ISTANBUL TECHNICAL UNIVERSITY  GRADUATE SCHOOL Ph.D. THESIS NOVEMBER 2021 INTEGRATED MANAGEMENT OF MIXED FLEETS OF ELECTRIC AND CONVENTIONAL VEHICLES UNDER ROUTING CONSIDERATIONS Reema Talab Hammad AL-DALAIN Department of Management Engineering Management Engineering Programme Department of Management Engineering Management Engineering Programme NOVEMBER 2021 ISTANBUL TECHNICAL UNIVERSITY  GRADUATE SCHOOL INTEGRATED MANAGEMENT OF MIXED FLEETS OF ELECTRIC AND CONVENTIONAL VEHICLES UNDER ROUTING CONSIDERATIONS Ph.D. THESIS Reema Talab Hammad AL-DALAIN (507162014) Thesis Advisor: Prof. Dr. Dilay ÇELEBİ GONIDIS İşletme Mühendisliği Anabilim Dalı İşletme Mühendisliği Programı KASIM 2021 İSTANBUL TEKNİK ÜNİVERSİTESİ  LİSANSÜSTÜ EĞİTİM ENSTİTÜSÜ ROTALAMA ETMENLERİ ALTINDA ELEKTRİKLİ VE KONVANSİYONEL ARAÇLARIN KARMA FİLOLARININ BÜTÜNLEŞİK YÖNETİMİ DOKTORA TEZİ Reema Talab Hammad AL-DALAIN (507162014) Tez Danışmanı: Prof. Dr. Dilay ÇELEBİ GONIDIS v Thesis Advisor : Prof. Dr. Dilay ÇELEBİ GONIDIS .............................. Istanbul Technical University Reema Talab Hammad AL-DAL’AIN, a Ph.D. student of İTU Graduate School of Science Engineering and Technology, student ID 507162014 successfully defended the thesis/dissertation entitled “INTEGRATED MANAGEMENT OF MIXED FLEETS OF ELECTRIC AND CONVENTIONAL VEHICLES UNDER ROUTING CONSIDERATIONS”, which she prepared after fulfilling the requirements specified in the associated legislations, before the jury whose signatures are below. Jury Members: Assoc. Prof. Dr. Özgür KABAK ............................. Istanbul Technical University Assoc.Prof. Dr. Yeliz EKİNCİ .............................. Bilgi University Date of Submission : 19 October 2021 Date of Defense : 12 November 2021 Prof. Dr. Hatice CAMGÖZ AKDAĞ.............................. Istanbul Technical University Prof. Dr. Gül TEKİN TEMUR .............................. Bahçeşehir University vi vii To my family viii ix FOREWORD This journey would not have been as amazing without the fascinating people I met along the road. I would first and foremost like to offer my deepest gratitude to my supervisor Professor Dilay ÇELEBİ GONIDIS for the opportunity to carry out this research project and for her tremendous encouragement, patience, motivation, and immense knowledge during my PhD. I would like to thank my research members, Doç.Dr. Yeliz EKİNCİ and Doç. Dr. Özgür KABAK, who shared their experiences and ideas, also for their many insightful recommendations and continuous support throughout my PhD study. I am forever indebted to my parents, my brothers, and my grandparents for their unconditional love and support throughout my life. October 2021 Reema AL-DALAIN (Management engineering) x xi TABLE OF CONTENTS Page FOREWORD ............................................................................................................. ix TABLE OF CONTENTS .......................................................................................... xi ABBREVIATIONS ................................................................................................. xiii SYMBOLS ................................................................................................................ xv LIST OF TABLES ................................................................................................. xvii LIST OF FIGURES ................................................................................................ xix SUMMARY ............................................................................................................. xxi ÖZET .............................................................................................................. xxiii INTRODUCTION .................................................................................................. 1 Background ....................................................................................................... 1 Scope of the Problem ........................................................................................ 6 The Purpose of the Thesis ................................................................................. 7 The Contributions .............................................................................................. 7 Research Significance ....................................................................................... 8 Thesis Structure ................................................................................................. 9 Key Terms ....................................................................................................... 10 THE USE OF ELECTRIC VEHICLES IN URBAN FREIGHT .................... 13 Fleet Management in Urban Freight ............................................................... 13 Sustainability in Urban Freight ....................................................................... 16 2.2.1 Urban freight transport strategies ............................................................ 18 2.2.2 SWOT analysis of electric vehicles in urban freight ............................... 20 2.2.3 Studies and experiences ........................................................................... 22 Vehicle Routing Problem ................................................................................ 25 2.3.1 Vehicle routing problem classifications .................................................. 27 2.3.2 Exact and heuristic method for the VRP ................................................. 31 Literature Review ............................................................................................ 33 2.4.1 The mixed vehicle routing problem with time window........................... 33 2.4.2 Fleet replacement model .......................................................................... 37 2.4.3 Literature gap ........................................................................................... 41 METHODOLOGY AND MATHEMATICAL MODELS ............................... 45 Methodology ................................................................................................... 45 Mathematical Models ...................................................................................... 47 3.2.1 Replacement model ................................................................................. 47 3.2.2 MVRPTW ................................................................................................ 51 COMPUTATIONAL TEST ................................................................................ 57 The Economic Calculations ............................................................................. 57 4.1.1 Salvage value ........................................................................................... 58 4.1.2 Operating cost .......................................................................................... 59 4.1.3 CO2 emissions ......................................................................................... 60 4.1.4 The maintenance and repair costs ............................................................ 61 4.1.5 Energy taxes............................................................................................. 62 xii 4.1.6 Discount Rate ........................................................................................... 62 Results and Discussion .................................................................................... 62 4.2.1 Competitive analysis ................................................................................ 64 4.2.2 Sensitivity analysis .................................................................................. 66 CASE STUDY ....................................................................................................... 69 Background ...................................................................................................... 69 Description and Modelling Approach ............................................................. 71 Sensitivity Analysis ......................................................................................... 74 CONCLUSION ..................................................................................................... 83 Summary .......................................................................................................... 83 Conclusion ....................................................................................................... 84 Limitations ....................................................................................................... 86 Future Work ..................................................................................................... 86 REFERENCES ......................................................................................................... 89 CURRICULUM VITAE ........................................................................................ 101 xiii ABBREVIATIONS AC : Alternative Current BBA : Branch and Bound Algorithm CO2 : Carbon Dioxide CV : Conventional Vehicles CVRP : Capacitated Vehicle Routing Problem DC : Direct Current DCVRP : Distance Capacity Vehicle Routing Problem DVRP : Distance Vehicle Routing Problem EAFO : European Alternative Fuels Observatory ECVs : Electric Commercial Vehicles EPA : Environmental Protection Agency EV : Electric Vehicles EVI : Electric Vehicles Initiative GA : Genetic Algorithm GHG : Greenhouse Gas G-VRP : Green Vehicle Routing Problem HDT : Heavy-Duty Truck HFVRP : Heterogeneous Fleet Vehicle Routing Problem HVRP : Heterogeneous Vehicle Routing Problem ICCVs : Internal Combustion Commercial Vehicles IEA : International Energy Agency Kg : Kilogram KWh : Kilowatt Hour LCV : Light Commercial Vehicle LNG : Liquefied Natural Gas MBFM : Mixed Bus Fleet Management MDT : Medium Commercial Vehicle Mpg : Miles per gallon MVRP : Multi-Depot Vehicle Routing Problem MVRPTW : Mixed Vehicle Routing Problem With Time Window NLABC : New Life Additional Benefit-Cost PVRP : Periodic Vehicle Routing Problem SA : Simulated Annealing SDVRP : Split Delivery Vehicle Routing Problem SOC : State of Charge SVRP : Stochastic Vehicle Routing Problem TCO : Total Cost of Ownership TDVRP : Time Dependent Vehicle Routing Problem TSP : Travelling Salesman Problem UCC : Urban Consolidation Centers UK : United Kingdom VRP : Vehicle Routing Problem VRPB : Vehicle Routing Problem With Backhaul xiv VRPTW : Vehicle Routing Problem With Time Window xv SYMBOLS 𝒂𝒊, 𝒃𝒊 : The time window for each customer i. 𝑨𝒌 : Maximum age of vehicle. 𝑩𝒕 : Budget at the beginning of year t. 𝑪 : Annual working days. 𝑫𝒕 : Annual miles that needed to be travelled at year t. 𝒆𝒇𝒌 : CO2 emissions cost of age type k vehicle per mile ec : CO2 emissions penalty er : Electricity inflation rate 𝑭𝒇𝒌 : Fuel tax cost of age type k vehicle per mile. FV : Future value. 𝒅𝒊𝒋 : The distance between customers i and customer j. 𝒅𝒓 : Discount rate. 𝑮𝒌 : Vehicle capacity of type k. g : Gallon 𝑯𝒌 : The operational range of vehicles of type k. 𝒉𝒇𝒌 : Initial number of age , type k vehicle at time zero. I : set of customers vertices. K : Type of truck. 