Ubiquitous city logistics an illusion, or can we create model(s) to measure and reduce congestion?

(1)

Ubiquitous city logistics an illusion, or can

we create model(s) to measure and

reduce congestion?

August 12, 2020 Word count: 13.537

Student:

Anouk de Vries

S3612961

a.m.de.vries.11@student.rug.nl

First supervisor:

dr. ir. P. Buijs

Second supervisor:

dr. ir. S. Fazi

University of Groningen

Faculty Economics and Business

(2)

2

Abstract

The e-commerce market is growing, leading to more and more vehicles in the city centre, crowded cities and pressure on the city logistics. Resulting in uncertain and increasing travel times in the city centre. Those disadvantages of the growing e-commerce market are also known as congestion. To prevent cities from being totally blocked and exploding travel times, action needs to be taken. Package delivery companies are willing to invest in innovative vehicles to reduce congestion. However, before investing a lot of money in new vehicles, managers want to know how much congestion is reduced by the vehicle. To measure this congestion impact, congestion measurements are needed.

Existing literature about congestion only focus on congestion levels for a specific region or specific road segment. Measurements to measure the congestion caused by a specific vehicle on the street are missing. Without those measurements it is impossible for managers to prove that the new vehicles will reduce congestion. Therefore, this thesis developed multiple possible congestion measurements into a congestion metric specific for delivery vehicles. Resulting into more and better insights in congestion impacts of delivery vehicles in city centres.

(3)

3

Preface

(6)

6

1. Introduction

People’s habits and demands in ordering goods are changing, including ordering online for delivery at home. This leads to more small orders every day and same day deliveries or within less than two hours deliveries (Fuldaur, 2019). As a consequence, parcel delivery becomes a painful topic, especially in city centres, which have become silted (ten Kate & Heeger, 2019). Predictions suggest that without intervention the problem will only get worse. The demand for urban last-mile delivery will grow with 78% by 2030 and without intervention this will lead to 36% more delivery vehicles in the top 100 cities worldwide (Hillyer, 2020). Looking at the Netherlands, without intervention, the number of delivery vehicles in the streets will increase from 40 till 50 to 70 till 80 per day, leading to more nuisance such as wrong parking, which triggers congestion (Transport Online, 2018). When no actions will be taken, each road user will face an additional eleven minutes of travel time per day if congestion will rise by over 21% (Deloison, et al., 2020). The biggest challenges in reducing or controlling the congestion in city centres, are reduction of the traffic movements and improvement in utilization of the available space in the city centres (Shopping Tomorrow, 2019).

Parcel delivery companies are actively working towards reducing or controlling the congestion in city centres and are willing to invest in initiatives to reduce congestion. These companies are already investing in multiple innovations, such as the adoption of light electric vehicles (LEV) and electric (cargo)bikes. Compared to the vans that are currently used, these innovative vehicles are smaller, which seems advantageous for congestions reduction (de Weerd, 2018). It is, however, not possible to empirically validate the congestion reduction caused by using these innovations because of the lack of suitable congestion measurements. Besides that, investing in LEV costs a lot of money for parcel delivery companies. The investing costs of a LEV is about 40 percent higher than the traditional delivery vans (Groen, et al., 2019). Without such measurements, how could managers convince their companies to invest such an amount of money in those innovations?

(7)

7

region or road segment. Examples of capacity measurements are the Volume-Over-Capacity-ratio, measuring how successful a road section can handle the assigned amount of road users (Çolak, Lima, & González, 2016) and the length of the road multiplied with number of lanes of the road (Mohan Rao & Ramachandra Rao, 2012). Examples of speed measurements, are calculating the delay time of an additional road user (Vickrey, 1969) and combining speed levels with the queuing theory (Kaddoura & Nagel, 2018). Unfortunately, those measurements are only applicable to simple traffic structures, such as highway roads which exist of long road segments with the same maximum speed level and number of lanes. City centres, however, are characterised by more difficult traffic structures While driving through the city centre, delivery vehicle may face many situations in which it is not possible to drive the maximum allowed speed, because of cyclists riding on the same road, pedestrian crossings, crossroads and road narrows. In those situations, the capacity and speed measurements are not suitable to use because they are not directly linked to congestion.

So, while academic literature already proposes several congestion measurements, they are not applicable to measuring the congestion impact of a specific delivery vehicle, and hence cannot be used to indicate the congestion impact of a specific delivery vehicle present in a certain street or area. Filling the gap of those missing congestion measurements specific for parcel delivery companies is crucial for the future. The lack of congestion measurements may therefore namely from a barrier for parcel delivery companies to invest in delivery modes that cause less congestion. To stimulate companies to invest in vehicles that cause lower congestion, measurements have to be developed. This leads to the following research question:

“Which factors contribute to the impact of congestion caused by parcel delivery

vehicles in inner cities and how can their impact be measured?”

(8)

8

2. Theoretical background

2.1 E-commerce trends and city logistics

In general, the main task of city logistics is the coordination of the delivery of freight from many different shipper-carrier-consignee commercial relations (Benjelloun & Crainic, 2008). Next to that, city logistics is about finding a balance between benefits and disadvantages for the city. Freight moving in and out of the city can be seen as a benefit. However, these movements cause environmental, social and economic nuisance, which are disadvantages. Therefore, city logistics is focused, on the one hand, on efficiency and fulfilment of demands of urban customers and businesses. On the other hand, city logistics is focused on reduction of negative externalities such as congestion and emissions (Rodrigue & Dablanc, 2020). Finding the balance between those sides is complicated because of the different trends and challenges playing a role in the busyness and scarcity of available space in the inner city (Shopping Tomorrow, 2019):

• Not enough space available to facilitate the traffic and not enough loading- and unloading sections;

• Vehicle restrictions from the municipality lead to more movements of smaller trucks inside the city centre because big, heavy trucks are banned;

• Accessibility of city centres for motorised transport decreases because of reduction of the maximum speed, limitation of the available parking spaces, stricter enforcement on wrong parking, and prohibition of motorised transport at certain locations between certain timeslots;

• As nowadays everything can be ordered online, the volume to be delivered to private customers is increasing. This leads to more disruption and unsafety on the roads in city centres.

(9)

9

Several studies already showed many advantages of the LEV. Van Amstel, et al. (2018) concluded that LEV could replace ten till fifteen percent of total delivery vehicle movements. The reduction of delivery vehicle movements seems to be a good opportunity for LEV to reduce congestion as less movements lead to less time spending on the road. Besides that, Van Amstel, et al. (2019) concluded that the LEV can also contribute to more happiness. They stated that the drivers of the LEV receives postive reactions of both customers and general public on the streets, compared to more complaining reactions to drivers of larger delivery vehicles during the unloading or delivery process. Both results seem to have a good impact on congestion creation by delivery vehicles. However, it cannot directly be assumed, based on those insights, that the LEV reduces congestion in reality. To empirically validate congestion reduction, congestion measurements for measuring the congestion created by delivery vehicles are needed.

