Collaboration in city distribution with Urban Consolidation Centers and a White Label Last Mile

(1)

Collaboration in city distribution with

Urban Consolidation Centers and a

White Label Last Mile

Combined Thesis

MSc. Supply Chain Management

MSc. Technology and Operations Management

University of Groningen – Faculty of Economics and Business

(2)

2

ABSTRACT

Purpose: The purpose of this thesis is to investigate collaboration in last mile delivery with Urban

Consolidation Centers (UCC) and a White Label Last Mile. UCCs in combination with a White Label Last Mile are considered as an initiative to reduce the negative impacts of last mile delivery such as congestion and gas emissions. Practice shows that a successful development of such a network is hard to accomplish. This thesis investigates what the differences are between the scenario of a UCC with a White Label Last Mile and the current situation of Logistical Service Providers (LSPs) doing their own last mile delivery.

Methodology and approach: This study uses a deductive study approach. A quantitative model is

developed to assess the differences of a UCC network in comparison to the current situation of last mile delivery where every LSP focusses on its own processes. The developed quantitative model is used in a business case for a company which is interested in developing a UCC with a White Label Last Mile. Performance indicators in this study are the travelled kilometres in the city distribution, the total costs of the city distribution and emissions.

Findings: A situation in which White Label Last Mile is combined with a UCC is outperforming the

conventional situation in which every LSP execute the last mile by themselves for all three performance indicators that are investigated. The travelled kilometres in city distribution show a reduction up to 80%, the costs of distribution up to 25% and the emissions up to 90%, where the first two performance indicators could even have a bigger reduction when the UCC network grows with more LSPs.

Practical implications: A lot of initiatives for the development of a UCC fail despite their benefits. A

reason for this is that stakeholders are not willing to participate. This study gives a better insight why it is interesting for these stakeholders to invest in a UCC network, as a UCC network outperforms the conventional situation of last mile delivery for all included performance indicators.

Originality/contributions: The use of a UCC has already been widely investigated by several authors.

(3)

3

TABLE OF CONTENT

ABSTRACT ... 2 TABLE OF CONTENT ... 3 1. INTRODUCTION ... 4 2. THEORETICAL BACKGROUND ... 5

2.1 Last mile collaboration ... 6

2.2 UCCs in practice ... 7

2.3 Cargo bicycles in distribution ... 8

2.4 Last mile collaboration in a UCC with cargo bicycles ... 9

3. METHODOLOGY ... 9

4. CASE STUDY ... 13

4.1 Current situation ... 14

4.2 Desired situation ... 14

4.3 What is needed for a network with a UCC ... 15

5. COMPUTATIONAL STUDY ... 17

5.1 Simulation of different scenarios... 17

5.1.1 Scenario 1: UCC network based on the pilot from the business case with single LSP ... 18

5.1.2 Scenario 2: UCC network with multiple and varying LSPs ... 19

5.1.3 Scenario 3: UCC network with B2B LSPs ... 20

5.1.4 Scenario 4: UCC network with parcel delivery LSPs ... 22

5.2 Discussion ... 24

6. CONCLUSION ... 27

6.1 Summary ... 27

6.2 Managerial insights ... 27

6.3 Limitations... 28

6.4 Recommendations for further research ... 29

REFERENCES ... 31

APPENDIX ... 34

(4)

4

1. INTRODUCTION

Last mile delivery is the home delivery service for a customer (Punakivi, Yrjölä & Holmström, 2001) and is seen as a logistical challenge (Boyer, Prud’homme & Chung, 2009) as it is regarded as “the more expensive, least efficient and most polluting section of the entire logistics chain” (Gevaers, Voorde & Vanelslander, 2014 pp. 398). The increasing shopping through e-commerce has the effect that customers order more packages online that are delivered at their home, which increases the importance of last mile deliveries in a city (Arnold, Cardenas, Sörensen & Dewulf, 2018; Iwan, Kijewska & Lemke, 2016). This increases the negative impact that deliveries have on urban areas: gas emissions, congestion, noise, traffic safety issues and other disturbances (Nordtømme, Bjerkan & Sund, 2015). The largely used delivery model is inefficient and unproductive, as last mile logistics are fragmented and hence uncoordinated (de Souza, Goh, Lau, Ng & Tan, 2014).

Allen, Browne, Woodburn and Leonardi (2012) review the use of Urban Consolidation Centers (UCCs) in city distribution, which is an initiative to reduce the negative impact of last mile delivery by reducing goods vehicle traffic. Their main conclusion is that UCCs can generate both internal and external benefits, as making use of distribution via a UCC can improve supply chain performance and reduce environmental and social impacts (Allen et al., 2012). Simoni, Bujanovic, Boyles and Kutanoglu (2018, pp. 112) describe a UCC as a “transshipment point situated in the proximity of a city centre, where deliveries from logistic companies are dropped off, sorted, and consolidated in smaller vehicles such as minivans, electric vans and cargo bikes”. Despite the advantages that have been addressed before, developing a UCC means high operating costs and a high reliance on governmental subsidies, which in most initiatives lead to an early termination of the initiative (Simoni et al., 2018; Verlinde, Macharis & Witlox, 2012; Kin, Verlinde, Lier & Macharis, 2016). These disadvantages led, together with the lack of support of transporters to participate in UCC networks (Park, Park & Jeong, 2016), to the fact that only limited initiatives for UCCs have been successfully executed. In the literature, the focus of research has mostly been in delivery networks where each company has its own UCC. Muñoz-Villamizar, Montoya-Torres & Vega-Mejía (2015) found several benefits of collaboration in last mile delivery, however last mile collaboration has not been investigated in combination with a UCC (Simoni et al., 2018).

(5)

5 in this topic is Fietskoeriers.nl, which is a platform in The Netherlands that has hubs in several cities where transporters could drop off their goods. In this platform, Fietskoeriers.nl will focus on the last mile distribution for the goods that are delivered at their hubs. The topic of a White Label Last Mile will be investigated in this thesis and therefore the following research question is developed: “What are the benefits of collaboration in a White Label Last Mile delivery setting with use of Urban Consolidation Centers?”