𝑳𝒊𝒋 : The travel time between customers i and customer j 𝑴𝒕𝒌 : Number of available vehicles of type k at time t. n : Lifetime of asset 𝒐𝒌 : Operational cost 𝑷𝒕𝒌 : The number of new type-k vehicles purchased at time zero PV : Present value 𝑸𝒇𝒌 : Maintenance cost of age type k vehicle per mile 𝒒𝒊 : Customer’s demand. 𝑹𝒇𝒕𝒌 : The number of age- ƒ, type-k vehicles salvage at the end of year t, r : Annual depreciation rate. 𝑺𝒊 : Service time of each customer. 𝒔𝒇𝒌 : The salvage value of an age type-k vehicle t : Time period. T : Maximum time. 𝑼𝒊𝒌 : The load of vehicle k after visiting customer i. 𝒖𝒇𝒌 : Utilization of age , type k vehicle. 𝒗𝒌 : The purchasing cost of a type k truck in dollars. 𝑾𝒇𝒕𝒌 : The number of age- ƒ, type-k vehicles used in year t, 𝑿𝒊𝒋𝒌 : 𝑋𝑖𝑗𝑘 = 1, if a vehicle type k traveling from i to j, OW 𝑋𝑖𝑗𝑘 = 0 𝒀𝒊𝒋 : The vehicle load from the customer i to j. 𝒇 : Age of vehicles. 𝒇𝒓 : Fuel inflation rate ƒ ƒ ƒ ƒ ƒ ƒ https://www.abbreviations.com/term/1388682 xvi 𝒍 : Prime lending rate 𝝆 : Price index 𝝉𝒊 : The start time of each customer service. µ : Very large number. % : Percentage. $ : USD Dollar xvii LIST OF TABLES Page Table 1.1: The barriers of using electric trucks in urban freight. ............................... 5 Table 2.1 : The impacts of urban freight transport. ................................................... 16 Table 2.2 : Illustrates some of the strategies that have been applied in different cities. ................................................................................................................... 20 Table 2.3 : SWOT of electric vehicles in urban freight. ........................................... 21 Table 4.1: Vehicle’s specifications. .......................................................................... 58 Table 4.2: The salvage value for electric and conventional vehicles over the age of vehicles. ..................................................................................................... 59 Table 4.3: The Energy consumptions and costs for both types of vehicles. ............. 59 Table 4.4: The maintenance costs for electric and conventional vehicles over the age of vehicles. ................................................................................................. 61 Table 4.5: Results of MVRPTW model for different fleet compositions. ................ 63 Table 4.6: Computational results of model and our proposed model. ................... 65 Table 4.7: Computational results of the proposed model with CO2 cost equal to zero. ................................................................................................................... 67 Table 5.1: The purchase costs and Energy prices for both types of vehicles in Turkey the market. ................................................................................................. 72 Table 5.2: Results of MVRPTW model for different fleet compositions. ................ 72 Table 5.3: The percentage of used vehicles from both types for different CO2 emissions costs. ......................................................................................... 75 Table 5.4: The percentage of used vehicles from both types for different maintenance costs. ..................................................................................... 77 Table 5.5: The percentage of used vehicles from both types for different depreciation rates. ........................................................................................................... 78 Table 5.6: Computational results of Feng and Figliozzi (2013) model and our proposed for the average number of used vehicles without subsidies. ...... 80 xviii xix LIST OF FIGURES Page Figure 1.1: The number of units of electric trucks sold from each type . ................... 3 Figure 1.2: Electric truck registrations by region, 2015-2020 . .................................. 4 Figure 1.3: Dissertation organization . ...................................................................... 11 Figure 2.1: The principle of sustainable urban freight transport system . ................ 17 Figure 2.2: An example of vehicle routing problem presentation . .......................... 25 Figure 2.3: Vehicle routing problem variants . ......................................................... 28 Figure 2.4: Classification of Vehicle Routing Problem (VRP) models according to their degree of realism. ............................................................................ 31 Figure 2.5: Solution algorithms for VRP. ................................................................. 32 Figure 2.6: Illustrative example of different solutions for VRP with different sets of vehicles. ................................................................................................... 43 Figure 3.1: The general structure of the proposed approach. .................................... 46 Figure 4.1: Vehicles cost analysis. ............................................................................ 57 Figure 4.2: A graphical representation of C and R problems. .................................. 63 Figure 4.3: The fleet compositions over the planning time horizon of each type of vehicle in each year. ................................................................................ 64 Figure 4.4: The percentage of purchased vehicles for both types for different states of charge over the planning time horizon................................................ 67 Figure 5.1: Map of Istanbul municipality ................................................................. 70 Figure 5.2: Customer’s locations. ............................................................................. 71 Figure 5.3: Fleet composition for both types of vehicles over the planning time horizon when CO2 emissions cost equal zero. ........................................ 73 Figure 5.4: Fleet composition for both types of vehicles over the planning time horizon when CO2 emissions cost equal $28/ton. ................................... 73 Figure 5.5: The CO2 emissions produced in Turkey per capita. ............................... 74 Figure 5.6: The Overall GHG emissions by sector in Turkey. ................................. 75 Figure 5.7: The number of used vehicles from both types for different CO2 emissions prices....................................................................................... 76 Figure 5.8: The number of used vehicles from both types for different maintenance costs. ........................................................................................................ 77 Figure 5.9: Illustration of straight line method for depreciation rate. ....................... 78 Figure 5.10: The percentage of used vehicles from both types for different depreciation rate values. .......................................................................... 79 Figure 5.11: The percentage of vehicles used over the planning time horizon for different purchasing costs with and without subsidies. ........................... 79 xx xxi INTEGRATED MANAGEMENT OF MIXED FLEETS OF ELECTRIC AND CONVENTIONAL VEHICLES UNDER ROUTING CONSIDERATIONS SUMMARY Planning and managing a set of freight delivery vehicles to minimize the total costs have always been a pressing issue in the transportation industry. However, recently and due to the environmental challenges, minimizing the greenhouse gas emissions in freight transport has become just as important as minimizing costs. In the race of dominance, the conventional vehicles used to overtake the electric vehicles in many aspects such as acquisition cost and refueling-recharging time, however effective management of electric vehicles in freight operations along with efficient cost planning throughout their lifecycle is expected to increase their adoption rate in urban freight. For heavy and medium duty vehicles there is uncertainty attached to the adoption rate due to limited driving range and charging battery, where companies might face losses of profit if vehicles needed to stop many times and for long periods during the day. Therefore, merging the electric vehicles with conventional vehicles in urban freight fleets can help to overcome the additional constraints induced by the specific characteristics of electric vehicles. A common practice for fleet mixed decisions is to use techniques that have been developed for managing conventional vehicles. These techniques may fall short in managing fleets with electric vehicles effectively. In this thesis, an attempt has been made to present a new perspective to the problem of managing a mixed fleet of electric and conventional vehicles in urban freight by integrating two models; fleet size and mixed vehicle routing problem with time window and replacement model. Our study was motivated by the recent practice of involving alternative vehicles in existing fleets as a response to the recent global advocates of minimizing the greenhouse gas emissions generated from using conventional vehicles in the transportation sector. We first consider a fleet size and mixed vehicle routing problem with time window to minimize the operational cost for different fleet compositions of electric and conventional vehicles, in which many constraints such as time window, limited distance range, and capacity are considered. Then we feed the results into a replacement model to find the best fleet mix policy. The replacement model is used to decide the optimal time periods to replace the used vehicles with a new one, taking into consideration different economic costs such as: annual discount rate, and energy prices for both fuel and electricity, along with the initial fleet compositions, and the planning time horizon. The methodology is implemented on generated and real life problems. Results from the computational experiments show that efficient planning of electric vehicles in urban operations can increase their presence compared to conventional vehicles. This is an important insight, since it shows that the adoption rate of electric vehicles in urban freight fleets may increase with better planning techniques, related to electric xxii vehicles or with the increased experience of operational managers with electric vehicles. xxiii ROTALAMA ETMENLERİ ALTINDA ELEKTRİKLİ VE KONVANSİYONEL ARAÇLARIN KARMA FİLOLARININ BÜTÜNLEŞİK YÖNETİMİ ÖZET Kentselleşme ile birlikte sera gazı salınımı ve enerji kaynaklarındaki azalma çok büyük problemler haline gelmiştir. Şehir nakliyeciliğinde elektrikli kullanımı, bu çok az miktarda karbondioksit salınımı, daha az gürültü, daha az kirlilik sağlamaları ve yenilenebilir enerji kaynakları ile çalışabilmelerinden dolayı bu problemlere bir çözüm olabilir. Elektrikli taşıtların bu avantajlarına ragmen, kısa sürüş menzili, yüksek fiyatları ve batarya maliyetlerindenn dolayı kentsel nakliyecliğinde kullanımı halen çok geride kalmıştır. Ayrıca bir çok ülkede şarj istasyonlarının olmaması elektrikli taşıtların bu alanda gelişmesi yönünde büyük engel teşkil etmektedir. Konvansiyonel taşıtlar elektrikli taşıtlar göre bir çok yönde daha üstünken, son zamanlarda elektrikli taşıtların yüksek teknoloji ile gelişitirilmiş olması, piyasaya sunulması ve çevresel problemlerdeki sorunların azaltılması, elektrikli taşıtları daha rekabetçi bir konuma getirmiştir. Elektrikli taşıtların etkili bir şekilde yönetimi ve ürün ömrü için