2.2 Congestion

The innovations mentioned above are meant to reduce congestion. Though, congestion is an ambiguous concept which makes it difficult to define and prove congestion reductions. Over the years, various definitions have been created to get a description of congestion. In the early years, Vickrey (1969) defined different types of congestion: ‘single and multiple interaction’, situations where drivers cannot reach the allowed maximum speed because of other road traffic participants, ‘pure bottleneck situation and triggerneck situation’ wherein the bottleneck of the road is responsible for the congestion at the surrounding roads and lastly ‘network and control congestion’ which is the unavoidable congestion caused by traffic measurements such as traffic lights.

Recently, Ferrara et al. (2018) further specified network and control congestion into the congestion phenomena by distinguishing recurrent and non-recurrent congestion. The main difference between those two types of congestion is the predictability factor. Recurrent congestion is predictable because it occurs on a daily basis when traffic demand exceeds traffic supply (capacity). Think of traffic jams during peak hours. On the contrary, non-recurrent congestion is not predictable and related to unexpected events, such as traffic accidents, unexpected weather conditions, fluctuations in traffic demand (during holidays), roadworks and special events (festivals).

(10)

10

reliable roads. The reliability of the road is an important aspect because road users have less tolerance for unexpected congestion than expected congestion. As a result of that, road users are more likely to choose a reliable road instead of a road with unreliable travel times.

Another perspective of determining congestion is by using micro level factors and macro level factors (Mohan Rao & Ramachandra Rao, 2012). Micro level factors are related to the road users of the road itself and triggers the congestion. Macro level factors stimulate congestion and are related to the more general demand and use of the road.

All in all, there are a lot of perspectives and factors related to congestion. Though, because of the complexity of city logistics it is hard to assign one of these factors or definitions to delivery vehicles. However, slowly more and more attention is paid to congestion specifically in urban surroundings. Research of Atomode (2013) gives insights in traffic delay related to vehicle categories in urban areas. This study shows that delivery vehicles are ranked fifth with an average contribution of 8 minutes in traffic delay time, equal to 9% of the total traffic delay time in urban areas. Besides the type of vehicle categories, also the traffic circumstances and actions are ranked. Multiple actions in that list can be linked to delivery vehicles. Parking problems are ranked second, accounting for an average of 21 minutes of the total delay time, equal to 23.3% of the total delay time un urban areas. Turning and manoeuvring problems are ranked third, responsible for 8 minutes (9.5%) of the total delay time in urban areas.

Besides the traffic delay times, also parking situations got more attention, especially on-street parking. On-on-street parking namely reduces the capacity of the road in two ways. Firstly, it narrows the free space for traffic in the street. Vehicles will then be forced to perform their actions in a narrowed space and have to limit their speed. Secondly, on-street parking and unparking manoeuvres create complex situations, leading to congestion on urban roads (Biswas, Chandra, & Ghosh, 2017).

2.3 Congestion measurements

(11)

11

Besides measuring the congestion levels for a specific road section, speed measurements can also be used to indicate the effects of additional road users on the road to other road users. Vickrey (1969), for example, concluded based on speed levels that an increase of one road user, leads to an increase of ‘k+1’ times the delay of the road user itself. In other words, every minute of delay caused by the new road user, will also add k minutes of delay to the other road users. Another way to indicate the congestion level caused by specific road users is by combining transport simulation software data with the queuing theory. Based on their length and speed, vehicles absorb a part of the total flow capacity of a road. Exceedance of the total flow capacity will lead to congestion. Therewithal the amount of exceedance will determine the amount of congestion for the arriving road users (Kaddoura & Nagel, 2018).

Besides speed levels, also capacity measurements are used to get an indication of the level of congestion. Çolak, Lima, & González (2016) created the Volume-Over-Capacity ratio (VOC-ratio). This ratio measures how successful a road section can handle the assigned amount of road users. High VOC-ratio values indicates more congestion. Another, more general method to calculate the capacity of the road is by multiplying the total kilometres with the amount of lanes of the road (Mohan Rao & Ramachandra Rao, 2012). The closer the amount of vehicles on the road approaches the maximum capacity, the more congestion will arise.

At last, a specific way to estimate the congestion level is by using the ‘Stagnationfactor’. In the Netherlands, the Ministry of Traffic and Water Management uses a stagnation rate or factor to define the expected congestion levels for a specific road section. Those expected congestion levels are predicted by the speed category of the road and stagnation during peak hours (Rijkswaterstaat, 2007).

All in all, there already exist multiple options to measure congestion. Unfortunately, those are not applicable to the delivery vehicles driving through urban road networks, because the creation of congestion by a delivery driver/vehicle has not been taken in account. For example the speed measurements, within city centres, cars may have to lower their speed for crossing pedestrians or a group of cyclists, but this congestion is not caused by a delivery vehicle at that moment.

2.4 Urban congestion measurements

(12)

12

of traffic situations in urban areas and traffic delays caused by car accidents. These measurements are not sufficient to develop a total congestion metric for delivery vehicles. Although they can be used as a good starting point.

Firstly, recent literature focuses more specific on congestion measurements in urban areas by measuring the loss in freedom of movement by using the speed-flow envelope (Kumar Sharma & Swami, 2016). This speed-flow envelope, created by Maitra, Sikdar, & Dhingra (1999), visualizes the relationship between speed and volume on the road. A higher volume will lead to a greater reduction in the speed level. Speed reductions of a delivery vehicle occurs more often than passenger vehicles, as delivery vehicles have to stop often to deliver a package. The speed flow envelope states that every speed reduction means that the vehicle is facing a congested situation. However, this does not hold for delivery vehicles, as stopping in an empty street to deliver a package does not mean a congested situation. Therefore, the speed-flow envelope is not directly applicable to delivery vehicles.

(13)

13

Figure 2.1 Example of outgoing entropy values

The FBPR score and outgoing entropy values combined, indicate the congested street segments. Congested street segments are indicated by the combination of the highest five and ten percent of the FBPR scores, so the top five and ten percent of the traffic demand, and the highest five and ten percent of outgoing entropy values, so the top five and ten percent of the flow complexity.

Lastly, traffic delay caused by traffic accidents can be determined by using the equivalent property damage only (EPDO) formula. This formula is based on the severity of the accidents, time period of the day and length of the street segment. An advantage of this formula is that the formula takes into account the specific time period on the day in order to calculate the effect of the severity of accidents. The higher the EPDO formula, the higher is the traffic delay (Pulugurtha, M.ASCE, & Pasupuleti, 2010).