Practice and literature demonstrate that it is interesting to investigate the benefits of last mile collaboration in a network which makes use of a UCC. The contribution of this thesis will be in quantifying the differences between an UCC network with a White Label Last Mile and the current way of performing city distribution where every LSP delivers goods by themselves. A quantitative model is developed which will account for routing and fleet choice for the last mile transporter. Comparing this quantitative model with the current situation of last mile delivery in which each LSP delivers the goods by themselves, will result in an indication of the benefits of a shared UCC. This comparison will be based on the kilometres driven to deliver the goods, the costs of the city distribution and the emissions that belong to it.

This introduction will be followed by a research framework, which can be found in chapter 2, in which current theoretical contributions are discussed regarding last mile collaboration, the practical use of UCCs and the use of cargo bicycles in last mile distribution. In chapter 3, these three concepts will be combined and a quantitative model will be developed. The model in this research design will be used in a computational study which will answer the developed research question. In chapter 4, a case will be introduced which is used to retrieve data for the simulations. The collected data from this case will be used as input for the quantitative model, which will give results for the developed situations of a network without and with a UCC. In chapter 5, the results of the computational study will be described, after which they will be analysed and discussed. The thesis will finish with a conclusion which can be found in chapter 6. The conclusion starts with a summary of the thesis and an answer to the research question, whereafter a managerial insight is presented which is of interest for both academics as well as professionals. The thesis is finished with the limitations of the study and recommendations for future research.

2. THEORETICAL BACKGROUND

(6)

6 operating on the same level of the supply chain, who have similar or complementary transportation needs” (Vanovermeire, Sörensen, Van Breedam, Vannieuwenhuyse & Verstrepen, 2014, pp. 339). The goal of horizontal collaboration for LSPs is to increase their productivity. Cruijssen, Cools and Dullaert (2007) investigated the opportunities that cooperating LSPs can benefit from, of which increasing the customer service and reducing costs of non-core activities are the highest valued ones. Cruijssen, Dullaert and Fleuren (2007) present a broad literature review regarding horizontal cooperation in transport and logistics and its drivers, impediments and facilitators. I refer to this article as starting point for further research in the overarching subject of horizontal logistics cooperation.

2.1 Last mile collaboration

Collaboration in last mile delivery in urban areas is investigated by several authors, for example by Lindawati, van Schagen, Goh and de Souza (2014) who provide an exploratory study that investigates the motivations to collaborate in urban logistics. New perspectives are developed regarding stakeholder participation in such initiatives and key factors of influence are identified. Next to exploratory studies, quantitative models can be developed regarding the urban distribution of goods. Muñoz-Villamizar et al. (2015) indicate that a collaborative and mutual strategy in urban freight transport is a promising area to study and they therefore compare a collaborative and non-collaborative scenario in the urban distribution of goods. Their main finding is that collaboration in the last mile delivery leads to a reduction in transportation costs and an increase in service level because of higher utilization of vehicles. This thesis differs from the paper by Muñoz-Villamizar et al. (2015) as the latter one assumes that still multiple companies perform the last mile delivery where this is done by one neutral, White Label company in this thesis. Taniguchi and Van Der Heijden (2000) indicate that collaboration in transport systems will lead to less vehicles that collect and deliver the same amount of goods and they conclude that collaboration leads to a reduction in environmental impacts. Next to this, they quantify benefits regarding carbon emissions and required vehicles in the routing. Companies can become more competitive by collaborating with other companies, as this allows for a use of economies of scale and an increased resource utilisation (Montoya-Torres, Muñoz-Villamizar & Vega-Mejía, 2016).

(7)

7 (long-haul) transporters do not want to participate in a city logistics network in which one company takes the lead. A successful example from practice of a neutral transporter in a logistics network can be found in the German city Kassel (Taniguchi & Van Der Heijden, 2000). A distribution network which makes use of a UCC has some differences compared to a conventional distribution network. Figure 1, which is created by Clausen et al. (2016), shows how a UCC works in comparison to a situation which reflects how currently last mile delivery is organised in urban areas.

Figure 1. Visual representation of the difference between current last mile delivery (left) and last mile

delivery by making use of a UCC (right) (source: Clausen et al. (2016))

2.2 UCCs in practice

(8)

8 A recent example of a pilot that did not even succeed in initiating can be found in the municipality of Haarlem (NL), where a feasibility study was performed to investigate the possibilities of a small-scale Urban Distribution Center. This Urban Distribution Center would function the same as a UCC, however the conclusion of the study was that its creation was not feasible for Haarlem. Two reasons for this were (1) a lack of support by entrepreneurs in Haarlem and (2) the several attempts that have been made in the Netherlands (and other countries) to develop a UCC network where none of them have become fully operational on their own. Lindawati et al. (2014) found two other arguments that hinder stakeholders to participate in an urban logistics collaboration initiative: the perceived benefits of collaboration and the risk of losing competitive intelligence. On the other hand, some examples of a successful development of a UCC can also be found (e.g. Taniguchi, 2015). Björklund, Abrahamsson and Johansson (2017) have developed seven critical factors for initiating a viable business model in which city logistics are done via a UCC. Three of these seven factors are of high importance for a successful development: (1) scaling up to variable costs, (2) logistics competence to access potential value streams and (3) the ability to take full advantage of advanced IT. To conclude, practice shows that the development of a UCC can be relatively hard despite its advantages, however when the identified factors are considered the development of a UCC can be executed successfully.