etkin maliyet planlamaları ile birlikte şehir içi nakliyede adaptasyonunu arttırması beklenmeketdir. Dünyadaki bir çok firma çevresel problemlerden dolayı konvansiyonel taşıtlarla birlikte elektriksel taşıtların etkili bir şekilde yönetimi ve ürün ömrü için etkin maliyet planlamaları ile birlikte şehir içi nakliyede adaptasyonunu arttırması beklenmeketdir. Dünyadaki bir çok firma çevresel problemlerden dolayı konvansiyonel taşıtlarla birlikte elektriksel taşıtları da kullanmaya başlamıştır. Konvansiyonel taşıtların elektirikli taşıtlarla bilirkte kullanımı elektriksel taşıtların kısa sürüş menzili ve şarj edilme gibi problemlerinin ortadan kaldırılma konusunda etkili olmuştur. Filo için kullanılan yaygın bir uygulama geleneksel taşıtların geliştirilmiş teknikleriyle birlikte kullanımı ile yönetmektir. Bu karma teknikler elektriksel taşıtların kullanılmaya başlaması ile azalmıştır. Bu tez kapsamında, elektriksel taşıtların şehir içi nakliyede kullanılması alanında adaptasyon düzeyi, operasyonal planlamaların filo yatırımlarındaki etkisine bakılarak araştırılmış ve ayrıca güzergah birleştirilmesi, filo bileşimlerdeki düzenlemelerin elektrikli taşıtların şehir içi nakliyedeki üzerine etkisi incelenmiştir. Elektrikli karma filo modelleri ile ilgili daha önceki çalışmalar taşıtların yaşına bağlı olarak sabit nakliye maliyetlerini varsayar. Bu varsayım yenileme ilkesini belirlemeyi kolaylaştırmasına ragmen, gerçek yaşam durumları için belirsiz ve gerçekçi değildir çünkü rota kısıtlamaları operasyonel maliyette çok önemli bir yer tutmaktadır Problemi daha net bir şekilde tanımlayabilmek için, şehir içi nakliyede elektrikli ve konvansiyonel taşıtlar için araç rotalama ve filo bileşim kararlarını birlikte ele alan yeni bir entegre metotla en iyi taşıt yenileme ilkesi belirlenmiştir. Bu modelde operasyonel maliyetler Zaman Pencereli Karma Araç Rotalama problemlerine (ZPKAR) göre hesaplanmıştır. Bu hesaplamada, taşıt yenileme modelindeki olası xxiv tüm filo bileşimleri için zaman penceresi, mesafe kısıtı ve kasapasite gibi sınırlamalar göz önüne alınarak hesaplama yapılmıştır. Elde edilen sonuçlar yenileme modelinde giriş verisi olarak kullanılmıştır. Yenileme modeli, kullanılmış taşıtların değişimi için en uygun zamanı ekonomik maliyetler, yıllık amortisman oranı, elektrik ve yakıt harcamaları gibi faktörlere göre belirlemektedir. Bununla birlikte başlangıç filo bileşimleri, planlama süresi, bütçe, taşıt başına katedilen mil, yıllık taşıt sayısı gibi etmenler de göz önünde bulundurulmaktadır. Yenileme modelindeki taşıt özelliklerine ek olarak, taşıtın maksimum yaşı, edinim maliyeti, tamir onarım maliyetleri, bakım maliyetleri, karbondioksit gaz salınımı, vergiler ve taşıt cinsi gibi parametreler de yer almaktadır. Tüm bu kriterler yenileme modelinde en önemli unsuru tahmin etmemizi sağlamaktadır. Bu unsur ise, tüm taşıtlar için ZPKAR tarafından belirlenen rotalar sonucu ortaya çıkan yıllık katedilen mesafedir. Geliştirdiğimiz model elektrikli ve konvansiyonel taşıtların karma filo performansları için yapılan klasik değerlendirmenin ötesinde bir analiz olanağı sunmaktadır. Bu sayede farklı karma filo bileşimlerinin, operasyonel aralıkların ve maliyetlerin filo karma ilkesine etkisi analiz edilebilmektedir. Önerilen yöntem gerçek veya kuramsal hayat problemlerine uygulanabilmektedir. Hesaplamalı deney sonuçlarımıza göre, etkili bir rota planlama ile elektrikli taşıtların şehiriçi yük taşımacılığında kullanımı konvansiyel taşıtlara kıyasla artmaktadır. Bu çok önemli bir sonuçtur, çünkü etkili bir planlama ile elektrikli taşıtların şehiriçi yük taşımacılığına adaptasyonun mevcut duruma göre artırılabileceğini göstermektedir. Önerilen yöntem herhangi bir karma filo planlama problem için doğrudan kullanılabilecektir. Fakat yöntemin çok karmaşık hesaplamalı bir yapıya sahip olmasından dolayı, problemin boyutu müşteri sayısı yönünden arttıkça, tam bir çözüm sunması çok zordur. Ayrıca araç rotalama modeli, aracın şarj için depoya dönmesi durumunda bazı modifikasyonlar gerektirecektir. Elektrikli taşıtların adaptasyon oranına etkisini görebilmek için sonuçların genellenmesi gereklidir ve bu konuda başka çalışmalar gerekmektedşr. Bunun için taşıt özellikleri ve sera gazı salınım maliyetleri de göz önünde bulundurulmalıdır. Ayrıca elektrikli taşıt endüstrisinde önemli teknolojik ilerlemeden dolayı ve yakıtlardaki sabit değişime göre, şu anki mevcut sonuçlar gelecekte güncellenecektir. Duyarlılık analizi sonuçlarına göre, filo bileşiminde daha çok elektrikli taşıt kullanabilmesi için, bataryaların ömür süresi artırılmalıdır, bu da çeşitli optimizasyon teknikleri ile sağlanabilir. Bununla birlikte bataryaların ömür süresi ile ilgili bir çok belirsizlik halen mevcuttur. Gelecekteki çalışmalar için, model şarj olanakları eklenerek genişletilebilir, çünkü elektrkli taşıtlar rota esnasında tamamen veya kısmi olarak şarj edilebilirler. Sonrasında farklı şarj stratejilerinin batarya ömür süreleri üzerindeki etkisi analiz edilebilir ve bundan dolayı elektrkli taşıtlar daha rekabetçi hale getirilebilir. Karbondioksit gaz salınım maliyetleri ve batarya değişim seçenekleri ile gözlemlerimize göre geliştirdiğimiz entegre model temel filo yönetim modeline kıyasla elektrikli taşıtların adaptasyon oranın arttığını söylemektedir. Ayrıca önerdiğimiz model konvansiyonel ve elektrikli taşıt maliyetlerinin birbirine çok yakın olduğu durumlarda daha etkili olacaktır. Bununla birlikte güvenilir sonuçlar için gelecekte daha çok örneklem üzerine çalışılmalıdır. Şehir nakliyeciliği kapsamlı bir konu olup bir çok faktör, dikkate alınacak hususlar ve problemler içerir. Son zamanlarda, elektrikli taşıtların kullanımı yoğun ilgli görmüştür, yalnız bu taşıtların şehir nakliyeciliğinde kullanımı için adaptasyonun arttırılması ile ilgili daha çok araştırmaya ihtiyaç vardır. Gelecek çalışmalarda şarj konusunu da kapsayan modellerim önemi artacaktır. xxv Modelleme yönünden ise, stokastik seyahat süresi ve hizmet süresi gibi dinamik parametreler de ele alınabilir. Bu sayede dinamik filo boyutu ve karma araç rota problemleri gelecekte çalışılabilir. xxvi 1 INTRODUCTION This introductory chapter presents the background, the scope, and the purpose of this thesis. In addition, the importance of the main topic “the impact of effective management on the adoption rate of electric vehicles in a fleet composed of electric and conventional vehicles in urban freight” is illustrated. The main contributions of this thesis are also mentioned, followed by the outline of this thesis and the key terms. Background The greenhouse gas emissions have a direct contribution in climate changes and global warming problems. It has been reported that approximately 24% of direct CO2 emissions in 2019 were caused by the transportation sector IEA (2019), and the percentage is expected to increase due to the fast growth of population along with the new online shopping behaviors, and home delivery. This drives many researchers to characterize the current logistic system as unsustainable from three different perspectives; economical, environmental, and social (Montreuil et al, 2012). To address this issue, efforts have been made to encourage the transportation industry to adopt decarbonization strategies to cope with the growth of the environmental challenges. Increasing energy efficiency, reducing the intensity of emissions, and switching from fuel to electricity are among the major decarbonization opportunities in urban areas (Quigley, 2019). Therefore, presenting genuine alternatives for the use of conventional vehicles within urban areas can help to move toward sustainability. Electric vehicles (EVs) are a good candidate to reduce the greenhouse gas (GHG) emissions in the transportation sector, since they have the ability to be charged by renewable energy; therefore, they produce almost zero emissions. Over the last decades, a lot of effort has been put into increasing the adoption rate of EVs in urban freight distribution as a response to the environmental challenges. However, it remains limited and mostly deployed as light commercial vehicles ((Bunsen et al 2019), where in Europe the light truck registration in 2020 exceeded 37000 units, 2 while in China it was more than 3400 units. For the rest of the world the registration of light electric trucks was around 19000 units (IEA, 2021). Besides being eco-friendly, electric vehicles can improve the efficiency of the transportation sector, as the cost per mile for electric vehicles is significantly lower than that of conventional vehicles. In addition, electric vehicles have fewer moving components compared to conventional vehicles, meaning that the maintenance and service will most likely be lower, beside the fast rate of growth and market penetration (Rahimi and Davoudi, 2018), also they can produce less noise (Teoh et al, 2018). Despite all those advantages, the adoption rate of electric vehicles in urban freight distribution faces many challenges such as limited driving range, high purchase and battery costs (Taefi et al, 2016), beside the absence of charging infrastructure (Ajanovic and Haas, 2016), and time required to recharge (Amirhosseini and Hosseini, 2018), along with technical and market challenges related to user’s acceptance, safety regarding battery technology, and performance issues (Wu et al, 2017). Charging stations continue to have a major impact on the presence of electric vehicles in urban freight. European cities invest highly in public charging of electric vehicles. Oslo, Paris, London, and Amsterdam reached about 4,300, 4,700, 5,800, and 9,100 charge points, respectively in 2019 (Hall and Lutsey, 2020). South Korea’s government has planned to deploy 10,000 fast chargers by 2022 (Research and Market, 2020). Similarly, India has targeted to install 2,700 charging stations by 2023 inside cities with more than 4 million residents (Research and Market, 2020). As a result, the electrification of medium and heavy-duty trucks increasingly gained attention as they were promoted to be the ideal solution to decrease the greenhouse gas emissions in urban freight distribution. Different governments have presented ambitious policies to increase the adoption of electric vehicles in different sectors including the development of new charging methods and infrastructure. China’s government is among the governments that has made active progress in developing government policies and regulation in subsidies for both electric vehicles and charging stations to increase the market share of electric vehicles (Song et al, 2020). The urban freight transport system mainly concerns business-to-business delivery between different warehouses, shops, retailers etc. Due to the increase in the population and online shopping behaviors, the business-to-consumer deliveries have 3 also increased. This results in a growing number of small deliveries, which are usually made by vans and cargo motorcycles. This is the costly, most polluting, and inefficient part of the transportation system. Electric vehicles may potentially replace these deliveries, as they are more suitable in short-distance transport for small freight volumes inside cities, which may reduce the environmental challenges related to the use of conventional vehicles. Mirhedayatian and Yan (2018) investigate the policies supporting EVs in urban freight transport by establishing a theoretical framework combining an optimization model with economic analysis to evaluate individual company’s actions in response to policies for electric vehicles. According to the authors, there are three main policies: purchase subsidies, low-emission or congestion zones, and vehicle tax exemptions. Quak et al. (2016) evaluate the feasibility of using electric powered vehicles in city logistics practices from a carrier's perspective. The authors argue that technological performance are among the most important factors, along with the limited production and availability of electric vehicles, more specifically the heavy duty electric freight vehicles. Moreover, the development of the technology is necessary for further optimization of EFVs integration into daily practices of transport operators. Figure 1.1 shows the number of electric trucks sold from different types of trucks in 2018. As the figure shows, the heavy-duty trucks have a small share of the market, unlike the light- and medium-duty trucks, which make up 98% of the global battery electric trucks, where more than 160,000 units sold from light duty trucks in 2018. (Scriven, 2019). Figure 1.1: The number of units of electric trucks sold from each type (Scriven, 2019). 