 2.** Comparison of the previous and current forms of the bike-sharing system in the city of Igoumenitsa.


The automatic bike-sharing system of Chania's municipality consists of four renting stations (Defkalionos Square, Katehaki Square, Talo Square, and Markopoulou Square) and 70 bicycles in total of which 50 are available for use. The stations consist of info kiosks where the users can borrow a bicycle by using a touch screen [31].

#### **3. Results**

#### *3.1. Timeline of COVID-19 Related Events*

The timeline of the events related to the COVID-19 pandemic in Greece that led to a general quarantine lockdown is presented in Table 3. The lockdown was announced on 23 March 2020 and endured until 4 May 2020, when the general government announced the beginning of easing restrictions.

**Table 3.** Timeline and milestones of COVID-19 pandemic in Greece for the period February 2020–May 2020.


In order to fully understand the imposed restrictions concerning the mobility of the Greek people, the central government on 23 March 2020 announced that all nonessential movements throughout the country were restricted. Since that date, Greek citizens were allowed to move outside their houses only for the following cases: (i) moving to or from one's workplace during work hours, (ii) going to the pharmacy or visiting a doctor, (iii) going to a food store, (iv) going to the bank for services not possible online, (v) assisting a person in need of help and (vi) going to a major ritual (funeral, marriage, baptism) or movement, for divorced parents, which is essential for contact with their children, and (vii) moving outdoors for exercising or taking one's pet outside, individually or in pairs.

#### *3.2. Exploratiry Data Analysis*

The analysis of the data provided by the administrator of the BS system in Igoumenitsa for the period July 2019–April 2020/May 2020 (depending on the parameter) is presented in Figures 2–4. Specifically, Figure 2 presents the evolution of the number of yearly subscriptions per month.

The increase in yearly subscriptions is important, especially for the period January 2020–April 2020, which is 68%, while the increase for the period February 2020– April 2020 is 38%. In order to better understand the importance of this increase not only in relation to COVID-19 but also in relation to the weather conditions, it must be noted that the average temperature in Igoumenitsa for January was 14 ◦C, for February was 15 ◦C, for March was 16 ◦C, and finally for April was 19 ◦C [32]. The average rain precipitation based on historical data for the city of Igoumenitsa is the following: January = 120 mm, February = 125 mm, March = 100 mm, and April = 60 mm [33]. Figure 3 presents the evolution of the total number of registered users for the period July 2019–April 2020.

**Figure 2.** Evolution of the number of yearly subscriptions per month for the time period July 2019–April 2020.

**Figure 3.** Evolution of the total number of the bike-sharing system's registered users in the city of Igoumenitsa for the time period July 2019–April 2020.

**Figure 4.** Evolution of Igoumenitsa's bike-sharing system—number of rides per month and average duration.

It is clear that after the imposition of restriction measures concerning people's mobility, the number of registered users in the bike-sharing system in Igoumenitsa was significantly increased. Specifically, in the period February 2020–April 2020, the increase was 97% while the respective increase between March 2020 and April 2020 was 35%. Although restrictions were in force on 23 March 2020 concerning the mobility of people, alongside extended police controls, the number of registered users was significantly increased. Moreover, as presented in Figure 4, the number of rides per month doubled during the period February 2020–May 2020. Specifically, the increase (for the above-mentioned time period) was 269%, while between March 2020 and May 2020 was 93%. Figure 4 also presents the average duration of the rides per month. Since the restrictions were terminated, the number of rides during June 2020 was decreased to half compared to May 2020.

Another interesting finding regarding the characteristics of the system's performance in the above-mentioned three cities regard the summed up (total) number of rides per day, before (1–23 March 2020), and during the lockdown (24 March 2020–4 May 2020) (see Figure 5 and Table 4). The usage of BS bicycles was significantly increased during the lockdown in all cases. It is estimated that the residents of these cities used the BS bicycles not only for leisure trips but also in order to move for their daily needs.