2.3 Cargo bicycles in distribution

(9)

9

2.4 Last mile collaboration in a UCC with cargo bicycles

Three elements have been discussed in this theoretical background: logistics/last mile collaboration, Urban Consolidation Centers and distribution with cargo bicycles. Combining these three elements creates an interesting subject to investigate: a distribution network where transporters collaborate in a UCC where a neutral company is present to perform the last mile distribution with cargo bicycles. Investigating UCCs where transporters collaborate is a topic which is proposed by Simoni et al. (2018). In their paper, a mathematical model is developed for a UCC based on a multi-depot vehicle routing problem with a heterogeneous vehicle fleet. The paper presents a case study for the city of Austin (US) where one company has a UCC from where a variety of vehicles do the delivery of goods. The paper and case present this for city logistics and provide useful information on the implementation of a UCC. Simoni et al. (2018) conclude that the biggest obstacle in a network like this is the large upfront costs. Therefore, they propose other transporters to join the initiative and share the UCC to reduce facility costs. This is in line with the conclusion of Muñoz-Villamizar et al. (2015), who state that collaboration in last mile delivery will lead to a cost reduction.

A given recommendation for future research by Simoni et al. (2018) is to extend their model to a model in which several transporters are represented in a shared UCC. This recommendation is in line with the conclusion of Janjevic, Lebeau, Ballé Ndiaye, Macharis, Van Mierlo and Nsamzinshuti (2016), who investigate several scenarios for urban city distribution with UCCs. Janjevic et al. (2016) state that there are numerous uncertainties with regard to the implementation of a UCC, for example the vehicle fleet, the location of the UCC(s) and the operational agreements. They investigate three of their developed scenarios for the Brussels-Capital Region in Belgium and evaluate these scenarios on five performance measures: cost of transport, congestion, emissions, noise and gross margin. In the end, they conclude that such scenarios should also be further evaluated with consultation of local stakeholders (Janjevic et al., 2016). All in all, a collaborative distribution network for urban areas needs to be investigated in which the perspective of several stakeholders is involved. This will be done in this thesis, where situations in which LSPs perform their own last mile are compared with situations in which a UCC is present with a White Label company that performs the last mile distribution.

3. METHODOLOGY

(10)

10 be taken into account in the model: the amount of kilometres driven in the distribution, the total costs of the city distribution and the emissions of the vehicles that are used.

Simoni et al. (2018) provide a quantitative model which investigates a multi-depot vehicle routing problem for UCCs with heterogeneous vehicle fleet. This model gives a clear overview of what is needed for a quantitative model in situations where a transporter has its own UCC. As stated before, collaboration in last mile delivery gives several benefits (Muñoz-Villamizar et al., 2015). Simoni et al. (2018) conclude their study with the recommendation to extend their model to represent several transporters that make use of a shared UCC. This extension is in line with the idea that last mile collaboration is beneficial over the current situation where all transporters focus on their own business activities. Janjevic et al. (2016) recommend to investigate different scenarios of UCC networks and consult local stakeholders in them, to come up with a scenario that will lead to the highest acceptance for all involved stakeholders. This study will contribute to this because a quantitative model will be developed for the situation in which multiple transporters are present in a UCC. The insights retrieved from the models by Simoni et al. (2018) and Muñoz-Villamizar et al. (2015) and the paper by Janjevic et al. (2016) are valuable for this and therefore serve as a source of inspiration for the model that will be developed in this study.

Some assumptions need to be developed for the problem formulation for a distribution network which makes use of a UCC. First, goods come from several transporters and are dropped off at the UCC, from where they are distributed to their destinations in the city centre. In this, all transporters are responsible to get their own goods at the UCC and the neutral company is responsible for the city distribution. The long-haul transportation will be disregarded in the model that will be developed, as its focus is on city distribution. For the city distribution and the UCC, it is assumed that the UCC has an infinite capacity for storage. This simplifies the model in advance and constraints could be added in further research to investigate the effects of a capacitated UCC. A third assumption is that all transporters have enough vehicles to get all the goods at the UCC. One of the questions that could be solved with the model is how many cargo bicycles the neutral company needs, assuming they already have some available right now. The fourth assumption is regarded to this, as the neutral company has the preference to use cargo bicycles which are limited in capacity in comparison to the vans and trucks.

The model itself has the goal to minimize the total costs of delivery through a UCC, where the costs consist of distance-related transportation costs to the final destinations, the fixed costs of using a specific vehicle type and the operational costs of using a UCC. The model has the following decision variables: X(k) corresponding to 1 if the UCC is open and 0 otherwise

(11)

11 The sets of the model are:

H set of available vehicles

I set of demand nodes

K candidate facility for UCC N I ∪ J set of all nodes

The inputs of the model are:

l(i) demand at customer node i f(k) fixed costs of opening UCC k h(k) operational costs of UCC k

c(hij) transportation costs between node i and node j using vehicle h g(h) fixed costs of using vehicle type h

u(h) capacity of vehicle type h

The objective function of the model will be:

𝑚𝑖𝑛 ∑ ∑ ∑ 𝑐(ℎ𝑖𝑗) ∗ 𝑍(ℎ𝑖𝑗) 𝑗∈𝑁 𝑖∈𝑁 ℎ∈𝐻 + ∑ 𝑔(ℎ) ∗ 𝑌(ℎ) ℎ∈𝐻 + ∑ 𝑓(𝑘) ∗ 𝑋(𝑘) 𝑘∈𝐾 + ∑ ∑ ∑ ℎ(𝑘) ∗ 𝑍(ℎ𝑗𝑖) 𝑗∈𝑁 𝑖∈𝑁 ℎ∈𝐻

Next to the objective function, the following constraints are developed to make the model representable for a real-world situation:

Constraint 1: Every customer j must be on exactly one route

∑ ∑ 𝑍(ℎ𝑖𝑗) = 1 ∀ 𝑗 ∈ 𝐼

𝑖∈𝑁 ℎ∈𝐻

Constraint 2: Flow conservation

∑ 𝑍(ℎ𝑖𝑗)

𝑖∈𝑁

− ∑ 𝑍(ℎ𝑗𝑖)

𝑖∈𝑁

(12)

12 Constraint 3: Elimination of subtours

∑ ∑ ∑ 𝑍(ℎ𝑖𝑗)

ℎ∈𝐻 𝑗∈𝜃 𝑖∈𝜃

≤ |𝜃| − 1 2 ≤ |𝜃| ≤ 𝐼, 𝜃 ⊆ 𝐼

Constraint 4: Vehicle capacity

∑ 𝑙(𝑖)