To get useful insights from the EPDO formula, the categories of traffic accidents need to be identified. In the Netherlands, there are five different categories: ‘light damage (UMS-)’, ‘more than light damage (UMS+)’, ‘light injury’, ‘more than light injury’, and ‘serious bodily injury’ (Overheid.nl, 2015). An example of the EPDO formula, using the categories of the Netherlands is:

𝑤1 × 𝐿𝐷 + 𝑤2 × 𝑀𝑇𝐿𝐷 + 𝑤3 × 𝐿𝐼 + 𝑤4 × 𝑀𝑇𝐿𝐼 + 𝑤5 × 𝑆𝐵𝐼 𝑙𝑒𝑛𝑔𝑡ℎ × 365

(14)

14

(15)

15

3. Methodology

Currently, managersof parcel delivery companies face the challenge that they want to reduce congestioncaused by their delivery vehicles, but they do not know how to measure this specific type of congestion. Therefore, this thesis adopts a design science approach to develop and test novel solutions to measure the created congestion by delivery vehicles. Design science emphasizes the process of exploration through design and seeks to explain the process of exploring new solution alternatives to solve problems and to improve the problem-solving process (Simon, 1973). On top of that, design science is a good method to solve so called ‘ill-structured’ management problems, the problem is known but it is not known how to solve it (Homström, Ketokivi, & Hameri, 2009).

The design science method consists of three cycles (Hevner, 2007). The first cycle, the relevance cycle, initiates the design science process with the actual problem and requirements of the application environment. In this thesis, this cycle consists of a design sprint conducted to fully understand the problem and requirements of a parcel delivery company and its relevant stakeholders. The second cycle, the rigor cycle, verifies the past knowledge to make sure innovative artifacts are developed during the project. This thesis builds on past knowledge on congestion measurements which are already developed metrics suitable for highways. Lastly, the design science process includes the design cycle, which carefully balances the efforts of constructing and evaluating. The design process of this thesis consists of a morphological analysis to get an overview of all the possible measurement artifacts, keeping in mind the relevance and rigor cycles.

3.1 Relevance cycle

(16)

16

sprint were presented to a broad audience consisting of traffic experts, traffic employees from different municipalities, teachers of both universities and universities of applied sciences. An in-depth explanation of the design sprint process can be found in appendix A.

Generally, data collection for the design sprint is done via semi-structured interviews on the street, a focus group and via observations during a ride along with some drivers of a delivery vehicle. The semi-structured interviews are conducted with traffic participants on some of the busiest streets in Amsterdam. Questions related to their online shopping and delivery experiences and busyness in the city were asked. Besides that, there was room for the interviewee to tell his/her own story related to the topic. The focus group was conducted with inhabitants of the city centre of Amsterdam. Lastly, riding along with some drivers is done within the delivery company PostNL. Different routes were driven through the city centre of Amsterdam, by different drivers with respect to experience and age. During the ride and the stops for delivering packages, an observation study is performed according to an observation form (see appendix B).

3.2 Rigor cycle

In order to define urban congestion, firstly academic literature about existing congestion definitions is examined to get both an overview and an idea about existing thoughts and aspects of congestion. Next to that, the interviews and rides along are used to discover the most congestion causing aspects or actions belonging to parcel delivery vehicles. The combination of both academic literature and insights of the interviews and rides along, leads to a new definition of urban congestion specifically focused on parcel delivery vehicles.

(17)

17

infrastructure and water management and searching online, resulted in several available data sources which can be used to provide data suitable as benchmark data.

After identifying all the suitable data sources, the creation of the measurements in more detail, or, in other words the exact formulas, starts. When creating the formulas, academic literature is used to identify and get used to current measurements of traffic busyness and congested situations. After, both collecting all the knowledge about current measurements, the suitable data sources and identifying the congestion causing moments around parcel delivery vehicles, the formulas to measure congestion related to delivery vehicles in the city centre are created, see design cycle, based on this acquired knowledge.

3.3 Design cycle

The design of the congestion metric was informed by a morphological analysis. Accordingly, to create the metric, several possible ways to capture the input required for the metric was done by collecting all the listed factors of the interviews and the ride along and decide, with the group of PostNL employees, which factors are the most important ones and able to use. The morphological analysis is used because it allows to investigate many possible relationships and configurations in a certain problem context (Ritchey, 2013).

In performing the morphological analysis, the following steps are taken, in line with the guidelines proposed by Ritchey (2013). The analysis starts with identifying and defining all relevant parameters or dimensions of the problem. Within this thesis those are the congestion factors related to delivery vehicles in city centres. After defining the parameters or dimensions the relevant conditions or values of the parameters will be defined.

To identify the dimension of the morphological analysis, the results of the design sprint are used as input data. The issues faced during the interviews, the ride along and discussions with the PostNL team already lead to some prototypes with several dimensions. Those prototypes are investigated and compared with each other and as a result six factors were appointed that caused congestion. After creating the first concept of all factors, again a session with all participants of the Design Sprint is used to discuss and evaluate this first concept.

(18)

18

(19)

19

4. Results

The results of this thesis are divided into three parts. The first part presents the new definition of congestion specific to urban areas and delivery vehicles. Compared to previous congestion definitions, this definition also takes into account all factors of urban areas and actions of delivery vehicles, which do not play a role in congestion on highways. The second part describes the process of creating the measurements supported by the description and measurement options of the measurement factors used for congestion caused by delivery vehicles (detailed information about the formulas can be found in appendix C). Thirdly, the results of the research to suitable and available data sources are described. Lastly, the measurement options are summarized in a morphological report leading to a ‘starting option’ and ‘future option’ for measuring the congestion impact of delivery vehicles.

4.1 Definition of urban congestion

Parcel delivery companies are already experimenting a lot with new vehicles to reduce the congestion caused by delivery vehicles during the delivery process. However, due to the lack of a clear-cut definition of urban congestion, it is difficult to focus these efforts across different initiatives and companies.

A considerably part of this thesis project was focused on developing an empirically based definition for urban congestion. This definition takes into account both multiple perspectives of users in the urban area and the perspective of the driver to perform his job.

(20)

20

Combining the results of the interviews and the ride along, quickly a first thought of the definition of urban congestion arises, namely: “Urban congestion is crossing the way of other people with your delivery vehicle”, for example by turning left on a crossroad or blocking the street with the delivery vehicle for other traffic participants. Later this definition changed as crossing seemed a bit too specific. The term crossing was often understood as exactly crossing the road of the other traffic participant by performing for example a turning action. Therefore, the definition changed to “Urban congestion is nuisance by interrupting the road for other road users with your delivery vehicle”. This definition is totally focused on causing nuisance to other road participants by vehicle actions specifically related to congestion. Other potential negative externalities, such as environmental and noise impacts are not taken into account in this research.