**Figure 5.** Comparison of summarized (total) number of rides per day before and during the lockdown in the cities of Chania, Rhodes, and Igoumenitsa.


**Table 4.** Number of rides for Igoumenitsa, Chania, and Rhodes during time periods before and during the lockdown.

Furthermore, the analysis of the collected data revealed that the majority of the rides during February, March, and April 2020 potentially concerned urban trips as well as leisure purposes, as the bicycles were parked at stations different than those rented [34]. This assumption was concluded by taking into consideration the locations of the stations and the most important landmarks in the examined cities (retail market, touristic and cultural sightseeing, coastal line, etc.). During the implementation of the current bike-sharing system in the city of Igoumenitsa, the operator estimated that over 400 L of fuel was saved due to the modal shift from private cars to bicycles, preventing almost 700 kg of CO2 from being emitted into the atmosphere.

In the framework of the SUMPORT project, key performance indicators (KPIs) were developed in order to be used for monitoring and evaluating the performance of the bikesharing (and not only) systems [35]. The indicators used for monitoring and evaluating the performance of Igoumenitsa's BS system were the following: (a) The number of rented bikes (provided by the operator), (b) km made by the bicycles (provided by the operator), (c) the number of avoided cars (calculated in house based on assumptions. Specifically, it is assumed that two rides would be equal to one avoided car), (d) the number of avoided km due to the usage of the BS system (calculated in-house based on assumptions and specifically the average distance covered by a car is equal to 15 km) and (e) the total amount of saved CO2 (calculated in-house based on assumptions and specifically that each car produces 200 g CO2 per km). The comparison of the performance of the BS systems in Igoumenitsa, Rhodes, and Chania is based on the above-mentioned indicators. Table 5 presents the performance of Igoumenitsa's BS system using the above-mentioned indicators, while Figure 6a,b presents the performance of Igoumenitsa's system using combo charts for two different sets of variables: (a) The number of rides per month and the number of avoided cars—avoided km and (b) the number of rides per month and the number of avoided cars—avoided emissions (CO2).

**Table 5.** Performance of Igoumenitsa's BS system using the SUMPORT key performance indicators (own process).


**Figure 6.** (**a**) Combo chart for Igoumenitsa's BS system relating number of rides per month and avoided cars to number of avoided km. (**b**) Combo chart for Igoumenitsa's BS system relating number of rides per month and avoided cars to number of avoided emissions (kg of CO2).

The same indicators were applied for the cases of the cities of Rhodes and Chania. The results are presented in Tables 6 and 7 (see also Figure 7a,b for the city of Rhodes and Figure 8a,b for the city of Chania), respectively.




**Figure 7.** (**a**) Combo chart for Rhode's BS system relating number of rides per month and avoided cars to number of avoided km. (**b**) Combo chart for Rhode's BS system relating number of rides per month and avoided cars to number of avoided emissions (kg of CO2).

**Figure 8.** (**a**) Combo chart for Chania's BS system relating number of rides per month and avoided cars to number of avoided km. (**b**) Combo chart for Chania's BS system relating number of rides per month and avoided cars to number of avoided emissions (kg of CO2).

A comparison of the evolution of the total number of rides as well as the average trip's duration for the period July 2019–May 2020 for the above-mentioned three cities is presented in Figures 9 and 10, respectively. The black dotted frame represents the time period during which the restricting measures were applied.

Another interesting finding from the first-level analysis of the provided data (the second-level analysis or in-depth analysis of the data concerned the exploration of the existence or not of correlations among the parameters of the bike-sharing systems based on the choices of the users) concerns the choice of the users to return or not return the bicycles to the renting stations from which they rented them. It can be assumed that this users' choice can assist in building an origin–destination matrix. An origin point is considered the renting station from which the user rents the bicycle and a destination point refers to the renting station to which the user returns the bicycle. The analysis of the collected data in an aggregated manner is presented in Figure 11. Specifically, for the city of Chania, two approaches were followed and are presented regarding the Talo renting station, as it was operational only for the year 2019. Therefore, the first approach takes into consideration this fact while in the second the station is totally ignored and excluded from the analysis.