𝑖∈𝐼

∗ ∑ 𝑍(ℎ𝑖𝑗)

𝑗∈𝐽

≤ 𝑢(ℎ) ∗ 𝑌(ℎ) ∀ℎ ∈ 𝐻

The following constraints need to be added to make the decisions variables binary: 𝑋(𝑘) ∈ {0,1} ∀𝑘 ∈ 𝐾

𝑍(ℎ𝑖𝑗) ∈ {0,1} ∀𝑗 ∈ 𝑁, ∀ℎ ∈ 𝐻 𝑌(ℎ) ∈ {0,1} ∀ℎ ∈ 𝐻

(13)

13 happens when more/less LSPs join the UCC network than expected. It could be possible that the last mile delivery network becomes more or less efficient when a big amount of goods is transported through the UCC. Knowledge is needed about this and could be retrieved via this study. It could mean a clear prospect for stakeholders what happens when they join the last mile delivery network by dropping off their goods at the UCC.

4. CASE STUDY

As stated before, developing a UCC is a topic which is of high interest for both practitioners as well as researchers (e.g. Clausen et al. (2016); Heeswijk et al. (2017); Taniguchi (2015)). A company which is highly interested in developing a UCC in which the last mile distribution is done by cargo bicycles is the Dutch Cycloon Post & Fietskoeriers. Cycloon is a company that is active in city logistics and which offers postal services. Next to this, Cycloon has developed the network Fietskoeriers.nl, which provides last mile delivery with cargo bicycles throughout the Netherlands. Transportation in cities is mostly done by bicycle messengers, only if a specific good does not fit into a cargo bicycle a green gas van is used. These vans are also present to execute the long-haul transportation between the cities where Fietskoeriers.nl is present. Currently, the number of cities where bicycle messengers deliver goods for Fietskoeriers.nl is 34. However, there is enough room to grow as every bicycle messenger company can join the network for a new city where Fietskoeriers.nl is not present yet.

Fietskoeriers.nl believes that the solution to the problems in city distribution is the (cargo) bicycle. They have an infrastructure available that includes 600 bicycle messengers in 34 cities, which has never been accomplished before in the Netherlands. There is no company that has been able to develop a likewise network before, where both hubs and last mile distributors are combined. They currently focus on the Business-To-Customer (B2C) market, where companies send their goods to the customers. This is done with a same-day guarantee, which means that a customer can order a package in the morning and receive this from a bicycle messenger in the evening. The Business-To-Business (B2B) market is also served, however the size of this market is somewhat limited for Fietskoeriers.nl.

(14)

14

4.1 Current situation

In the current situation of logistics and city distribution, all LSPs are focusing on their own activities and optimizing their processes. The LSPs pick up the goods they have to transport, or take it from their warehouse. From there, the goods are transported to the city where they need to be delivered. This delivery can be done with vans or trucks. When arriving at the city, there are two possible ways for the goods to arrive to the customer: (1) the goods are directly delivered at the customer or (2) they are first delivered at a(n) (intermediate) hub of the transporter whereafter the goods are delivered to the customer. This latter option is mostly done in the B2C market, where the last mile delivery is executed by vans and the long-haul transportation of goods is done by trucks.

All LSPs strive for the best delivery schedule and have the goal to optimize their processes, for example by minimizing the total distribution costs. They do this for example by minimizing the total amount of kilometres in their routing, which implies the lowest transportation costs and fastest delivery. Another objective for determining an optimal delivery schedule is to take the most economical trucks for example, as these trucks will use the least fuel or have the lowest emissions. However, because every LSP only focus on their own activities, the total distribution network might not be optimal. In practice, it is possible that several vans from different companies need to deliver a good at the same location/street. This will mean that a specific area is visited by a lot of vans from different companies, sometimes even in the same time interval. This could be seen as far from optimal for the total distribution network in a city, because these vans could literally drive after each other from one of their addresses to another. In the end, this contributes to the negative impact that deliveries have on urban areas for example in the form of congestion.

4.2 Desired situation

(15)

15 and group all these goods, which come from several transporters. The next thing that will be done is the delivery of all these goods, which will be done by cargo bicycles and, when needed, by green gas vans. In the end, it will mean that multiple streams of goods are coming into the UCC, whereafter only one stream of cargo bicycles will depart from the UCC in the city centre.

As stated before, Fietskoeriers.nl has a share in both the B2C as well as the B2B last mile. Their share in the B2C market is growing since their establishment in 2016. When investigating possibilities for collaborating in the B2C market, Fietskoeriers.nl found that the other companies present in this market do not want to collaborate in a White Label Last Mile. This is mainly because these companies are afraid of losing brand awareness as someone else than their employees will be delivering the packages to the customers. The focus of Fietskoeriers.nl in the B2C market is on acquiring (web)shops to choose for them instead of for the traditional package delivery companies.

In the B2B market, there is an interesting growth potential for Fietskoeriers.nl to collaborate with other transporters. These transporters are LSPs that do not specifically focus on the last mile distribution, but who have core activities like long distance transportation, warehousing or the transport of large goods. Despite the focus of these companies, all transporters have to deal with a last mile in their deliveries. Therefore, these companies could be interested in collaborating in a White Label Last Mile with UCCs.

4.3 What is needed for a network with a UCC

For creating a delivery network with a White Label Last Mile and UCCs where the distribution is executed with cargo bicycles, some specific things are needed. First, you need a location for the UCC where long-haul transporters can deliver their goods. The location for this UCC is important as it should be easily accessible, not only for the long-haul transporters but also for the bicycle messengers. An ideal location would be, according to Kin et al. (2016), where the UCC is easily accessible by main roads but at the same time in relative close proximity to the delivery area. Fietskoeriers.nl currently has a location in Zwolle available which is in the proximity of the highway but also close to the city centre.