4.2 Congestion measurements

After creating the definition, the focus shifted to the creation of the measurement artifacts. The measurements are divided into five different factors, focused on the vehicle itself, playing a relatively huge role in causing congestion by parcel delivery vehicles and the happiness factor to measure the happiness of the other road traffic participants.

4.2.1 Possibility to pass delivery vehicle

(21)

21

cars can still not pass the delivery vehicle because the additional free space left to the vehicle is too small for a passenger car.

To measure the congestion impact, the route and both data about the width of the streets and width of the delivery vehicles, available via the license plate number, are needed. Sizes of the street can be collected via municipality data, or if no data available via the municipality, the free data source PDOK Viewer (PDOK, 2020) can be used. Based on this data, there are three options to measure the possibility to pass a delivery vehicle. Firstly, the number of suitable roads that are used can be checked. Secondly, the amount of correct driven kilometres can be measured. Lastly, time spent on correct parts of the roads can be used as measurement.

Option 1.1: Number of suitable roads used

As stated above a street is marked suitable when the street fulfils the road suitability conditions. The more suitable roads are taken by the driver, the less heavy congested situations will occur because roads are less often blocked by delivery vehicles and other road participants can pass the delivery vehicle. To indicate the suitability of the street, every smallest part of the street width per street taken in the route is needed. This can be collected from the municipality data or by using the measure instruments from the PDOK viewer to measure the street width (see figure D.1 in appendix D). Based on the smallest street width per street and the width of the vehicle from the licence plate, every street can be marked as suitable or non-suitable. To compare routes and delivery vehicles with each other, the percentage of suitable roads used during the route is calculated by dividing the number of suitable roads by the total number of roads.

Option 1.2: Correct driven kilometres

(22)

22

which streets are used. Combining those data sets, the amount of suitable meters can be calculated and be compared with the total route length. The more suitable meters are chosen, the relatively less congestion will be caused.

Option 1.3: Correct driven time

Congestion also depends on the time spent on those non-suitable street segments. The relatively more time is spent on non-suitable street segments, the higher the congestion will be. Again the lengths of the non-suitable and suitable street segments are used to indicate the correct driven meters. The GPS device in the delivery vehicle also collects the time being present at the specific locations based on the GPS coordinates and the total time spent on the total delivery route. With this data, driven time on suitable road segments can be compared to the total driven time during the delivery round.

4.2.2 Use of crossroads

Congestion is mainly associated with traffic jams and delays on highway roads. As urban congestion is less well known, logically, crossroads are almost never taken into account when thinking about congestion solutions. However, surprisingly, literature already showed us that congestion can be caused in different ways on crossroads. Which emphasizes the importance of taken into account crossroads, leading to three measurement options for measuring congestion related to the use of crossroads. Firstly, accidents are a determinant for the safety and traffic delays on crossroads. Secondly, the amount of traffic flows determines the level of congestion. Lastly, congestion on crossroads will be caused by turning movements in which other traffic flows are crossed.

Measuring the congestion caused by delivery vehicles can be done in several ways. The first option is to measure the safety of the crossroad by using the EPDO score. Second option is calculating the complexity of a crossroad. This can be measured with the Flow-Based Page Rank (FBPR) and outgoing entropy value (OEV) formulas. Lastly, congestion caused at crossroads can be measured by the turning actions on the crossroad. Especially when the main traffic flow of the crossroad is be hindered, this could cause congestion.

Option 2.1: Crossroad safety measure

(23)

23

number of accidents can be collected via the municipality. If not available via the municipality, the more time-consuming method, using the PDOK data source is needed. By selecting the options accidents based on RDW and road segments, the number and type of accidents per road segments are visible in the map. Using the measurement instruments of the PDOK software, the belonging length of the road segment can be measured.

The EPDO score gives an indication of the average traffic delay on that road segment in case of an accident, which is one of the determinants of congestion. The more severe accidents on a road segment the higher the EPDO score. The other way around, the smaller the road segments, ceteris paribus, the higher the EPDO score. The higher the EPDO score, the higher the traffic delay time will be on that specific road segment and thus the bigger the chance of congested situations. Crossroads within the top ten percent of the EPDO scores are marked as non-preferable. To measure the congestion, the percentage of relatively safe crossroads used is calculated based on all crossroads used during the delivery round.

Option 2.2: Crossroad complexity measure

A relatively new insight for package delivery compagnies is that the complexity of a crossroad is a good indicator for the congestion sensitivity of the crossroad. A more complex crossroad leads to a higher chance of congestion. The complexity of the crossroad is determined by the FBPR and OEV scores. The formulas can be found in appendix B. The more traffic demand at a crossroad, the higher the chance of traffic jams because of exceeding the capacity and the higher the FBPR score. The more possible outgoing movements on a crossroad, the higher the OEV score. Crossroads with both a top ten percent score on FBPR and a top ten percent score on OEV are indicated as complex crossroads with a relatively high chance of congestion. The amount of high chance of congestion crossroads is counted and divided by the total amount of crossroads used in order to measure the congestion caused by the delivery vehicle.

Option 2.3: Crossroad movement measure

(24)

24

account the turning actions during the delivery route to identify the hinder to other traffic participants is thus a complete new topic for parcel delivery companies.

Thus logically, crossing other traffic flows lead to congested situations as road users around the delivery vehicle have to wait before they can continue their way. However, crossing other traffic flows cannot be totally avoided at crossroads, but the number of situations in which other traffic participants are hindered can be minimalised. Within this measurement the congestion is calculated by counting the number of actions on a crossroad in which the main traffic flow is crossed as crossing the main traffic flow is less preferable than crossing a minor traffic flow.

4.2.3 Time of presence

Different parts in the city are busy at different times of the day. In the mornings and evenings, many vehicles and persons are on the roads and crossroads nearby the highway roads around the city. School areas will be crowded during the start and end time of the schooldays. And around sport clubs or sport facilities, it is busy during the training times and on matchdays. As package delivery companies expanding their delivery time windows more and more, they are also more and more present at busy time periods. Interestingly, those conflicting situations are not taken into account yet when creating the routes. However, the busier the traffic situation or the situation around the traffic roads, the greater the chance of roads becoming full and blocked. Besides that, there is a greater change of accidents. That is why this factor focuses on suitable and non-suitable time periods for being present at a particular location with a delivery vehicle. A street is classified as busy if at least 70 percent of the capacity of the road is already used, based on the intensity-capacity ratio (Rijkswaterstaat, 2007). Therefore, it is preferable to be present in streets in which less than 70 percent of the capacity used. Those are classified as non-busy.