**Figure 9.** Comparison of the total number of trips for the cities of Igoumenitsa, Rhodes, and Chania during the period July 2019–May 2020.

**Figure 11.** Percentage of BSS bicycles returned by the users to their initially rented station for the cities of Igoumenitsa, Rhodes, and Chania for the period July 2019–May 2020.

#### *3.3. Number of Rides Regression Model*

For the purpose of our analysis, daily data provided by the operator (Cyclopolis) of the BS systems were converted into weekly data, which were then used to construct an unbalanced panel dataset of the bike-sharing system in three Greek cities (Igoumenitsa, Rhodes, and Chania) from July 2019 to May 2020.

To analyze the performance of bike-sharing systems we used as a dependent variable the number of rides recorded per week. This variable measures the level of usage of bike-sharing systems in each of the aforementioned Greek cities. In addition, we used a set of independent variables that can be associated with the number of rides. In many cases, the development of a model is highly affected and dependent on the attributes of the collected data. The model developed in the framework of the present research activity aimed to identify whether the lockdown period and the restriction measures applied affected the performance of the BS systems in three Greek cities which can be measured by how ridership was changed before, during, and after the lockdown. For this to be achieved, the number of rides, their duration, and whether the users returned the bicycles to the stations from which they rented them or not, based on the following assumptions, were used:


Firstly, to control the effect of quarantine lockdown, we include a dummy variable that takes the value of 1 during the quarantine period, i.e., the weeks between 23 March and 4 May 2020 and zero otherwise. Moreover, we include an alternative dummy variable that takes the value of 1 during the quarantine period and beyond, i.e., all the weeks after 23 March 2020 and zero otherwise. We do so, in order to control whether the effect of quarantine lockdown on the performance of bike-sharing systems remains after the easing of restrictions on 4 May. Secondly, we include the average duration of the rides per week in each city. Finally, taking into account the availability of bike-sharing systems, the number of rental stations in each city is also included.

Accordingly, the empirical model is formulated as follows:

$$Y\_{\rm if} = b\_0 + b\_1 \mathfrak{x}\_{1\rm it} + b\_2 \mathfrak{x}\_{2\rm it} + b\_3 \mathfrak{x}\_{3\rm it} + \mathfrak{u}\_{\rm if} \\
\text{with } \mathfrak{u}\_{\rm if} = \gamma\_{\rm i} + \varepsilon\_{\rm it}$$

where,


In random-effects models, the random effect is a component of the composite error term, i.e., *uit* = *γ<sup>i</sup>* + *εit*, and is not correlated with any regressor. Both random effect and errors are independently identically distributed, *<sup>γ</sup><sup>i</sup>* <sup>∼</sup> *IID* 0, *σ*<sup>2</sup> *γ* and *<sup>ε</sup>it* <sup>∼</sup> *IID* 0, *σ*<sup>2</sup> *ε* . Our random-effects model is estimated by ordinary least square (OLS) regressions with robust clustered standard errors in order to control for both heteroscedasticity and correlation of the error terms [36]. To decide between the use of fixed effects or random effects, we applied the standard Hausman test which showed that the appropriate specification is the random-effects model.

The main results are presented in Table 8. In the first column, we estimate a randomeffects model including only our main variable of interest, namely Lockdown (*x*1). Then, in columns (2) and (3), we estimate random-effects models including the remaining independent variables (*x*1, *x*2). In all three columns, the estimated coefficient of lockdown is positive and statistically significant, indicating a positive effect of quarantine lockdown on the number of rides. Specifically, the estimated coefficient suggests that lockdown leads to an increase in the number of rides via bike-sharing systems, during the the lockdown period. It must be noted that for these calculations, rides with a duration of less than 5 min were excluded.


**Table 8.** The effect of lockdown on number of rides, comparing lockdown period and non-lockdown period, excluding rides with duration less than 5 min.