(16)

16 A third thing which is needed in a network with a UCC is the presence of both long-haul LSPs that are willing to collaborate in a UCC as well as bicycle messenger companies that are willing to receive these goods and distribute them in the city centre. Building stable relationships with these companies will take some time but this will contribute to a successful implementation of the network. Something which needs to be aligned in this is the use of cargo bicycles for the city distribution. The bicycle messenger companies need to have enough cargo bicycles to deliver all the goods on the agreed time. It is interesting to investigate how much cargo bicycles would be needed for a successful and in-time delivery of the goods; however, this is fully dependent on the amount of goods that will be delivered at the UCC. Something which should be considered here is that the amount of goods can differ per day; some days can be really loaded with lots of goods while the amount of goods on other days can be disappointing. Next to this, the size and weight of the goods will be diverse where some goods can be easily transported by a cargo bicycle while others will be too big or too heavy and need to be delivered by a van or even a truck. Information about the amount of goods, their size and their weight should be shared between the long-haul LSPs and the bicycle messenger companies, so all parties know what to expect and which goods can and cannot be distributed via a UCC. Next to the integration of physical processes and activities, the invoicing process between the companies should be aligned. The preparation for the pilot of Fietskoeriers.nl shows that both companies invoice in a different way. Fietskoeriers.nl invoices a price per package that needs to be delivered, no matter the size and weight of the good. The LSP that is willing to collaborate invoices based on the size and weight of the good and takes an additional percentage as a fuel surcharge. When collaborating, both companies should have an integrated invoicing system that works for both of them.

(17)

17 is to make it compatible for all types of software that are used by the long-haul transporters. A UCC network will be ready for its practical implementation when such an overarching software is developed and all other identified conditions mentioned are met.

5. COMPUTATIONAL STUDY

This chapter provides a computational study based on the quantitative model that has been developed in the methodology chapter. This model is implemented in Python 3.6 on a Microsoft device running Windows 10. The optimization software Gurobi Optimizer 7.5.2 has been used to solve the quantitative model. A second model has been developed to represent the conventional situation in which every LSP distributes goods in their own last mile. This quantitative model can be found in Appendix 1 and is implemented and solved in the same way as the model for the UCC situation.

5.1 Simulation of different scenarios

Both quantitative models that have been developed in this study will now be used to run simulations for different scenarios in city distribution. The described business case of Fietskoeriers.nl in Zwolle will be used to retrieve data that will serve as input for the model. The quantitative models will be simulated for the situation where the UCC is on the location that Fietskoeriers.nl has in mind, this means for the developed model that | k | = 1 because there is just one UCC. As described before, this location is close to the city centre but also easy accessible from the highway in Zwolle. This location is already used by the bicycle messenger company as their building. There will only be one UCC used in the simulation and it is assumed that the capacity of this UCC is infinite. In practice, this is not the case because the UCC has a limited capacity. However, the capacity will be enough in the pilot and start-up phase when is expected that only one or a few LSPs are connected to the network. In the future, it is expected that a lot of LSPs will drop their goods off at the UCC in Zwolle, which would make the capacity of the UCC inadequate for the amount of goods that get consolidated. Far before these flow of goods will overwhelm the UCC, research should be done in looking for alternatives like moving to a UCC with more capacity or developing an additional UCC in the city. In this computational study, four different scenarios are investigated which represent several directions that the business case could result in. All these scenarios will be described first, whereafter the results of the simulation per scenario will be presented.

(18)

18 exit of the highway. In the real-life situation, the routings of the LSPs will probably pass by the area around the UCC as the transporter is driving from one location to another, because of its central location between the highway and city centre. Next to this, starting all routings from this location, which is indicated with a red dot in the physical representations, makes comparing the two situations easier.

5.1.1 Scenario 1: UCC network based on the pilot from the business case with single LSP

The first scenario that is simulated represents the pilot that will be run for Fietskoeriers.nl and the LSP that is interested to collaborate in a UCC network. Based on information acquired form these companies can be concluded that the LSP currently delivers goods with a truck in the whole city. Therefore, condensed customer locations have been identified that represent the areas where customers are located. A physical representation of these customer locations, which are represented by yellow dots, can be found in figure 2. It is assumed that in total twenty-six locations have to be visited, two at each condensed customer locations. This will match the pilot, as is expected that on a daily base twenty to thirty packages will be dropped off at the UCC by the LSP. Based on the condensed customer locations, optimal vehicle routings can be developed for the LSP (see figure 3) and for the situation that the UCC of Fietskoeriers.nl is used (see figure 4). The vehicle routings result in information regarding the total kilometres that have been covered in the city distribution, the total distribution costs and the CO2 emissions that are concerned with this. The conventional situation, without UCC, can now be compared to the situation in which the LSP drops off the goods at the UCC of Fietskoeriers.nl. This comparison can be found in table 1.

(19)

19

Figure 3. Routing scenario 1 without UCC Figure 4. Routing scenario 1 with UCC

Performance indicator No UCC UCC Difference

Travelled kilometres 47,78 km 48,98 km +2,5%

Costs city distribution €126,83 €123,40 -2,7%

CO2 emissions 19.601 gram 1.800 gram -90,8%

Table 1. Performance indicators for scenario 1.

5.1.2 Scenario 2: UCC network with multiple and varying LSPs

(20)

20 scenario, six LSPs have to deliver in total sixty packages at the condensed customer locations, three LSPs do this by truck and three do this by van. On average, every LSP delivers ten packages to customers however it is assumed that half of the LSPs deliver more packages and the other half less. The performance indicators for scenario 2a are presented in table 2, for scenario 2b they are presented in table 3. There is no physical representation of the routings included for these scenarios, as the customer locations are the same as in scenario 1. The biggest difference is that extra routings are added in the situation without UCC.

Travelled kilometres 99,15 km 45,85 km -53,7%

Table 2. Performance indicator for scenario 2a.

Table 3. Performance indicator for scenario 2b.

5.1.3 Scenario 3: UCC network with B2B LSPs

(21)

21 the B2B market, most deliveries are done with trucks. Again, two sub scenarios are developed; the first, scenario 3a, in which only three LSPs join the network (with twenty-four deliveries) and the second, scenario 3b, in which both the LSPs and deliveries are doubled (six LSPs deliver sixty packages). For scenario 3a, the routing for the LSPs delivering by themselves can be found in figure 6 and for the UCC network in figure 7. The performance indicators for the two sub scenarios can be found in table 6 and 7.