4 For more details, Figure 1.2 shows the electric truck registration for the biggest market; China, Europe, and the United States from 2015 to 2020. As the figure indicates, China dominates the category, with more than six thousand new registrations in 2020. In Europe, the electric truck registrations increase to reach 23%, which is around 450 vehicles, while in the United States increased to 240 vehicles (IEA, 2021). Figure 1.2: Electric truck registrations by region, 2015-2020 (IEA, 2021). Different companies start producing electric vehicles that vary between light, medium, and heavy duty, such as delivery and pickup trucks, garbage trucks, buses, electric cargo motorcycles and many more. In 2020 there were around 30 medium- duty electrified models and 21 heavy-duty models offered in the market for sale worldwide (Lilley, 2020). Those numbers are expected to increase in the upcoming years, as the Electric Vehicles Initiative (EVI) set a goal to reach 30% market share for EVs by 2030 (including cars, buses and trucks) in order to meet the Paris Agreement targets (IEA,2020). Several companies announce the production of electric heavy-duty trucks including: big rigs, semi-trucks, and delivery vans. For instance, BYD is a Chinese manufacturing company that is considered as the world’s largest company that produces electric vehicles in terms of number of e-trucks sold. Another company that takes the lead in the electric truck industry is Rivian, which is an American electric vehicle automaker company. The company gains its fame when amazon announced https://www.iea.org/topics/transport/evi/ 5 purchasing 100,000 electric trucks from the manufacture in order to achieve their goal to be net zero carbon by 2040 (Downing, 2020). Tesla also has an ambitious plan to present two heavy-duty electric models: one with a 300-mile range and one with a 500-mile range, with expected price range ($150,000 to $180,000) by the end of 2021 (Socio, 2021). In order to achieve a tipping point in truck electrification, a near parity must be achieved with conventional vehicles, regarding the driving range, initial cost and charging infrastructure. Table 1.1 illustrated some of the barriers of using electric vehicles in urban freight. The barriers are grouped into three main categories: knowledge, technical, and economic barriers. Electric trucks are relatively new compared to conventional trucks; therefore consumers may have some concerns about their performance, reliability, and the technology advancement. Also there is a shortage regarding mentioning and clarifying the positive impact of electric vehicles on the environment. Providing potential consumers with detailed information about electric vehicles and their benefits may increase their presence and adoption rate in urban freight. Another important issue is the storage capacity of the electric vehicles batteries, which determines their traveled distance. Range anxiety is considered one of the major problems facing the potential users of electric vehicles. Therefore, battery technology and safety are among the most significant technical barriers. Table 1.1: The barriers of using electric trucks in urban freight. Type of barriers Barriers Literature Knowledge barriers  Lack of verification and evidence on the performance of using e-trucks in urban freight.  Lack of knowledge on the latest technologies used in e-trucks.  Lack of awareness of the impact of using e-trucks on the environment. (Adhikari et al, 2020) Technical barriers  Limited driving range.  User’s acceptance.  Safety regarding battery technology (Wu et al, 2017) Economic barriers  High acquisition cost  High battery costs (Amirhosseini and Hosseini, 2018) 6 The total cost of ownership (TCO) e-Truck for many models will be close to the total cost of ownership for ICE trucks if the battery technology, vehicle performance, and safety of electric vehicles improve. McKinsey estimates that the adoption rate of e- Truck will exceed 30% by 2030 for different vehicle classes: light commercial vehicle (LCV), medium-duty truck (MDT), and heavy-duty truck (HDT) (Furnari et al, 2020). It is expected that the growth of the adoption rate will be faster for light commercial vehicles due to many reasons; first the high similarities with passenger electric vehicles regarding the battery and used technology. Second, the light duty trucks are used in last mile deliveries, where the distances in these routes are typically short, which means that the limited range of electric trucks will not be a barrier, and the battery will not need recharge during the routes. Unlike medium and heavy trucks which are mainly responsible for delivering goods between different warehouses and large distribution centers, where the daily routes are usually long and therefore the battery needs to be charged many times during the visits. Scope of the Problem Truck manufacturers racing to bring electric vehicles into the urban freight market motivated by the world’s keen interest in reducing the greenhouse gas emissions, government incentives, and drop in acquisition and battery costs. In the scope of this thesis, the adoption rate of electric vehicles in urban freight is investigated by studying the impact of operational planning on fleet investment decisions, and the effect of combining routing and fleet composition in increasing the adoption rate of electric vehicles in urban freight distribution. The adoption of electric vehicles in urban freight transport definition is inspired by the following definitions “The sequence of stages through which a consumer progresses from first awareness of an innovation to final acceptance.” (Mahajan and Yoram, 1985) and “The adoption process framework has been utilized to evaluate the potential viability of a new product” (Blattberg and Golanty, 1978; Pringle et al., 1982; Silk and Urban, 1978). Therefore, we can define the adoption of electric vehicles in urban freight by: Moving the electric vehicles from the niche market to the mass market by illustrating their viability and economic feasibility in urban freight. 7 The Purpose of the Thesis This research is motivated by the rapid electrification of the transportation sector as the recent promising technologies used in EVs offer an economical and technically feasible option to increase the sustainability in the transportation and logistics industry. However, some EV’s characteristics impose a negative impact on the adoption rate of the electric vehicles in urban freight operation such as limited driving range and long charging time. Therefore, the overall purpose of this thesis is: Studying the impact of effective management of a mixed fleet composed of electric and conventional vehicles on the purchase decisions over a planning time horizon. Fundamentally, the life cycle analysis and mixed vehicle routing model has been investigated in many researches for decades, although the analysis of alternative fuel vehicles performance specially electric vehicles still remain unfulfilled. Most importantly, none of the existing models integrated the fleet size and mixed vehicle routing model to calculate the operation costs for different fleet compositions. The main objectives of this thesis are listed below:  Understand environmental and economic factors that affect the adoption rate of electric vehicles in urban freight, in order to produce results that are directly relevant to real life situations.  Propose a novel integrated model that considers both routing and fleet composition decisions by integrating fleet size mixed vehicle routing problem with time window and replacement model. The Contributions The key contributions of this work are highlighted as follows: 1. The research presents a new perspective to the research arena by introducing an integrated model that considers both routing and fleet composition decisions for using electric vehicles in urban freight distribution. 8 2. The thesis presents a formulation to find the best fleet mix policy based on integration process and the understanding of the operational characteristics of a fleet composed of electric and conventional vehicles. 3. Our contribution also includes a scenario analysis rather than the traditional method of evaluating the performance of a mixed fleet of electric and conventional vehicles, which allows us to address the impact of different fleet compositions, operation ranges and costs on the fleet mix policy. 4. The inconsistency factors that affect the adoption rate of electric vehicles are considered and a sensitivity analysis to study these factors is presented. 5. The presented research addresses most of real life vehicle’s characteristics and economic factors that can be applied at any vehicle type and at any country. 6. A case study from a real life situation from the city of Istanbul-Turkey is presented to evaluate the performance of the proposed methodology. Research Significance In this section, we present the importance of the presented thesis from the literature point of view, where the main reasons are listed below: 1. The evaluation of using electric vehicles in fleet planning in urban freight has become more complex due to their fast technological advancement; therefore life cycle cost analysis over the age of vehicles is required. However such approaches are few in the literature (Davis and Figliozzi, 2013). 2. The incorporation of electric vehicles in the distribution freight activities especially in heterogeneous fleets shows a promising trend, with the rising importance of using sustainable strategies in road logistics and transportation. (de Armas et al, 2016) 3. The challenge of establishing a systematic and standard methodology for integrating the economic, environmental and social impact assessments is still lagging behind (Gundes, 2016). 