Notes: All regressions include random effects and are estimated with robust clustered standard errors. Lockdown dummy takes the value of 1 during quarantine period, i.e., the weeks between 23 March and 4 May 2020 and zero otherwise. Standard errors in parentheses. \*, \*\* denote statistical significance at 10% and 1%.

Concerning the rest of the independent variables, the coefficient of the weekly average duration of rides is insignificant in all regressions, implying that the duration of rides does not affect the number of rides. Similarly, the rental stations variable exerts an insignificant coefficient, indicating that the number of rental stations is not associated with the weekly number of rides. In column (4), we estimate the baseline model using fixed effects instead of random effects in order to see whether the estimated results depend on model specification. As can be seen, estimated results for the main independent variable, i.e., Lockdown (*x*1)*,* hold across both random and fixed effects models.

In Table 9, we re-estimated all regressions of Table 7 using an alternative dummy for lockdown which measures the effect of lockdown during and after the lockdown period. Yet again, rides with a duration of less than 5 min were excluded by these calculations. As can be seen, the results remain the same since the estimated coefficient of Lockdown is still positive and statistically significant in all regressions. Hence, our results can have an alternative interpretation: Lockdown leads to an increase in the number of rides both during the lockdown period and after the easing of restrictions. Regarding the rest independent variables, estimated results for the average duration of rides remain unchanged. On the contrary, the number of rental stations exerts a positive and statistically significant coefficient, indicating that it is positively associated with the number of rides.


**Table 9.** The effect of lockdown on number of rides, comparing period before and after lockdown, excluding rides with duration less than 5 min.

Notes: All regressions include random effects and are estimated with robust clustered standard errors. Lockdown dummy corresponds to the alternative dummy which takes the value of *1* during quarantine period and beyond, i.e., all the weeks after the 23 March 2020 and zero otherwise. Standard errors in parentheses. \*, \*\* denote statistical significance at 10% and 1%.

#### **4. Discussion**

The statistical analysis of the data provided by the BS systems' operator revealed that the lockdown affected the modal choice of the residents of the three above-mentioned and analyzed Greek cities. It seems that the users of the BS systems in Igoumenitsa, Rhodes, and Chania acknowledged the benefits of using bike-sharing bicycles and as a result continued to use the bicycles even after the end of the lockdown period, although no campaigns were implemented or measures were taken as those in other European cities aiming to promote the usage of BS systems. Even though a decrease was recorded after the end of the restriction measures, the BS systems continued to have an increased level of usage six or nine months ago. The examination of qualitative and quantitative data for a longer time period, regarding the performance of the BS systems, could reveal if the users developed a new pattern concerning their habits and modal choice or if the effect of the lockdown period leading to an increased level of BS systems usage, fades out as time passes. Toward this direction, our team has addressed an official request to the BS systems operator in order to provide us with data covering a longer time period mainly after the end of the first lockdown period.

Similar results to our research were found and conclusions were extracted for the cities of New York [37,38], Budapest [39], Thessaloniki [40,41], Nanjing [42], and the region of Sicily in Italy [43], regarding the level of ridership before, during, and after the imposition of restricting measures concerning urban mobility and specifically the usage of bicycles, the rides' duration [44], the people's perception towards BS systems due to the COVID-19 pandemic, and to bicycles in general [45,46]. Furthermore, during the early stages of the pandemic, in some cases, bike sharing expanded their memberships or improved accessibility for specific working groups [47]. However, there are several cases in which despite an initial increase in using BS systems during the first months of the pandemic, ridership declined for several months in 2020 [48–53]. There are also systems recovering from the effects of the pandemic and mainly lockdown measures applied in many countries; for example, in the UK, the latest data show a significant recovery [54]. A study estimating the effect of the pandemic on the London BS system over the period March–December 2020 indicated that although a reduction in cycle hires was recorded in March and April 2020, the demand increased after May 2020 and even more during April, May, and June 2020 the bikes were hired for longer trips (perhaps as the authors mention due to a shift from public transit) [54]. It is clear that people have reacted differently worldwide concerning the usage of BS systems during the pandemic, making each case unique. For example, in Seoul, the level of usage of the BS system was affected negatively by the number of daily new COVID-19 cases and positively as the necessity for social distancing was becoming a reality for the residents [34]. Moreover, a study regarding the impact of