Figure 5. Customer locations used in scenario 3

(22)

22

Table 6. Performance indicators for scenario 3a.

Table 7. Performance indicators for scenario 3b.

5.1.4 Scenario 4: UCC network with parcel delivery LSPs

(23)

23 city distribution because it was unclear how much these costs are. Therefore, a certain amount of costs should be added to the costs without UCC that are presented in tables 8 and 9, however how much is not clear. This means that the difference in distribution costs that can be found in table 8 and 9 does not represent the actual cost difference of the distribution, which is indicated with a *.

Figure 8. Condensed customer locations used in scenario 4

(24)

24

Costs city distribution €372,13 €456,17 +22,6% *

Table 8. Performance indicators for scenario 4a.

Costs city distribution €559,13 €682,64 +22,1% *

CO2 emissions 13.381 gram 1750 gram -86,9%

Table 9. Performance indicators for scenario 4b.

5.2 Discussion

(25)

25

Performance indicator Smallest difference Biggest difference

Travelled kilometres -47,8% -78,4%

Costs city distribution -5,0% -23,6%

CO2 emissions -85,1% -89,4%

Table 10. Summary of differences in the performance indicators for all scenarios

As the summarizing table presents, a network with a UCC and a White Label Last Mile outperforms the conventional situation for all three performance indicators. For the pilot with the single LSP that joins the UCC network, a reduction is observed for the total costs of the city distribution and for the CO2 emissions. A marginal increase is observed in the kilometres that are driven in the distribution. This is because the capacity of cargo bicycles is limited in comparison to the truck that is currently used, which creates a second routing for the delivery. However, the summarizing table above presents that when more LSPs join the UCC network under different scenarios, reductions are observed for all performance indicators. The pilot will result in a marginal increase in travelled kilometres and a decrease in distribution costs and CO2 emissions. However, the UCC network will become more successful when more LSPs join the network which results in a reduction for all performance indicators.

The reduction in kilometres that are driven to distribute all the goods has a big variety, ranging from almost 50% up to almost 80%. This reduction is dependent on the customer locations and the number of LSPs that join the UCC network. What can be seen in the developed scenarios is that the reduction in travelled kilometres increases when more LSPs join the network. This is an expected result, because all goods are grouped together in the UCC per customer area so the bicycle messenger does not have to visit the whole city but only some specific customer areas. In the conventional situations, the messenger has to drive his truck or van throughout the city as all the goods that need to be delivered are distributed over the city. When more LSPs join the network, it could be possible that the reduction in travelled kilometres could increase even more because the routings in the UCC situation will not change that much while more routings for trucks and vans will be added as these all have to travel throughout the city to do the last mile delivery.

(26)

26 extra process that takes place in the UCC. In the conventional situation, all goods are directly distributed to the customers, where the goods have an extra process of consolidation in the UCC network. This extra process means extra costs, as a location needs to be hired for the UCC and personnel needs to be paid for the unloading of the trucks/vans, consolidating and the loading of cargo bicycles in the UCC. That the distribution costs in the UCC network are still less than in the conventional situation is because the last mile delivery is done by cargo bicycles. The costs of using a cargo bicycle are much lower than the costs of using a truck or a van. This has to deal with two things: (1) a truck or van needs diesel to get powered where a cargo bicycle is powered by human strength and (2) the purchase price of a cargo bicycle is nothing in comparison to the price of a truck or van, which will decrease the depreciation costs.

What needs to be noted is that several assumptions have been made to calculate the total distribution costs, for both the conventional situation as well as the UCC network. Currently, the pilot that will be run between Fietskoeriers.nl and the LSP is prepared and both companies are investigating how much it costs them to perform last mile delivery. Unfortunately, the research of both companies came too late to include in this thesis. Therefore, not a clear representation could have been made for the facility costs for example. This is most clear in scenario 4, where a simulation is run for a parcel delivery LSP. Normally, these LSPs have a hub from where they distribute the packages, however it was not possible to acquire information about this. Therefore, these costs have been excluded in this simulation. Even for the UCC, it is not clear how much the facility costs are because Fietskoeriers.nl already has a facility available in Zwolle which is currently used for the operations they execute now. When the UCC is developed, goods can be dropped off here, however it is not sure how much space this will take at the location.

(27)

27

6. CONCLUSION

In this conclusion, first a summary of this thesis will be presented in which an answer to the developed research question will lead to a conclusion. Second, managerial insights based on this conclusion are discussed, whereafter the limitations of this thesis are presented. Last, recommendations for further research are given which are partly based on the limitations of the thesis.

6.1 Summary

In this thesis, the differences in city distribution have been investigated between the conventional way of last mile deliveries in which every Logistical Service Provider (LSP) executes this themselves and a situation in which an Urban Consolidation Centre (UCC) is combined with a White Label Last Mile. In doing this, the following research question have been created: “What are the benefits of collaboration in a White Label Last Mile delivery setting with use of Urban Consolidation Centers?”. To answer this question, three performance indicators have been introduced which have been investigated: total travelled kilometres, the total costs of city distribution and the emission of CO2. A quantitative model has been developed which was used to run simulations in the computational study. This computational study was based on information coming from a pilot that will be executed by Fietskoeriers.nl, a network that is willing to collaborate with LSPs in a White Label Last Mile setting with UCCs. Several situations are represented in the simulations of LSPs joining the White Label Last Mile network. In this, different configurations are developed regarding the delivery vehicles of the LSPs, the quantities of goods that need to be delivered and for customer locations. All in all, it can be concluded that a White Label Last Mile setting with a UCC outperforms the conventional way of last mile distribution. The total costs of city distribution will almost reduce up to 25%, where the total travelled kilometres and the CO2 emissions will reduce enormously up to almost 80% and 90%. It is expected that the percentages for the travelled kilometres and distribution costs will increase when more LSPs join the network. To answer the research question of this thesis, the benefits of last mile collaboration in a White Label Last Mile delivery setting with use of Urban Consolidation Centres are a large reduction in travelled kilometres, total distribution costs and CO2 emissions.