(25)

25

Option 3.1: Number of busy periods used

Every time a delivery vehicle is present in a street during the busy time period instead of the non-busy period, it is likely that the delivery vehicle causes congestion. Therefore, the more streets are used during busy time periods, the more congestion is caused. The amount of congestion caused can be measured by the percentage of streets used during the busy time period. Based on the data available via the municipality or via, for example, TomTom Traffic API, the busy time periods per street can be measured, using the intensity-capacity ratio of Rijkswaterstaat (2007). The GPS device of the delivery vehicle gathers the data during the route, based on the GPS coordinates and the corresponding time of presence, it can be determined when and where the vehicle was present. If the time frame of presence is (partly) equal to the time frame of the busy periods in that specific street. The presence in that street is counted as being present at a busy period.

Option 3.2: Time spent in busy periods

Sometimes a street is used during busy time periods, while the street was entered during a non-busy time period. In this case, only the last minutes spent in the street may have caused congestion. To make this difference visible, the total time spent during non-busy time periods and total time spent during busy time periods are important to know. The more time of the total delivery route has been spent during busy time periods, the more congestion may have been caused. The amount of congestion caused thus depends on the percentage of time spent in streets during busy time periods compared to the total time of the delivery route. Again the timeframe spent in the street is compared to the busy time frame(s) of the street. This time not only the number of same time periods are counted, but the specific amount of minutes during the busy time period are counted. This is possible as the GPS device also tracks the time being present at specific GPS coordinates.

4.2.4 Use of parking places

(26)

26

To measure the use of parking places, data about the number and sizes of the parking places along the route are needed. Many municipalities have a good overview of the number of parking places in their area. If cooperation with the municipality is not possible, the open data sources of the RDW can be used. The RDW has many different parking data sources, like static and dynamic data, number and specifications of the parking places. In combination with the sizes of the chosen vehicle it can be determined how many suitable parking places are available for the delivery vehicle per route. Logically, a parking place is suitable for a delivery vehicle when the sizes of the parking place are bigger than the delivery vehicle itself. The GPS-data from the delivery vehicle gives information about the stop location and duration of the stop. If there are doubts about the exact stop location, information about the usage of flashing lights can be used as an additional tool. When the driver uses flashing lights during the stop it is assumed that he/she has parked on the street and not at a parking place. The use of parking places can be measured and expressed in either the availability of suitable parking places, or the number of parking place stop actions, or the time spent on parking places.

Option 4.1: Parking place availability

Logically, without suitable parking places it is impossible to use parking places as stop location. The delivery vehicle has to be parked at another location at which, as expected, it is likely to cause more congestion than at a parking place. So, the more suitable parking places are available for the delivery vehicle, the less likely it is that the delivery vehicle causes congestion while packages are delivered. It is interestingly to see, that many parcel delivery companies do not take into account parking places in choosing the vehicle for the delivery process, but only the amount of packages. However, this measurement will not be the most accurate measurement, it will be a good starting point to get a first indication of the congestion impact of the vehicle. How much congestion may be caused, thus depends on the percentage of available suitable parking places compared to the total available parking places along the route.

Option 4.2: Number of parking places used

(27)

27

places along the route. From the GPS device in the vehicle it can be measured where the vehicle was parked and thus the suitable parking places used can be counted. Based on the parking place data in the region from the municipality or RDW, the total number of available and suitable parking places can be identified along the route.

Option 4.3: Time spent on suitable parking places

It may sounds strange, but some people will now think why not parking the vehicle at a suitable parking places and deliver the packages by foot from there this parking place. It may seems to cause the least congestion. But other people that also want to park their vehicle are now hindered. Imagine that a delivery vehicle uses a parking place for five hours in one street or parking location at a shopping centre, every time this will be the last parking place people have to park their vehicle further away from the destination they want to be. While, when the vehicle was only parked there for half an hour, less people at that location will be hindered. Besides that, the delivery process is not efficient anymore. Every route has an efficiency indicator which instructs how many packages need to be delivered in a certain time period. Efficiency may seems to be less important for the customers. Anyhow, a less efficient delivery process will surely influence the happiness of the customers. As a less efficient delivery process will lead to larger delivery times. Therefore, this measurement option calculates the congestion by measuring the amount of time spent at a specific parking place. Based on the stop time from the GPS device and the number op packages delivered from the scanner, the correct amount of stopping time can be measured. The correct amount of stopping time is equal to or lower than the time and packages set as efficiency indicator as mentioned above. The higher the correct stopping time, the less congestion is caused and the more efficient the trip has been.

4.2.5 Stopping on the road actions

(28)

28

Option 5.1: Number of stop actions

Logically, the more stop actions on the street are made, compared to stopping at a parking place, the relatively more congestion is caused. Based on the GPS data it can be retrieved where the driver has stopped the vehicle during the route and above that it is possible to check if the stop action was correct or not. Whether the stop action was correct, is related to the free space left requirements (see 4.2.1). Interestingly, a parking action is currently not just a parking action anymore, but can be specified in a correct or non-correct parking action. Based on those results, the caused congestion can be measured by comparing the number of non-correct stop actions with the total number of stop actions. The more non-non-correct stop actions are made, the more congestion is caused.

Option 5.2: Time spent on the stop action on the road

Knowing whether the vehicle has stopped on the road gives an indication of the caused congestion. Though, importantly, the time spent on that location gives more insights. As expected, the longer the duration of the stop, the bigger the chance of causing congestion. The caused congestion thus depends on the duration of the stop on the road compared to the total stop time. This measurement also takes into account the impact of the stop location on the efficiency of the delivery process. Again, the duration of the stop is not correct if it exceeds the efficiency rule. The more time is marked as not correct, the relatively more congestion will be caused.

4.2.6 Happiness

Both the happiness of the driver and the happiness of other traffic participants are taken into account. The logic behind this choice is that the objective measurements of congestion may not resemble the subjective perspective of congestion. Some people are, for example, happy with a reduction of 10% in the caused congestion, whereas other persons long for more reduction. Package delivery companies want to perform well for their customers. A lot of them are other traffic participants. Therefore, it is important that the happiness level of those customers is sufficient. Besides that, the happiness scores can also be used for validation of the factors in the congestion metric. Based on the opinions of the people in the street and the driver it is possible to evaluate whether every factor is equally important for that specific route. In this way, weights can be assigned to the factors in the metric.

(29)

29

may differ. Firstly, the application of those techniques for the other road traffic participants are described.

Option 6.1.1: Survey – Conducting surveys on the streets

The survey technique can be performed in several different ways. The first option is to ask traffic participants to fill in the survey while the delivery vehicle is present at the crossroad or in the street, so the traffic participants are currently facing the maybe congested situation because of the presence of the parcel delivery vehicle. A big advantage of this option is thus that the persons give their opinion at the moment they face the problem, so they do not have to imagine and think about a certain situation and it is thus possible to measure their direct feeling to the situation. A disadvantage of this option is that it is a quit expensive to perform, because persons should be placed on the street in order to hand out the surveys and be present at the same location as the delivery vehicle.