COVID-19 on the BS system in Slovakia, showed that during the lockdown period, the level of usage was decreased. However, after restrictions were relieved, a slight increase was recorded [55]. In a recently published study regarding the performance of BS systems in the United States [56], one of the main conclusions (five out of eleven BS systems provided feedback) was that bike-sharing moderate-frequency riders (1–2 times per month) may increase after the coronavirus restrictions are lifted.

Based on the above-described analysis, we can state with relative safety that the lockdown period affected the performance of the BS systems in the examined cities since during the lockdown period the level of usage of BS systems increased. This evolution can be capitalized by the local authorities by promoting and advertising the benefits of the BS systems, even those that were not initially (and most likely could not be) conceived, but ultimately were revealed during the COVID-19 pandemic [57–59]. Specially designed campaigns could be organized by the local authorities in order to inform the people of the benefits of using bicycles, by highlighting that BS systems' level of usage was increased during the lockdown period providing them a way out from the restriction measures while at the same time maintaining social distancing. Lastly, those campaigns could promote the usage of BS systems as a significant tool to achieve sustainable mobility [56].

Our study has some limitations. The data provided cover a time period of many months before the applied first lockdown in Greece and only a few months after the easiness of the applied measures. Although the change in the level of BS bicycles' usage was recorded, analyzed, and examined, a longer time period would provide more solid conclusions and results. Furthermore, our analysis is based on quantitative data for the BS systems and at the same time, we do not have access to data concerning other available transport modes in the examined cities. As a result, no comparison among the available transport modes can be made as, for example, public transport and BS bicycles. Finally, the absence of qualitative data regarding the reasons for users' choice to use more (or at least at the same level) BS bicycles during and after the lockdown period compared to the pre-lockdown period, does not allow us to perform an in-depth analysis.

As the pandemic continues, it has been proposed that BS systems continue to be constantly monitored by the operators and despite their level of usage, the users' choices (either these lead to an increase or decrease in their usage) should be investigated through rolling questionnaire-based surveys. As health protocols demand social distancing, it is difficult to perform in situ surveys and therefore it is recommended to exploit digital services such as the internet and smartphones in order to implement these surveys. Furthermore, it is crucial for the sustainability of the BS systems that operators fully comply with the health protocols applied which require decontamination of bicycles and all surfaces in the rental stations ensuring that BS bicycles do not become contagion sources.

If the values presented in Table 10 and specifically the number of rides per 1000 residents are expressed on a daily basis (30 days per month), then for the case of Igoumenitsa, the respective values are equal to 1.88 for March 2020, 2.48 for April 2020, and 3.65 for May 2020. For the case of Chania, the respective values are equal to 0.03 for March 2020, 0.17 for April 2020, and 0.20 for May 2020 while for the case of Rhodes are equal to 0.03 for March 2020, 0.55 for April 2020, and 1.07 for May 2020. When comparing the highest values of the number of rides per 1000 residents (May 2020) with the respective values for other BS systems, Igoumenitsa outperformed the BS system of Mexico City, Chania's BS system performed similarly to Seattle's BS system, and Rhodes' BS system performed similarly to London's BS system based on the findings of a report published in 2018 [60].


**Table 10.** Effect of lockdown on number of rides per 1000 residents in the cases of Igoumenitsa, Chania, and Rhodes for the period March 2020–May 2020.

#### **5. Conclusions**

Based on the presented data, during the first lockdown period in Greece, the residents of the above-mentioned Greek cities, for unknown (up to present) but assumable reasons, decided to use the installed and operational BS systems in their home cities. Although the increase had started a short time period before the lockdown period, it came to a peak during the lockdown. After the end of the lockdown, the residents tended to return to their prior moving patterns with an increased share for the BS systems compared to the pre-lockdown period. The average duration of the BS bicycles usage during the lockdown period was increased compared to the months before it (Igoumenitsa, Chania) and the months after it (Igoumenitsa, Chania, Rhodes) in 2020, despite the fact that during the lockdown the residents were allowed to move only for a period of one hour during daytime if permission was granted to them and for a small number of specific reasons.