6.2 Managerial insights

(28)

28 specialized in last mile deliveries. The model was run for a company that uses cargo bicycles for last mile logistics. When these cargo bicycles are replaced by cars or small vans, the reduction in CO2 emissions will decrease to an extent that is equal to the total kilometres travelled. Using electric vans and trucks will prevent this, however these vehicles will be more expensive in their usage as the buying and replacement are much more expensive than cargo bicycles.

What should be noted is that the scale of last mile logistics and deliveries is increasing. In the Netherlands, the total deliveries for both B2C and B2B are increasing every year; for example, in 2016 an increase of 12% was found in comparison to 2015 for package deliveries (Autoriteit Consument & Markt, 2017). This would mean for the network of Fietskoeriers.nl that the amount of goods that need to be delivered will also increase for the coming years.

Another thing that should be considered is, next to the growing market of deliveries, that when deliveries through the UCCs of Fietskoeriers.nl will be successful, more and more LSPs will join in this network. The network will start with a pilot with only one LSP, however more LSPs will only join the network when they are convinced of its benefits. For most of the LSPs, their main concern will be the total costs: when delivering the goods via a network with a UCC is cheaper than doing it by themselves, LSPs will probably decide to join the network. In this, it should be considered that some LSPs will not be willing to join the network as they are afraid of losing brand awareness. Fietskoeriers.nl might possibly become a victim of its own success when lots of LSPs join the network and the capacity of their UCCs become too small. To prevent this, they should always monitor whether the capacity of the UCC is enough and options should be available for creating an extra UCC or switching to a location with more capacity.

6.3 Limitations

(29)

29 included would have made the quantitative model more representative for the real-world situation. However, as the model also needed to be efficient and manageable, it was chosen to use a simplified model to represent the real-world situation.

A second limitation has to do with the data that has been collected for the input of the quantitative model. Currently, there are lots of uncertainties regarding the development of the UCC in Zwolle. The data that has been used to represent this UCC in the quantitative model might not be a correct display of the real-world situation because of this. As the UCC is currently in a starting phase and not feasible because critical parts like the software need to be developed, it is hard to come up with representable data for the quantitative model. When the UCC is feasible and in use, data should be collected which will be much more reliable than the data that is used now. At this moment, it is hard to say how the UCC network will develop and what exactly will happen when it is operational, which could make the data used in this thesis useless as the data could be completely different when the UCC network is developed. Next to this, data from the LSP might not have been a good reflection of their real-world situations. The data used in this study is based on one LSP and because of very busy times they could not provide detailed information regarding their current situation of last mile delivery. Data from them was acquired after a meeting between the LSP and Fietskoeriers.nl took place.

A third limitation of this study is the quantitative model itself. Because of the many constraints that are included, it becomes a large Mixed Integer Programming problem which is hard to deal with for a solver. In this thesis, the optimization solver Gurobi is used which is a software that is widely used by both academics and professionals. Even though this solver is widely used, it has some problems with the exponentially many constraints of the developed quantitative model which makes it not nicely scalable. The Gurobi solver has, just as other solvers, difficulties with solving such large models to optimality. Therefore, a small data set has been used in this study to represent condensed customer locations and some assumptions have been developed which made it possible to solve the model. This limitation seems like a major one for the conclusion of the thesis but should not be regarded as, because solving the model to optimality for large problems will not change the message of this study dramatically.

6.4 Recommendations for further research

(30)

(31)

31

REFERENCES

Allen, J., Browne, M., Woodburn, A. & Leonardi, J. (2012). The Role of Urban Consolidation Centres in Sustainable Freight Transport. Transport Reviews, 32 (4), pp. 473-490.

Arnold, F., Cardenas, I., Sörensen, K. & Dewulf, W. (2018). Simulation of B2C e-commerce distribution in Antwerp using cargo bikes and delivery points. European Transport Research Review, 10 (2), pp. 1-13.

Autoriteit Consument & Markt, (2017). Post- en Pakketten monitor 2016. [pdf] Available at:

https://www.acm.nl/sites/default/files/old_publication/publicaties/17545_Post-en-pakkettenmonitor-2016-08-08-2017.pdf [Accessed 13 June 2018].

Björklund, M., Abrahamsson, M. & Johansson, H. (2017). Critical factors for viable business models for urban consolidation centres. Research in Transportation Economics, 64 (2017). Pp. 36-47. Boyer, K.K., Prud’homme, A.M. & Chung, W. (2009). The last mile challenge: Evaluating the effects

of customer density and delivery window patterns. Journal of Business Logistics, 30 (1), pp. 185-201.

Calco Ambel, C. (2015). Too big to ignore – truck CO2 emissions in 2030. [pdf] Transport & Environment, Available at: https://www.transportenvironment.org/sites/te/files/publications/ 2015%2009%20TE%20Briefing%20Truck%20CO2%20Too%20big%20to%20ignore_FINAL .pdf [Accessed 30 May 2018].

Clausen, U., Geiger, C. & Pöting, M. (2016). Hands-on testing of last mile concepts. Transportation

Research Procedia, 14 (2016), pp. 1533-1542.

Conway, A., Cheng, J., Kamga, C. & Wan, D. (2017). Cargo cycles for local delivery in New York City: Performance and impacts. Research in Transportation Business & Management, 24 (2017), pp. 90-100.

Cruijssen, F., Cools, M. & Dullaert, W. (2007). Horizontal cooperation in logistics: Opportunities and impediments. Transportation Research Part E, 43 (2007), pp. 129-142.

Cruijssen, F., Dullaert, W. & Fleuren, H. (2007). Horizontal Cooperation in Transport and Logistics: A Literature Review. Transportation Journal, 46 (3), pp. 22-39.