Option 6.1.2: Survey – IBeacons

The second option to conduct a survey research is by using IBeacons. IBeacons are placed in the parcel delivery vehicles and register people around the vehicle via mobile phone apps. Those registered people receive the survey on their phone after being present nearby a parcel delivery vehicle. This is a cheaper way to reach people facing the problem situation. However, IBeacons do not reach a lot of people at the moment, because the software is not included in many apps yet. Besides that, it is not guaranteed that everyone will definitely fill in the survey.

Option 6.1.3: Survey – Panel data

At last, surveys can be sent to panels. Advantage is the fact that people will fill in the survey, though there are not facing the problem situation at that moment. However, the answers of those panels can be valuable for indicating the importance of the factors per route.

Option 6.2.1: Interviews

(30)

30

Option 6.3.1: Observations

Observations are quite easy, but expensive to perform. Besides that the disadvantage is that you are guessing the people’s reaction regarding to situations by checking their movements or faces. Again, with the results of this method it is more difficult to create a happiness score than with the results of a survey.

Next to the options to measure the happiness of the other road traffic participants, there are also several options to measure the happiness of the driver of the parcel delivery vehicle.

Option 6.4.1: Survey

Surveys can be sent to the drivers via an anonymous source or via the company itself. It is preferred to use the anonymous source so that drivers do not get the feeling that the company is checking on their work. Besides that, a survey can also be sent after every route. However, then the question raises whether the drivers will fill in the survey honestly as it may become annoying to fill in the survey every time.

Option 6.4.2: Interviews

Interviews can also be held with the drivers. Again, an advantage is the fact that more information may be gathered, however the question raises whether drivers are willing to be interviewed.

Option 6.4.3: Observations

At last observations can be done. It is quite good to experience the situations yourself and ask the driver some questions during a ride along. Though, it can be quite expensive to have an observation person riding along multiple times. This method can lead to valuable information for validating the factors of the route. The observer namely experiences which factors are influencing congestion the most during that specific route.

4.3 Data sources used

(31)

31

to know exactly the time and location of the vehicle. Besides that it measures when the motor of the vehicle is shut down and started again and thus the stop time of a vehicle is also known. Lastly, there can be invested in additional tool of the GPS device which measures the use of the flashing lights. The accuracy of the GPS coordinates can be optimized by finding the optimal location at the vehicle to place the GPS device. Secondly, the license plate of the vehicle or vehicle type can be used to gather the exact sizes of the vehicle. Next to the stopping location and time, the scanner of the delivery boy measures the time and number of packages delivered during the stop. Though, to fulfil the measurements, external data is needed as well.

The interviews performed by interviewing people from different municipalities, the ministry of infrastructure and water management and research on the web, resulted in several available and suitable sources to gather the data of the region in which the vehicle is operating. First of all, a cooperation with the municipality in which the vehicle is operating is possible. The municipality can offer a lot of data. Most municipalities has a good overview of the traffic structures in their region. They know how many lanes a road have, where the crossroads are and sometimes even data about traffic busyness and traffic accidents. Besides that, mostly, they know the number and locations of the parking places. If cooperation with the municipality is not possible, many other (open) data sources are available to get the needed data. One of these open data sources is PDOK Viewer from PDOK. PDOK Viewer consists of a map in which different data can be shown, such as the number and type of accidents on a specific road segment, type of road, number of lanes and above that, the map consists of measurement options which make it possible to measure the length or width of road segments (PDOK, 2020). Another open data source which can be used is the open data source of the national road service (RDW). The open data of the RDW gives valuable static and dynamic parking data, such as number of parking places, time slots, charging posts for electric vehicles and many more (RDW, 2020). Lastly, for the region analysis, the non-free data source TomTom traffic can be used to get insights in the traffic busyness. TomTom has built a historical database which provides insights into the traffic situation on the road throughout the whole day, since 2008 (TomTom N.V., 2019). With the help of this database, the historical data makes it possible to identify traffic busyness patterns in the streets and the real time traffic data indicates the busyness at the moment the delivery vehicle enters the street.

4.4 Morphological report

(32)

32

Based on all the pros and cons of each measurement option two advices have been made. The first advice, the green line, indicates a starting point to quickly start measuring congestion. The line represents a situation that can easily be set up because no additional tools and great data transformations are needed. The red line represents a ‘future solution’. To set up this option, more time and some additional tools are needed. Besides that, it will take more time to collect and transform data.

Table 4.1 Morphological report

4.4.1 Starting option – Green line

(33)

33

data first, before the measurements and data collection of ‘keeping enough free space next to the delivery vehicle on every road segment’ are started.

To quickly start measuring the congestion, the following factors will be used: use of crossroads, time of presence, use of parking places and happiness. To measure the congestion related to the use of crossroads, the option of measuring the impact of movements on crossroads is advised. This option can easily be started, because data about the main traffic flows and location of the crossroads can be easily collected. Besides that, the GPS data of the route is already available. After the collection of this data, no data transformation is needed. The only step that has to be taken, is a check whether the main traffic flow is crossed. By simply counting these crossing actions the caused congestion can already be measured.

Time of presence is measured with the use of ‘time spent in busy periods method’. Traffic data sources about traffic busyness are already available and GPS data includes the time and location of the delivery vehicle. Having those two data sources, the most accurate and detailed method for measuring the time of presence can immediately be started. The ‘time spent during busy hour periods’ is more accurate than the measurement option ‘number of street visits during busy hour periods’. For example, a period in which a delivery driver enters a street at a quiet moment, but leaves it at a busy moment, could be seen as one visit at a busy street, while most of the time the street was quiet. Entering multiple streets for just a few seconds during busy periods, will lead to a high measured congestion level, while in fact the impact was lower. The caused congestion related to parking places cannot be measured very detailed, as the GPS accuracy may not be optimal. Though, a first step in measuring congestion related to parking actions can be taken by using the ‘parking place availability’ measurement. Within this measurement the number and sizes of parking places along the route are needed and the sizes of the vehicles are needed. With this data already a first choice of vehicle can be made. For example, a vehicle that does not fit in any available parking places along the route, will not reduce congestion compared to smaller vehicles. Within this method the likelihood of congestion can be measured by comparing delivery vehicles or comparing routes.