9 Thesis Structure This thesis is divided into three main parts; conceptual, mathematical, and application as illustrated in Figure 1.3, this allows the reader to follow the detailed investigation of this thesis in a more logical way. A more detailed overview of each chapter is presented below: Chapter 1- Introduction: gives a general introduction to the problem that we proposed in this dissertation. And illustrate the scope of the problem, the motivation, and the main contributions. Chapter 2- The Use of Electric Vehicles in Urban Freight: begins with defining the fleet management in urban freight, and illustrating the principle of sustainable urban freight transport system. This follows with various experiments that have adopted the use of electric vehicles in their operating fleets are also represented. Since this thesis presented a novel methodology, the literature is reviewed for both mixed vehicle routing problem and replacement models highlighting different aspects of logistics and urban freight. Chapter 3- Methodology and Mathematical Model: discusses the approach we used to investigate presence of electric vehicles in a fleet composed of electric and conventional vehicles in urban freight over a planning time horizon. In addition, this chapter presents the two models used to calculate the optimal replacement policy for a mixed fleet. The first model is the mixed vehicle routing model with a time window, which is used to calculate the operational costs for different fleet compositions, while the second model is the replacement model. The purpose of the replacement model is to obtain the best replacement policy for a set of mixed fleet taking into consideration many economic and vehicular characteristics. Chapter 4 – Computational Experiments: In this chapter, we presented the economic calculation of various parameters that is used in the models. Also computational experiments for the proposed methodology are obtained. Chapter 5- Case Study: We choose to conduct a case study from a retail company in the city of Istanbul-Turkey. Conducting a case study in this thesis aims to investigate the performance of the presented methodology on a real life situation. 10 Chapter 6- Conclusion and Future Work: illustrate the key points of our finding in the presented thesis with a general summary of the contributions. In addition, the outlines of future research are also presented in this chapter. Key Terms  Fleet management: administrative approach companies follow to organize vehicles in a way that enhance the efficiency, reduce costs, and provide compliance with government regulations (Rouse, 2017).  Fleet size: determining the number of vehicles in the fleet that is able to satisfy a complete transportation order while avoiding high costs related to fleet underutilization (Żak et al, 2011).  Fleet mix: determining the optimal mix of vehicles from different types. (Bojovic and Milenkovic, 2008).  Fleet composition: determining the fleet size and mix (Etezadi and Beasley, 1983).  CO2 emissions: colorless, odorless and non-poisonous gas formed by combustion of carbon and in the respiration of living organisms and is considered a greenhouse gas (OECD,2013).  Logistics: The process of strategically managing the movement and storage of materials, finished parts and inventory through the organization and its marketing channels in a way that maximized the future profitability through the cost-effective fulfillment of orders (Behrenbeck et al, 2007). 11 Figure 1.3: Dissertation organization. 12 13 THE USE OF ELECTRIC VEHICLES IN URBAN FREIGHT This chapter aims to investigate relevant studies and literature related to electric vehicles in urban freight transport context and will be discussed in four sections. The first section will present the definition of fleet management in urban freight, followed by a discussion of the sustainability in urban freight operations. A review of solution strategies and applications is also discussed. Furthermore, overviews of several extensions of well-known variants of vehicle routing problem are reviewed. Finally, a literature related to fleet size and mixed vehicle routing problem with time window and replacement model is introduced. Fleet Management in Urban Freight Fleet management can be defined as an administrative approach companies follow to organize vehicles in a way that enhance the efficiency, reduce costs, and provide compliance with government regulations (Rouse, 2017). Depending on the nature of the problem, different approaches and methods can be formulated to solve fleet management problems, such as linear programming, nonlinear programming, goal programming, mixed integer and dynamic programming models. Linear programming for fleet management was discussed early in the literature by (Dantzig and Ramser, 1959). The authors used a linear programming model to minimize the number of tankers in order to meet a fixed schedule. Williams and Fowler (1980) developed a model to deal with the environmental constraints of vehicle acquisition policy and time-dependent fleet loads, the requests of vehicles were generated from probability distributions specific to the demand time series. Mixed integer programming formulation for large scale fleet management under a variety of side constraints, due to marketing, operational, maintenance restrictions was presented by (Rushmeier and Kontogiorgis, 1997). Ziarati et al. (1999) present branch and cut to select the type and number of vehicles that minimize the fixed and operational costs. Calvete et al. (2007) introduced a goal programming approach to solve medium- sized delivery problem for a heterogeneous fleet of vehicles and provide an optimal 14 solution in a reasonable time. Mathew et al. (2010) formulated a non-linear optimization problem for maximizing the total weighted average remaining life of the fleet subjected to different constraints, such as budget, demand, and non- negativity. The authors presented two solution approaches to solve the problem, genetic algorithm (GA) and a branch and bound algorithm (BBA). Jin and Kite- Powell (2000) introduced a dynamic model that optimizes utilization and replacement decisions where the optimal acquisition and retirement strategies were included. For more on this literature, (Oakford et al. 1981, 1984). According to different articles and researches, the fleet management depends on several criteria (Rogic et al, 2007); the fleet size is among the most significant one, along with the size of the operative zone (the covering area of the operations) which is divided into three main categories: local, regional, and national. The third criterion is the routes of the vehicles, and it could be fixed daily routes, or variable daily routes. Another criterion is the time tolerance in delivering goods. In additions, the demand of a customer is also considered as a significant factor in fleet management planning, since it has a direct influence on the fleet size and fleet mix, as the demand of customers can vary in size, location of customers (local, regional) and many other characteristics, as a result, different types and/or size of fleets may be needed to meet the customers demand. Different terms used to define the movement of goods and services within, into, and through urban areas. According to( Lindholm, 2012), ‘urban freight transport’, ‘urban goods transport’ and ‘urban distribution’ are among the most common terms. Since this thesis addresses the planning of a mixed fleet of electric and conventional vehicles freight in urban areas, we choose to use the urban freight transport term as it includes all types of transports in urban areas. It is important to have a clear definition of the term urban freight transport to use through this thesis, therefore, we decide to go with the definition of Allen et al.(2000), as they provide an inclusive definition: (1)” all types and sizes of goods vehicles and other motorized vehicles used for (core) goods collections and deliveries at premises in the urban area.” (2) “all types of goods vehicle movements to and from urban premises including goods transfers between premises, ancillary goods deliveries to urban 15 premises, money collections and deliveries, waste collections and home deliveries made from urban premises to customers.” (3)” service vehicle trips and other vehicle trips for commercial purposes which are essential to the functioning of urban premises”. In recent decades, fleet management in urban freight has increasingly drawn attention as an important topic in the research arena. Ambrosini and Routhier (2004) presented a number of studies and surveys in the area of urban freight transport and good movements. The authors compared the methods and results from different countries in Europe and Asia. Delaître and Routhier (2010) introduced two models that focus on the location and size of delivery in urban areas in order to help decision making processes to develop optimal delivery scenarios by estimating the number of goods vehicles in order to estimate the inconvenience on the overall traffic of the city. Holguín-Veras et al. (2020) presented an extensive discussion of initiatives that could be used to enhance the efficiency of urban freight activity in public-sector including financial approaches; logistical management; and demand/land use management. Several factors affect the development of urban freight transport, such as customer behaviors, rapid technological advancement and economic factors, and with the ongoing urbanization these factors become even more in the future and have a higher influence on the movement in urban freight. Recently, there has been a growing trend of replacing stores by e-commerce and home delivery, due to many economic factors and health concerns. From the first sight, it seems like e-commerce can potentially decrease the movements and transportations, since the customers do not need to come to the stores. However, in reality, customers may replace or examine their products many times before purchasing. Consequently, there will be a growth in freight activities inside cities. The problem with the freight activities, that it has a direct impact on several aspects of daily life such as increasing greenhouse gas emissions, congestion, and pollution. Therefore, there is a need to understand and target the development of the urban freight system in conjunction with our goals of reducing emissions and enhancing life in urban areas. One option is to move towards a sustainable system. Urban freight transport activities are indispensable for the growth of economy and individual income in urban areas, since they have a positive impact on increasing the 16 commercial activities and the development inside cities, in addition to the economic growth. However, they cause a variety of negative impacts especially on the environment since most of the vehicles in urban areas use non-renewable energy sources to power, therefore; they produce a large amount of greenhouse gas emissions, noise, pollution, in addition to the waste of resources as shown in Table 2.1. A variety of social impacts can be noted as a result of urban freight operations, including accidents, traffic congestion, besides the health concerns. Table 2.1 : The impacts of urban freight transport. Type of impacts Impacts Literature review Economic impacts  Increase the development inside cities.  