De Souza, R., Goh, M., Lau, H.C., Ng, W.S., Tan, P.S. (2014). Collaborative Urban Logistics – Synchronizing the Last Mile. A Singapore Research Perspective. Procedia – Social and

Behavioral Sciences, 125 (2014), pp. 422-431.

Ec.europa.eu, (2017). Reducing CO2 emissions from vans. [online] Available at: https://ec.europa.eu/clima/policies/transport/vehicles/vans [Accessed 30 May 2018].

(32)

32 Gruber, J., Kihm, A. & Lenz, B. (2014). A new vehicle for urban freight? An ex-ante evaluation of electric cargo bikes in courier services. Research in Transportation Business & Management, 11 (2014), pp. 53-62.

Van Heeswijk, W., Larsen, R. & Larsen, A. (2017). An urban consolidation center in the city of Copenhagen: a simulation study. BETA working papers, 523, pp. 1-34.

Iwan, S., Kijewska, K. & Lemke, J. (2016). Analysis of parcel lockers’ efficiency as the last mile delivery solution – the results of the research in Poland. Transportation Research Procedia, 12 (2016), pp. 644-655.

Janjevic, M., Lebeau, P., Ballé Ndiaye, A., Macharis, C., Van Mierlo, J. & Nsamzinshuti, A. (2016). Strategic scenarios for sustainable urban distribution in the Brussels-capital region using urban consolidation centres. Transportation Research Procedia, 12 (2016), pp. 598-612.

Kin, B., Verlinde, S., Van Lier, T. & Macharis, C. (2016). Is there life after subsidy for an urban consolidation centre? An investigation of the total costs and benefits of a privately-initiated concept. Transportation Research Procedia, 12 (2016), pp. 357-369.

Lia, F., Nocerino, R., Bresciani, C., Colorni, A. & Luè, A. (2014). Promotion of E-bikes for delivery of goods in European urban areas: an Italian case study. Transport Research Arena, 2014 (Paris), pp. 1-10.

Lindawati, van Schagen, J., Goh, M. & de Souza, R. (2014). Collaboration in urban logistics: motivations and barriers. International Journal of Urban Sciences, 18 (2), pp. 278-290.

Maes, J. & Vanelslander, T. (2012). The use of bicycle messengers in the logistics chain, concepts further revised. Procedia – Social and Behavioral Sciences, 39 (2012), pp. 409-423.

Mason, R., Lalwani, C. & Boughton, R. Combining vertical and horizontal collaboration for transport optimisation. Supply Chain Management: An International Journal, 12 (3), pp. 187-199. Melo, S. & Baptista, P. (2017). Evaluating the impacts of using cargo cycles on urban logistics:

integrating traffic, environmental and operational boundaries. European Transportation

Research Review, 9 (30), pp. 1-10.

Montoya-Torres, J.R., Muñoz-Villamizar, A. & Vega-Mejía, C.A. (2016). On the impact of collaborative strategies for goods delivery in city logistics. Production Planning & Control, 27 (6), pp. 443-455.

Muñoz-Villamizar, A., Montoya-Torres, J.R. & Vega-Mejía, C.A. (2015). Non-Collaborative versus Collaborative Last-Mile Delivery in Urban Systems with Stochastic Demands. Procedia CIRP, 30 (2015), pp. 263-268.

Nordtømme, M.E., Bjerkan, K.Y. & Sund, A.B. (2015). Barriers to urban freight policy implementation: The case of urban consolidation center in Oslo. Transport Policy, 44 (2015), pp. 179-186. Park, H., Park, D. & Jeong, I.J. (2016). An effects analysis of logistics collaboration in last-mile

(33)

33 Punakivi, M., Yrjölä, H. & Holmström, J. (2001). Solving the last mile issue: reception box or delivery

box? International Journal of Physical Distribution & Logistics, 31 (6), pp. 427-439.

Simoni, M.D., Bujanovic, P., Boyles, S.D. & Kuatnoglu, E. (2018). Urban consolidation solutions for parcel delivery considering location, fleet and route choice. Case Studies on Transport Policy, 6 (2018), pp. 112-124.

Slabinac, M. (2015). Innovative solutions for a “last-mile” delivery – A European experience. 15th

international scientific conference Business Logistics in Modern Management, 2015, pp.

111-129.

Taniguchi, E. (2014). Concepts of city logistics for sustainable and liveable cities. Procedia – Social

and Behavioral Sciences, 151 (2014), pp. 310-317.

Taniguchi, E. & Van Der Heijden, R.E.C.M. (2000). An evaluation methodology for city logistics.

Transport Reviews, 20(1), pp. 65-90.

Vanovermeire, C., Sörensen, K., Van Breedam, A., Vannieuwenhuyse, B. & Verstrepen, S. (2014). Horizontal logistics collaboration: decreasing costs through flexibility and an adequate cost allocation strategy. International Journal of Logistics: Research and Applications, 17 (4), pp. 339-355.

Verlinde, S., Macharis, C. & Witlox, F. (2012). How to consolidate urban flows of goods without setting up an urban consolidate centre? Procedia – Social and Behavioral Sciences, 39 (2012), pp. 687-701.

Wrighton, S-. & Reiter, K. (2016). CycleLogistics – moving Europe forward! Transportation Research

(34)

34

APPENDIX

Appendix 1. Quantitative model for conventional situation without UCC

This model is representable for one Logistical Service Provider (LSP). When a simulation is run in which multiple LSPs are included, this model can be copied for all LSPs. The decision variables of the model are:

X(k) corresponding to 1 if the UCC is open and 0 otherwise

Z(hij) corresponding to 1 if node j is served after node i by using vehicle h Y(h) corresponding to 1 if vehicle h departs from depot.

The sets of the model are:

H set of available vehicles

I set of demand nodes

K candidate facility for depot N I ∪ J set of all nodes

The inputs of the model are: l(i) demand at customer node i f(k) fixed costs of opening depot k h(k) operational costs of depot k

c(hij) transportation costs between node i and node j using vehicle h g(h) fixed costs of using vehicle h

u(h) capacity of vehicle h