(34)

34

4.4.2 Future option – Red line

Within this option it is assumed that every tool and all data is available. So, the most detailed measurement options can be advised. Starting with the ‘correct driven time’ option for measuring the possibility to pass the delivery vehicle. As the GPS data is optimized it is possible to track the exact location of the delivery vehicle and check the space left next to the delivery vehicle. It is advised to use the time measurement option as it is the most accurate one of the three options. The ‘number of suitable roads used’ does namely not offer an exact indication about how much congestion is caused because the period of time is not taken into consideration. The same holds for the measurement based on the driven kilometres. A vehicle that spends more time on the non-suitable kilometres because of a low speed level, the measurement gives the same amount of caused congestion as a vehicle that drives quickly through non-suitable road segments. Therefore, measuring the amount of time spent on the non-suitable parts is more important and gives a more detailed insight in the caused congestion.

Secondly, the use of crossroads is now advised to be measured by using ‘the complexity of the crossroads’. The risk of measuring congestion by the movements on crossroads is that the turning action could also be counted as cause of congestion while no other traffic participants are present at the road. The crossroad complexity is based on the average traffic demand per time period and number of exits on a crossroad. This will not change frequently and is therefore a more secure method to indicate the caused congestion. Measurements based on the safety of the crossroad are not advised as accidents are a more volatile determinant than the average traffic demand. These measurements will lead to less accurate congestion measurements.

Congestion related to the time of presence is again advised to be measured with the ‘time spent during busy periods’ option. The amount of time spent during busy periods is more accurate and detailed than the number of times spend in busy periods.

(35)

35

For stopping on the road, the same reasoning holds as parking places. Only observing the number of stop actions gives less accurate and detailed results. Therefore, again, it is valuable to take into account the time period in the measurement. Also this option will benefit the efficiency of the delivery process.

Lastly, happiness can now be extended with observations and interviews. Both methods are a good complement to the surveys. Observations and interviews can be performed at the place of the presence of the delivery vehicle and may lead to more and other interesting results than results from surveys. Though, the disadvantage of observations and interviews are that the results out of those methods are more difficult to transform into a score. Also the IBeacon could be used in the future. The advantage of the IBeacon is that it is quite a cheap solution to reach a lot of people around the delivery vehicle. Though this method can only be used if the number of apps or users of the IBeacon will grow. In that way the chance of reaching people around the vehicle grows.

4.5 The congestion metric

The chosen measurement options can be gathered in a metric to calculate the caused congestion. The metric follows several steps which are explained below and can be found in appendix E.

The first step in the congestion metric is creating the benchmark values. Based on those benchmark values it can be indicated how much congestion is caused. The benchmark values are mainly collected via the data from the region analysis of the driven route. This data can be collected via an agreement with the municipality, or online data sources such as PDOK Viewer and TomTom. The collected values are the denominators of the measurement options of the morphological report, for example the number and locations of suitable parking places or the suitable streets. The collected data will be put in the right format, for example expressing the suitable street segments in kilometres. After transforming the data in the right format, the data is stored in the database of the metric as benchmark values.

(36)

36

After creating the benchmarks, the congestion metric can be filled with the current results. Firstly, the results of the delivery trip will be collected. As this data comes from different data sources, like GPS data from the vehicle and scanner data from the scanner used during the delivery, the data first needs to be combined in one data file. After combining all the data, the data will be checked and put in the right format. After that, the scores of the factors per route are sent to the congestion metric. In the congestion metric the scores are stored as the current results per route and per factor.

At last, the current happiness level is collected. By using valid methods as, observations, surveys and interviews, the current happiness level of both the traffic participants and the driver are measured. This measurement will lead to a happiness grade or score and will be sent to the congestion metric. Within the congestion metric, the results are marked as current happiness results and stored in the metric.

(37)

37

5. Conclusion and discussion

This study shows that it is possible to measure the congestion created by the delivery vehicles in city centres. To measure this specific type of congestion several factors relating to the delivery vehicle and delivery process must be taken into account. Those factors are the possibility to pass by the delivery vehicle, use of crossroads, time of presence, parking places, stopping on the road and happiness. The data needed to make it possible to measure these factors is identified and it is shown how to use this data to exactly measure the caused congestion. Besides that, this study shows that companies can choose the complexity and accuracy of the congestion measurements. By performing quick and easy calculations companies can quickly start measuring their congestion. Or companies who are either focussing more on congestion reduction or want to outcompete other companies on congestion performance, can choose for the more accurate options. The last advantage is that the metric shows that it is easy to choose for different measurement options and that not every factor is needed to start measuring. Therefore every company can choose their own factors and even add some other factors by themselves.

5.1 Theoretical implications

First of all, a more general definition related to urban congestion is defined. Existing definitions focus specifically on speed levels and capacity implications which causes congestion. Though, this new definition: “urban congestion is nuisance by interrupting the road for other road users with your delivery vehicle”, is applicable to every situation in the city and not specific related to a certain action which will not occur or impact the congestion at some places.

(38)

38

measurements. The same holds for the Stagnationfactor of Rijkswaterstaat (2007). An additional check of the delivery vehicle location is needed to know if the parcel delivery vehicle contributed to the congestion in the street or not. Next to that, this study contradicts the speed-flow envelope of Maitra, Sikdar, & Dhingra (1999), which indicates that a vehicle is hindered because of congested situations when the speed of the vehicle is reduced. This may also be the case for delivery vehicles, but not always. A delivery vehicle can also slow the speed in an empty street to deliver a package, though this does not mean that the vehicle has to slow down because of congestion on the road. The speed-flow envelope was created to measure specifically in urban areas, though this study shows us that the speed-flow envelope is not applicable to all kinds of traffic modes in the city centre. Measurements created in this study are focused on the actions of the vehicle and measures based on those actions the impact related to congestion specific to that vehicle. Next to that, those measurements are focussed on multiple congestion determinants. Current measurements were only focussed on one congestion determinant, for example, accidents (Pulugurtha, M.ASCE, & Pasupuleti, 2010) or the complexity of the road network (Wen, Chin, & Lai, 2017).

5.2 Managerial implications

(39)

39

5.3 Limitations and future work

The morphological report is a good starting point for innovative solutions to measure congestion caused by delivery vehicles. Though, creating those congestion measurements have only just begun. Unfortunately, due to COVID-19 virus, the streets became empty and people had to stay home as much as possible. Leading to, actually, ‘congestion-free cities’ in which it was not possible to test the created measurement options, observe new delivery vehicles in the city centre or perform interviews with traffic participants to ask them about their experience or happiness related to the congestion on the urban streets. Therefore, the input data for the morphological report is now only based on a few interviews and observations during the design sprint. Next to that, because of the importance of the happiness of both the driver and other traffic participants, the factor cannot be left out in the analysis of measuring the congestion impact of delivery companies. That is why it is chosen to discuss with both internal and external people or companies about the options of measuring happiness of both the driver and other traffic participants. Those methods of measuring the happiness are described and comparisons of the different methods are made to investigate both the advantages and disadvantages per method. Besides that, the morphological report is now only focused on the main actions of the delivery vehicle. But to get the most accurate and detailed congestion measurements there are a lot of other small determinants which can be taken into account, such as weather conditions and driver experience.