It is necessary to better understand the reasons for the (recorded and presented in this paper) performance of the BS systems in the three Greek cities before, during, and right after the first lockdown period in Greece, to perform a questionnaire-based survey addressed to the users asking them for the reasons that led them to use the BS systems during the lockdown. Furthermore, a request has been planned for submission to the operator of the examined BS systems to provide our team with more data covering mainly the time period after the lockdown period, which will allow us to compare the evolution of the BS systems' performance for a longer time period.

An interesting finding of the above-presented analysis concerns the evolution of the total number of rides after the lockdown. While in the city of Igoumenitsa the total number decreased (but yet remained higher with the respective numbers before the implementation of the restriction measures during the lockdown), in the cities of Rhodes and Chania the total number of rides kept rising. However, based on Table 10, it is clear that the average daily rides in Igoumenitsa in May 2020, which was the most significant among the three cities, were about 30. Considering there is a population of over 9000 in Igoumenitsa, this amounted to about 0.03% of people who opted to use the bikes in the BS systems in the city. While the effect was statistically significant, in reality, the effect was considered very minor.

The airport at Chania served flights from 1 July 2020 and beyond due to the restriction measures applied worldwide [64]; therefore, the increase was not caused by visitors. In the city of Rhodes, the decrease in arrived flights at the island's airport compared to those in 2019 for the period April 2020–June 2020 was 100% [65]. Moreover, in this case, the increase in the BS system was not caused by visitors. In order to understand the reasons for this phenomenon, as mentioned above, a questionnaire-based survey addressed to the users of the BS systems operating in the currently studied cities is under development.

The current situation (COVID-19 pandemic and the measures applied worldwide) is unique for modern human history in terms of the geographical area and the duration of the applied restricting measures/lockdown. Estimations and assumptions can be made in order to explain the phenomenon and identify the effects (if identified) of the lockdown on the users' modal choice and modal shift. However, during the first lockdown period in Greece, the residents of three Greek cities in which BS systems were installed and are currently operating decided to use them more than the previous time period. This fact is indisputable; however, the reasons that led to this evolution can be only assumed based on the availability of qualitative data for that time period.

The first and more solid assumption is that the increase in the demand for the BS systems was a countermeasure to the restriction measures implemented due to the COVID-19 pandemic as people chose BS bicycles as a safe transport mode ensuring social distancing, freedom of their movements, and environmental benefits. Specifically, the possible benefits identified by the usage of BS bicycles should be capitalized by local decision-makers based on the fact that sustainable mobility has been, is, and will keep being, one of the most important objectives of the European Commission [66–69]. Moreover, there is a high possibility that users that did not own a bicycle and therefore could not exercise (one of the reasons that the residents could go outside for a period of one hour was to exercise), chose the bike-sharing systems as an excuse to go outside and exercise in an effort to decrease the pressure (mainly psychological) created by the lockdown (such a choice was given by the Greek government using a specially developed mobile service) or a transport mode in order to move across those cities covering and serving their needs.

**Author Contributions:** E.B. Conceptualization, methodology, validation, investigation, and data curation, writing—original draft preparation, writing—review and editing; S.B. validation, writing—review and editing; S.F. formal analysis, writing—original draft preparation; E.P. validation, writing—review and editing; H.S. validation, investigation and data curation, writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank the authorities of the Regional Unit of Thesprotia for providing the data, as well as the permission for using them, related to the SUMPORT Project, the authorities of the municipalities of Rhodes and Chania for the permission of using the bike-sharing systems' data and finally Cyclopolis for providing the necessary data regarding the performance of the bike-sharing systems in the Greek cities of Igoumenitsa, Rhodes, and Chania.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article*