Increase individual income.  Increase the commercial activities.  Economic growth.  (Browne and Allen, 2011) Environmental impacts  Greenhouse gas emissions.  Noise.  Waste of resources such as (oil, materials tiers).  Pollution.  Use non-renewable resources.  (Stefanelli et al, 2015)  AustriaTech,2014) Social impacts  Health concerns.  Accidents  Traffic congestion.  Regulations and policies.  Urban expansion.  (Foltyński, 2014a)  (Behrends et al, 2008)  (Datz et al, 2009) Those impacts lead to unfavorable outcomes such as: health and safety concerns, global warming, and high transport costs, which eventually, reduce the quality of life in urban areas. Those impacts vary depending on the size and the infrastructure of the urban areas (Foltyński, 2014b). It is worth mentioning that the availability of data regarding the impacts of urban freights is poor compared to passenger transport. Sustainability in Urban Freight Moving toward sustainable alternatives and solutions to diminish the negative impacts in urban freight is a priority in urban freight planning. According to 17 Behrends et al (2008) the sustainable urban freight must meet the following objectives:  Ensuring the accessibility offered by the transport system to all categories of inhabitants, commuters, visitors and businesses, in line with the objectives below.  Reducing the negative impact of the transport system on the health, safety and security of the citizens, in particular the most vulnerable ones.  Reducing air pollution and noise emissions, greenhouse gas emissions and energy consumption (including contributing to meeting legislative requirements on air quality and environmental noise  Improving the efficiency and cost-effectiveness of the transportation of persons and goods, taking into account the external costs. .  Contributing to the enhancement of the attractiveness and quality of the urban environment. Figure 2.1 shows the principle of sustainable urban freight transport system based on the sustainable development principles (Behrends et al, 2008). It includes all parts that a sustainable system requires. Figure 2.1: The principle of sustainable urban freight transport system (Behrends et al, 2008). 18 In general, we can consider a freight transport system to be a sustainable one, if it has a contribution to economic growth, environment, and social equity. Moving toward sustainable urban freight, whether by enhancing the current system or by introducing new technologies, faces many challenges such as lack in regulation and policies. Russo and Comi (2012) investigated the sustainability of urban freight in several European cities. The authors argue that the most important aspects to make urban mobility more sustainable is to promote a sustainable development strategy, monitoring and controlling the different types of costs generated by freight mobility in the urban area. The results also show that high cost investments are not required to obtain good results in terms of environmental goals. According to (Eleonor and Blinge, 2014) having awareness about freight transport is an important factor in increasing the sustainability in urban freight transport and can help in the policy making process. The authors also raise an important issue related to sustainability; that there is lack of motivation and knowledge for the local authorities to deal with the problem and rarely there is anyone responsible for freight transportation, which results in postponing the sustainable inside cities. 2.2.1 Urban freight transport strategies Managing fleet for urban freight operations needs planning on tactical and operational levels as it faces continuous changes related to the increase in demand and just in time delivery, which require creative solutions and techniques from a theoretical and mathematical view. Researchers have introduced a number of strategies and practices to optimizing the urban freight operations, some of them are discussed below.  Road pricing The road pricing is the process of charging drivers for using the main roads. Despite being an easy strategy to execute, road pricing continues to be implemented at a low rate in many countries around the world. The main goal of this strategy is to decrease the traffic congestion and the environmental impact in urban areas. Road pricing targets all road participants, whether they are passengers or freight transport, but they affect the passenger’s activities more than freight transport, since the companies tend to charge the customer with the extra cost resulting from the road pricing (Hans Quak, 2011). Ruesch (2004) presented an overview of road pricing in urban freight 19 context. The authors implemented pricing schemes in European urban areas. They conclude that road pricing leads to more sustainable freight operations and has the ability to improve the efficiency of logistics. The authors also recommend that pricing should be linked with regulations, loading factors, and vehicle size and type.  Urban consolidation centers Allen et al. (2007) define the urban consolidation centers as “A logistics facility situated in relatively close proximity to the geographic area that it serves (be that a city centre, an entire town or a specific site such as a shopping centre), to which many logistics companies deliver goods destined for the area, from which consolidated deliveries are carried out within that area, in which a range of other value-added logistics and retail services can be provided.” The urban consolidation centers (UCC) considered as the most famous pooling solution to deal with city logistic in European cities (Chwesiuk and Kijewska, 2010; Van Duin and Muñuzuri, 2015), they are able to reduce the movement of vehicles inside cities, and therefore, reducing air pollution, noises and energy consumption. Faure et al. (2016) stated that the effectiveness of UCC is related to their locations and numbers.  Alternative fuel vehicles Efforts to shift the fuel vehicles to more eco-friendly vehicles in urban freight have been put. Experiments in different countries around the world using electric trucks, hybrid propulsion, and compressed- natural- gas trucks took place (Hans Quak, 2011). Different experiments show that using alternative vehicles has a positive impact on decreasing the pollution and noises and they are more energy efficient. Bethoux (2020) investigated the hydrogen fuel cell vehicles in road transport, the authors examined to what extent hydrogen fuel cell vehicles can satisfy the demands of the car industry and the possibility to implement the large scale FCV in road transport and their effect on the environment and the economy. It is worth mentioning that electric vehicles are among the most investigated vehicles to be used in urban freight operations to reach sustainability in urban freight operations.  Policies and regulations Setting policies to manage the freight delivery is one of the most important steps in urban freight sustainability, the main target of those policies is to limit the vehicles 20 (trucks and freight delivery) access to the urban areas in a way that minimize congestion, pollution and improve air quality, such as setting delivery time windows, low emission zones, set a special time for deliveries, loading and unloading operations,and determining specific characteristics to the vehicles that are allowed to enter the urban areas (tonnage, size, age). One of the challenges facing the vehicle restriction strategy is the tendency for companies to start their business early in conjunction with rush hours, since most employees favor working within a fixed schedule, especially in the public sector, encouraging the employees to have a more flexible schedule with fixed hours could help more in establishing the vehicle restriction strategy widely. Table 2.2 shows some of the strategies that are implemented in some cities around the world. Table 2.2 : Illustrates some of the strategies that have been applied in different cities. Strategies City Literature Roading price London Stockholm Tokyo (Russo and Comi, 2012) (Wappelhorst et al, 2020) Consolidation centers London Tenjin (Browne et al, 2011) (Browne et al, 2005) Alternative fuel vehicles New York Australia London Amsterdam (Browne et al, 2011) (Forde, 2020) (Baster et al, 2014) Loading and unloading Turin (Diana et al, 2020) Zero-Emission Zones Rotterdam, Shenzhen (Transport Decarbonization Alliance, 2020) Among all strategies presented above, we choose focusing on using alternative vehicles, more specifically using electric vehicles in urban freight. 2.2.2 SWOT analysis of electric vehicles in urban freight SWOT analysis is a strategic tool for evaluating the internal and external factors of an organization’s resources. The SWOT analysis stands for: Strength, Weakness, Opportunity, and Threats. Despite being simple, SWOT is a very efficient tool in assessing the current status of electric vehicles and estimating the future threats. 21 Table 2.3 illustrates the strength, weakness, opportunities, and threats of using EVs in urban freight distribution. Table 2.3 : SWOT of electric vehicles in urban freight. Strength Weakness  Low fuel costs ( Rahimi and Davoudi, 2018),(Mouhrim et al, 2018)  High purchase costs (Ahani et al, 2018) (Lebeau et al, 2013)  Low maintenance costs (Feng and Figliozzi, 2013), (Kleindorfer et al, 2012), (Macharis et al, 2013)  Limited driving range (Ahani et al, 2018) (Zhao and Lu, 2019)  Eco- friendly (Foltyńsk, 2014b) (Stefanelli et al, 2015), (Aljohani and Thompson, 2018).  Limited editions and models (Wu et al, 2017)  Low noise (Teoh et al, 2018),(Ruesch, 2004), (Macrina et al, 2019)  Charging infrastructure (Nallusamy et al, 2016), (Ashkrof et al, 2020)  Limited maintenance workshop (Adhikari et al, 2020). Opportunities Threats  Increase in battery capacity (Redondo- iglesias et al, 2019)  Lack in regulations and policies (Mirhedayatian and Yan, 2018), (Taefi et al, 2016), (Green et al, 2014)  New vehicles have a higher driving range (Sanguesa et al, 2021), (Davis and Figliozzi, 2013)  Unstable of energy prices (Adhikari et al, 2020)  Technology development is better than ICVs (Ahani et al, 2018).  