(40)

40

References

Adler, M. W., van Ommeren, J., & Rietveld, P. (2013). Road Congestion and Incident Duration. Economics of Transportation, 2(4), 109-118.

ANWB. (2020, June 30). Lading op auto's en aanhangwagens. Retreived from ANWB: anwb.nl/juridisch-advies/in-het-verkeer/verkeersregels/afmetingen-van-autos-en-aanhangers

Arnott, R. (2015). A Bathtub Model of Downtown Traffic Congestion. Access, 46, 26-33.

Atomode, T. I. (2013). Assessment of Traffic Delay Problems and Characteristics at Urban Road Intersections: A Case Study of Ilorin, Nigeria. IOSR Journal Of Humanities And

Social Science (IOSR-JHSS), 12(4), 6 - 16.

Banfield, R., Lombardo, C. T., & Wax, T. (2015). Design Sprint: A Practical Guidebook for

Building Great Digital Products. California: O'Reilly Media Inc.

Benjelloun, A., & Crainic, T. G. (2008). TRENDS, CHALLENGES, AND PERSPECTIVES IN CITY LOGISTICS. Transportation and land use interaction, proceedings

TRANSLU, 4, 269-284.

Biswas, S., Chandra, S., & Ghosh, I. (2017). Effects of On-Street Parking In Urban Context: A Critical Review. Transportation in Developing Economies, 3(10), 1 - 14.

Brown, J. L. (2018, June 27). Empathy Mapping: A Guide to Getting Inside a User’s Head. Retreived from UXBooth: https://www.uxbooth.com/articles/empathy-mapping-a-guide-to-getting-inside-a-users-head/

Çolak, S., Lima, A., & González, M. C. (2016). Understanding congested travel in urban areas. Nature Communications, 7(1), 1-8.

Consultant, P. (2015). Week 4: Double Diamond Framework. Retreived from The Faces of Amnesty: https://interactiondesign17.wordpress.com/2017/02/09/week-4-double-diamond-framework/

Dalton, J. (2019). Great Big Agile. Berkely: Apress.

(41)

41

Deloison, T., Hannon, E., Huber, A., Heid, B., Klink, C., Sahay, R., & Wolff, C. (2020). The

Future of the Last-Mile Ecosystem. Cologny/Geneva: World Economic Forum.

Design Council. (2015, March 17). What is the framework for innovation? Design Council's

evolved Double Diamond. Retreived from Design Council:

https://www.designcouncil.org.uk/news-opinion/what-framework-innovation-design-councils-evolved-double-diamond

Ferrara, A., Sacone, S., & Siri, S. (2018). Freeway Traffic Modelling and Control. Berlin: Springer.

Fietsersbond.nl. (2020, July 1). Fietspaden. Retreived from Fietsersbond:

https://www.fietsersbond.nl/ons-werk/infrastructuur/fietspaden/#:~:text=In%20het%20'Handboek%20Wegontwerp'%2 C,standpunt%20van%20kenniscentrum%20CROW%20Fietsberaad.

Fuchs, C., & Golenhofen, F. (2019). Mastering Disruption and Innovation in Product

Management - Connecting the dots. Cham: Springer.

Fuldaur, E. (2019, May 17). The Last Mile issue: how can we solve urban delivery problems? Retreived from Tomorrow. MAG by Tommorow.City:

https://www.smartcitylab.com/blog/mobility/the-last-mile-issue-how-can-we-solve-urban-delivery-problems/

Gibbons, S. (2018, December 9). Journey Mapping 101. Retreived from Nielsen Norman Group: https://www.nngroup.com/articles/journey-mapping-101/

Groen, M., Vos, G., Verweij, K., Otten, M., Tol, E., de Goffau, W., . . . Nesterova, N. (2019).

Laadinfrastructuur voor elektrische voertuigen in stadslogistiek. Topsector Logistiek.

Haug, J., & Schuurman, F. (2015). Voetpaden voor iedereen. Utrecht: Bouw Advies Toegankelijkheid.

He, F., Yan, X., Liu, Y., & Ma, L. (2016). A Traffic Congestion Assessment Method for Urban Road Networks Based on Speed Performance Index. Procedia Engineering,

137, 425 - 433.

Hevner, A. (2007). A Three Cycle View of Design Science Research. Scandinavian Journal of

(42)

42

Hillyer, M. (2020, January 10). Urban Deliveries Expected to Add 11 Minutes to Daily

Commute and Increase Carbon Emissions by 30% until 2030 without Effective Intervention. Retreived from World Economic Forum:

https://www.weforum.org/press/2020/01/urban-deliveries-expected-to-add-11-minutes- to-daily-commute-and-increase-carbon-emissions-by-30-until-2030-without-effective-intervention-e3141b32fa/

Homström, J., Ketokivi, M., & Hameri, A.-P. (2009). Bridging Practice and Theory: A Design Science Approach. Decision Sciences, 40(1), 65-87.

Interaction design foundation. (2020, 01 20). Problem Statement. Retreived from Interaction design foundation: https://www.interaction-design.org/literature/topics/problem-statement

Kaddoura, I., & Nagel, K. (2018). Simultaneous internalization of traffic congestion and noise exposure costs. Transportation, 45(5), 1579-1600.

Kumar Sharma, H., & Swami, B. (2016). Congestion Characteristics of Interrupted Flow for Urban Roads with Heterogeneous Traffic Structure. 2016 5th International Conference

on Transportation and Traffic Engineering (ICTTE 2016) (pp. 1-6). MATEC Web of

Conferences.

Lidgey, B. (2018, July 16). What is a lightning talk? Retreived from Medium: https://medium.com/@benlidgey/what-is-a-lightning-talk-6344fe239e3e

Maitra, B., Sikdar, P., & Dhingra, S. (1999). MODELING CONGESTION ON URBAN ROADS AND ASSESSING LEVEL OF SERVICE. Journal of Transportation

Engineering, 125(6), 508-514.

Manley, E., & Cheng, T. (2010). Understanding road congestion as an emergent property of traffic networks. 14th WMSCI, 1 - 7.

Michinov, N. (2012). Is Electronic Brainstorming or Brainwriting the Best Way to Improve Creative Performance in Groups? An Overlooked Comparison of Two Idea-Generation Techniques. Journal of Applied Social Psychology, 42(1), E222-E243.

Mohan Rao, A., & Ramachandra Rao, K. (2012). MEASURING URBAN TRAFFIC CONGESTION – A REVIEW. International Journal for Traffic and Transport

Ubiquitous city logistics an illusion, or can we create model(s) to measure and reduce congestion?