Hydrogen vehicles (Manoharan et al, 2019), (Bethoux, 2020)  Fast charging (Ajanovic and Haas, 2016), (Baster et al, 2014)  Environmental awareness (Adhikari et al, 2020) The main strength of using electric vehicles is the low running cost. According to EIA (2021) the gasoline prices equal $3.02/gallon. Meanwhile, the prices of electricity equal $0.0982/KWh. Secondly, (Feng and Figliozzi, 2013) reported that 22 the maintenance costs for electric vehicles are 50% less than that of conventional vehicles. Also the positive impacts of using electric vehicles on the environment as they produce less noise and pollution. On the other hand the weakness related to its high initial cost, despite the government’s incentives and subsidies. In addition, the limited driving range of electric vehicles due to the limited capacity of the battery is considered one of the main weaknesses of EVs along with the absence of charging infrastructure in many cities around the world. Also the scarce maintenance workshop, where the potential customer cannot make the purchase knowing that there is only a limited number of workshops that are able to fix the vehicles. Finally, the limited editions and models, even though the companies started offering different models and sizes, they still lack an extensive lineup of products compared to conventional vehicles. The opportunities lie in the new battery capacity, where the battery charge lasts longer, also the continuous technology improvements in electric vehicles, which results in a significant increase in the market demand of electric vehicles. The rapid advancement of electric vehicles charging technology should also be highlighted as an essential attribute in spreading the existence of electric vehicles. The first significant threat facing electric vehicles is the lack of regulation and policies. Secondly, the fluctuation in energy prices due to many economic factors. Finally, electric vehicles may face a prominent threat from the hydrogen vehicles, which are also considered as eco-friendly vehicles. 2.2.3 Studies and experiences The idea of adopting green vehicles for urban freight distribution is not new. These vehicles are powered by green energy such as electricity. However, actual use and implementation of those alternative vehicles inside cities have started recently. In this section, we reviewed the initiatives implemented to reduce the environmental impacts of daily freight transport activities in the following companies:  Heineken The cities of Amsterdam/ Rotterdam have been supporting the use of electric vehicles in freight operations by using different incentives, such as government subsidies, and traffic regulation exemptions for logistics operators that use electric vehicles, for both vans and trucks. Heineken is a company that delivers their products to the shops and bars in cities throughout the Netherlands using 220 trucks, each 23 covering between 100 and 250 km per day. They started adopting electric vehicles gradually in their operating fleet, currently they operate with 23 and 28 trucks from their Rotterdam and Amsterdam depots respectively. And they are targeting a 100% electric truck for secondary distribution centers in the near future, using renewable energy to recharge the vehicles. Recently, Heineken, deploy six electric freight vehicles (with 12 tons) in Amsterdam and one electric truck (with 19 tons) in Rotterdam. Although they are currently using electric vehicles in their fleet, those are the first e-trucks as large as 12 and 19 tons to be operated in Heineken. Those trucks in Rotterdam and Amsterdam operate almost exclusively in the city centre. Their daily average cutting distance is 60 kms, with an average drop count of 13 to 17 deliveries.  IKEA Ikea is a Swedish company that designs and sells ready-to-assemble furniture, As of November 2020, there are 445 IKEA stores operating in 52 countries around the world. Ikea set a goal to become climate positive and clean transport in city center by cut CO2 emissions in all stages of their value chain by 2030. They are taking significant steps on this front by using electric vehicles (EVs) or other zero-emission solutions. In New York City, Ikea is targeting to use 40 electric vehicles to service all five boroughs by May 2021. They are focusing now on building charging infrastructure at IKEA stores as it works toward their main target of having 100% zero emission deliveries by 2025 (Conshohocken, 2021). In addition, the Australian branch of IKEA has also committed to use all electric vehicles in their delivery fleet by 2025. The company now operates with a fleet of 100 trucks for large shipments and 250 trucks for small ones. They are using 7 electric trucks in delivering goods in Sydney, Perth and Melbourne.  DHL DHL is a company for package delivery, and express mail service ranging from domestic parcel delivery to international express. The company operates with nearly 100,000 vehicles all over the world. The company is aware of the amount of CO2 emissions produced by those vehicles. Therefore, they decide to take further steps towards making the cities greener and reducing the impact of commercial vehicles on the environment. Recently, they started using, more eco-friendly way to distribute https://en.wikipedia.org/wiki/Ready-to-assemble_furniture 24 deliveries from the depots to their final recipients. In 2017, DHL set a plan, called Mission 2050, to reduce its net carbon emissions to zero by 2050 (Reid, 2019). The short term target for 2025 is to reduce emissions by operating clean services for 70% of the company’s pickup and delivery services. Today, DHL uses more than 3,200 electric bikes, plus 9,000 other e-bikes and e-trikes all over the world, with potential increase in some markets. In London, DHL Express has launched ten new electric courier vans as part of UK fleet (DHL, 2021).  Amazon In support of the Paris agreement, Amazon had committed to be net zero carbon by 2040 by using electric vehicles and inventing new alternative delivery solutions. In 2019, Amazon ordered 100,000 electric delivery vehicles, which is considered as the biggest order ever for electric vehicles. Their plan is to operate with 10,000 new electric vehicles as early as 2022 and 100000 vehicles on the road by 2030 (Coren, 2019). Currently, Amazon operates using hundreds of electric vehicles around the world, and integrating charging infrastructure to use. In addition they are using e- cargo bikes for deliveries in some European cities and New York City.  Office delivery in London In 2009, London made the decision to trial a new urban delivery system to reduce the environmental impact of the freight operation inside the city (Browne et al, 2011). The trial involved the use of an urban micro-consolidation centre along with electric vans and electric cargo tricycles. The micro consolidation centre was located in the city of London, which is the historic core of London with an area of 2.9 km2. The electric vans and tricycles deliver the parcels from the centre to the final customers, which were all located in London as well. The results indicate that the use of a micro-consolidation centre along with a total fleet of electric vans and cargo tricycles reduced the total distance driven per parcel delivered between the centre and the customer delivery locations by 20%, and the CO2 equivalent emissions per parcel delivered have also reduced by 54%. The trial emphasizes that even in a supply chain system in which goods are highly consolidated, the potential to achieve sustainability, reduce greenhouse gas emissions and total distance travelled inside cities. In addition, the system operated in the https://discover.dhl.com/business/business-ethics/the-path-toward-zero-emissions 25 presented trial has a direct contribution in improving the air quality and reducing the noises. Vehicle Routing Problem The vehicle routing problem is the problem of finding the optimal set of routes to perform all transportation demands with a specific fleet, more precisely, managing which vehicle handles which customers and the sequence of visiting those customers in a way that minimizes the total cost or distance. The high interest of vehicle routing problem (VRP) is not only motivated by their complexity as optimization problems but also by their relevance to real-world application. As a consequence, the academic and industrial world focuses more on different variants of VRP. More than 70 years have elapsed since Dantzig and Ramser (1959) presented the first vehicle routing problem, at that time called the dispatching problem. The authors’ concerned about finding the optimum between a bulk terminal and a large number of service stations supplied by the terminal routes for a fleet composed of gasoline delivery trucks. To solve the problem, the authors presented a simple matching-based heuristic. Clarke and Wright (1964) extend Dantzig and Ramser (1959) work and formulate a VRP with more restrictions and constraints like different vehicle capacities. The authors proposed a simple but effective heuristic to find a near-optimal solution. Figure 2.2: An example of vehicle routing problem presentation. 26 The vehicle routing problem is classified according different factors, but mainly:  The network structure, where the operation is performed on the locations of the customers, which are identified as vertices of a graph called node routing problem. In contrast, the operations that are performed on the arcs are called the arc routing problem.  The type of transportation requests, which is all the requests of all goods that are distributed from the depot to a set of customers. The main types are: delivery and collections, point-to-point transportation, alternative and indirect services, split and non-splits services, multimodal services, dynamic and stochastic routing  The constraints, such as capacities, route length, time window, multiple use of vehicles  The objective function, which can be a single objective optimization function, or multiple criteria optimization function. Dantzig–Fulkerson–Johnson formulate the travelling salesman problem (TSP) as an integer linear program (Dantzig and Ramser, 1959), later on the formulation was extended to create a two index vehicle flow formulations to present the vehicle routing problem as shown below (Laporte, 1992a):  Model formulation On a graph G = (V, E) where V = {0,. . ., n} is the vertex set, and E = {(i, j)| i ≠j; i, j ∈ V} is the arc set. V represents the customers and the depot, where i=0 denotes 0, and vertices from i=1,…,n corresponds to the customers. 𝐶𝑖𝑗 present the costs of going from i to j, K is the number of vehicles, 𝑟(𝑠) present the minimum number of vehicles needed to serve set S and 0 denote the depot. 𝑀𝑖𝑛 ∑ ∑ 𝐶𝑖𝑗𝑋𝑖𝑗 𝑗∈𝑉𝑖∈𝑉 (2.1) S.t: ∑ 𝑋𝑖𝑗 𝑖∈𝑉 = 1 ∀𝑗 ∈ 𝑉{0} (2.2) 27 ∑ 𝑋𝑖𝑗 𝑗∈𝑉 = 1 ∀𝑖 ∈ 𝑉{0} (2.3) ∑ 𝑋𝑖0 𝑖∈𝑉 = 𝐾 (2.4) ∑ 𝑋0𝑗 𝑗∈𝑉 = 𝐾 (2.5) ∑ ∑ Xij j∈S𝑖∉𝐒 ≤ |𝑆| − 𝑟(𝑠) (2.6) 𝑋𝑖𝑗 {0,1} ∀ 𝑖, 𝑗 ∈ 𝑉 (2.7) The objective function is to minimize the total costs. The first and second constraints denote that the number of vehicles entering each vertex must equal the number of vehicles leaving the vertex. Constraints (2.4, 2.5) ensure that the number of vehicles leaving the depot must equal the number of vehicles entering the depot. Constraint (2.6) is the sub tour elimination. While the last constraint shows the integrality constraints. 2.3.1 Vehicle routing problem classifications Different variants of vehicle routing problem were discussed in the literature; such as vehicle routing problem with backhauls (VRPB), heterogeneous or mixed fleet (HFVRP), periodic vehicle routing problem (PVRP) and split delivery vehicle routing problem (SDVRP), In addition to the capacitated vehicle routing problem (CVRP). The VRPB is concerned about delivering goods from depot to customers and vice versa, where the delivering of goods from depot to customers is known as linehaul, and the backhaul goods picked up from the customers to the depot. The HFVRP refers to a fleet of vehicles at the depot with different vehicle specifications, like capacity and distance range; it has been studied first by (Golden et al, 1984). PVRP is a variant where the customers require repeated visits with a planning time horizon. The SDVRP is concerned about visiting the same customer more than once to satisfy the customers demand (Irnich et al, 2014). 28 Figure 2.3: Vehicle routing problem variants. The capacitated vehicle routing problem is one of the basic models in the vehicle routing problem family, which has been extensively studied in the literature. The problem can be described as follows; a set of m identical vehicles with identical capacities and are located at the depot, the objective is to determine a set of routes that satisfy all customer demand with least cost, so that each route does not exceed 29 the vehicle capacity and that each customer visits only once. Time windows are a common extension in vehicle routing problem formulations where the service time of each customer is determined in advance. For illustrative purposes, the capacitated vehicle routing problem with time windows formulation is introduced here and it is an extension to the one presented in the previous section. On a graph G = (V, E) where V = {0,. . ., n} is the vertex set, and E = {(i, j)| i ≠j; i, j ∈ V} is the arc set. V represents the customers and the depot, where i=0 denotes 0, and vertices from i=1,…,n corresponds to the customers. Let 𝑎𝑖, 𝑏𝑖 are time window for each customer 𝑖 ∈ 𝑉, and 𝜏𝑖 is the start time of each customer service, 𝑠𝑖 is the service time, and 𝑡𝑖𝑗 is the travel time from customer i to j. 𝑞𝑖 presents the customer demand. 𝑄𝑖 denotes the capacity of the vehicle after leaving customer i, while G presented the vehicle capacity. 𝑀𝑖𝑛 ∑ ∑ 𝐶𝑖𝑗𝑋𝑖𝑗 𝑗∈𝑉𝑖∈𝑉 (2.8) S.t: ∑ 𝑋𝑖𝑗 𝑖∈𝑉 = 1 ∀𝑗 ∈ 𝑉/{0} (2.9) ∑ 𝑋𝑖𝑗 𝑗∈𝑉 = 1 ∀𝑖 ∈ 𝑉/{0} (2.10) ∑ 𝑋𝑖0 𝑖∈𝑉 = 𝐾 (2.11) ∑ 𝑋0𝑗 𝑗∈𝑉 = 𝐾 (2.12) ∑ ∑ Xij j∈S𝑖∉𝐒 ≤ |𝑆| − 𝑟(𝑠) (2.13) 𝑋𝑖𝑗 (𝜏𝑖 + 𝑠𝑖 + 𝑡𝑖𝑗 − 𝜏𝑗 ) ≤ 0 , ∀𝑖, 𝑗 ∈ 𝑉 (2.14) 𝑎𝑖 ≤ 𝜏𝑖 ≤ 𝑏𝑖, ∀𝑖 ∈ {1,2, … , 𝑛} (2.15) 30 𝑋𝑖𝑗 (𝑄𝑖 + 𝑞𝑗 − 𝑄𝑗) ≤ 0 ∀𝑖, 𝑗 ∈ 𝑉 (2.16) 𝑄𝑖 ≤ 𝐺 ∀𝑖 ∈ 𝑉 (2.17) 𝑋𝑖𝑗 {0,1} ∀ 𝑖, 𝑗 ∈ 𝑉 (2.18) Constraint (2.14) specifies the relationship between the service time of customer i and customer. Constraint (2.15) indicates the time window of customer i. Constraints (2.16, 2.17) are the capacity constraints, where constraint (2.16) ensure that the capacity of the vehicle after leaving customer i, is reduced by the value of customer i demand, and that the value of the vehicle’s capacity after leaving customer i, doesn’t exceed the capacity of the vehicle. Constraint (2.18) indicates the integrality constraints. The distance constraint is another extension of the vehicle routing problem. In this variant the vehicles have limited driving range that cannot be exceeded during the route. Let the maximum driving range of vehicles denote by 𝐷𝑘, and 𝑑𝑖𝑗the distance between customers. The constraint can be defined as follow: ∑ ∑ 𝑋𝑖𝑗𝑘 𝑁 𝑗=1 𝑁 𝑖=1 . 𝑑𝑖𝑗 ≤ 𝐷𝑘 , ∀𝑘 ∈ {1,2, … , 𝐾} (2.19) In general, vehicle routing problem (VRP) models are classified into three main levels according to their degree of realism as shown in Figure 2.4. The first one is the classical-basic VRP models, a theoretical problem, they mainly used to solve methods in controlled environments, and where their performance can be assessed before executing them in practical applications. Solving them can be done by exact or approximate methods. The classical advanced vehicle routing problem models are characterized by a higher level of realism; such as integrated routing and logistics, large-scale problems and multi-objective functions. The Rich VRP models are more complex, where metaheuristic algorithms used to solve them such as genetic algorithms, ant colony optimization, and local search. 31 Figure 2.4: Classification of Vehicle Routing Problem (VRP) models according to their degree of realism ( Caceres-Cruz et al, 2014). 2.3.2 Exact and heuristic method for the VRP Vehicle routing problem is an NP hard problem, where exact approaches can only solve small instances. Few exact approaches introduced in the literature to solve VRP. Laporte (1992) classified exact approaches introduced in the literature into three classes: • Direct tree-search methods • Dynamic programming methods • Integer linear programming methods. Branch and Bound based on the following k-degree center tree was proposed early by (Christofides et al, 1981). They have successfully solved VRPs ranging in size from 10 to 25 vertices. Fisher and Jaikumar (1981) ) improves Christofides et al. (1981) method by defining K-tree to be a set of n + K edges that span the graph. This algorithm improves the optimal solutions for a number of difficult problems with 100 customers for well-known problems and with 25–71 customers for several real problems (Eilon et al, 1974) was first to propose dynamic programming to solve vehicle routing problem. Only a few researches were presented in the literature to solve the vehicle routing problem using dynamic programming, since the problem is considered weak for large size problems. Balinski and Quandt (1964) was first to develop a set partitioning problem to solve VRP for truck delivery. 32 Route construction heuristics is a traditional heuristic approach that select nodes based on minimization criterion in a sequential way until a feasible solution is obtained (Bräysy and Gendreau, 2005), one of the most know heuristic is the savings heuristic of (Clarke and Wright, 1964) which was basically introduced to solve VRP problem. The other heuristic is the nearest neighbor heuristic, the routes start by finding the nearest un-routed customer to the depot, then finding the unrouted customer that is close to the last added one, until all unrouted customers are included. Sweep heuristic was proposed by (Gillett and Miller, 1974), Solomon (1987) was the first to solve vehicle routing problem with time window constraints using sweep heuristic, the logic behind the heuristic is to divided the problem, into clusters stages and scheduling stages (Bräysy and Gendreau, 2005). The parallel route building was proposed by Potvin and Rousseau (1993) This algorithm is based on the idea of initializing many routes at the same time (parallel), and uses a generalized regret measure to select the next unrouted customers for insertion. Improvement heuristics is based on improving the initial solution by performing neighborhood search iteratively. Most improvement algorithms are used for the vehicle routing problem such as string relocation and string exchange. Figure 2.5: Solution algorithms for VRP. 33 Literature Review In this section, abroad literature on mixed vehicle routing problem with time windows is first presented in section 2.4.1, followed by a review on replacement model in section 2.4.2. 2.4.1 The mixed vehicle routing problem with time window Vehicle fleet composition can be found in the literature within two premier categories (Du et al, 2016; Etezadi and Beasley, 1983): (1) vehicle fleet size problems, which refer to the decision of determining the number of homogeneous fleet of vehicles, and (2) vehicle fleet composition problems, which refer to the problem of deciding the fleet size and mix simultaneously for a mixed fleet of vehicles. Fleet composition problems are usually discussed in combination with other transportation problems. Hoff et al. (2010) introduced a comprehensive literature review that discusses the industrial aspects of combined routing and fleet composition in maritime and road transportation. The authors classified the research articles related to this integration into four classes: fleet size mixed vehicle routing problem, heterogeneous fixed fleet vehicle routing problem, fleet size mixed vehicle routing problem with time windows, fleet size mixed vehicle routing problem with multiple depots. Golden et al. (1984) presented the first article that relaxed the homogeneous fleet assumption in vehicle routing problem, where different types of vehicles assumed to be available at the depot. Their formulation is classified as a fleet size and mix vehicle routing problem. Van Duin et al. (2013) investigated the impact of routing constraints, electric vehicle characteristics and driving environment on the cost differences between electric and conventional commercial vehicles. They formulated the problem as a mixed integer programming model and then used sequential insertion heuristic to solve the problem. The authors have represented the vehicle acquisition costs as a daily fixed cost of the vehicle used. This approach, despite its simplicity, still requires analysis of different fleet mix scenarios, and restricts the use of the model for multi-period fleet investment decisions. Salhi et al. (2013) introduced fleet size and mixed vehicle routing problem with backhauls, the authors presented new ILP formulation to minimize the total cost of routes originating and terminating at the depot, taking into consideration the capacity and route length limitations. Sassi et al. (2015) handled a 34 mixed fleet of electric and conventional vehicle routing problem, where different charging technologies and time dependent charging costs were considered. They introduced a charging routing heuristic to generate initial solutions as well as an inject-eject-based local search method with three different insertion strategies. Rezgui et al. (2015) introduced the electric Modular Fleet Size and Mix Vehicle Routing Problem with Time Window, which is a new type of logistics scheme for urban freight delivery that takes into account the possibility of recharging at a customer location. As a solution method, the authors developed an approach based on a genetic algorithm. The experimental results obtained on benchmarks from the literature show that with the modularity feature, using electric vehicles for freight delivery in urban environments is economically interesting. Li et al. (2016) formulate mixed bus fleet management (MBFM) taking into consideration four different types of buses (compressed natural gas bus, diesel bus, electric bus, hybrid-diesel bus) in order to solve the problem, the authors presented a new approach new life additional benefit-cost (NLABC) which aims maximizing the total net benefit. The presented formulation allowed them to determine the optimal fleet size and composition, while including the routing problem. Macrina et al. (2019) modeled a mixed vehicle fleet composed of EVs and CVs, where partially recharging the EVs was allowed. The authors proposed an iterated local search metaheuristic to solve the formulated model. Several computational experiments on modified benchmark instances were conducted. They concluded that the use of partial recharge may lead to more effective and sustainable solutions. Hiermann et al. (2019) introduced a mix of conventional, hybrid, and electric vehicles routing problem. They designed a hybrid genetic algorithm to solve the problem. Rezgui et al. (2019) presented a fleet size and mix vehicle routing problem with EV modular to achieve sustainable urban freight deliveries. The modules can be released at customer’s locations to overcome length restrictions in some urban areas, to recharge the battery or to help respect delays when performing the tours. Alizadeh Foroutan et al. (2020) considered a green vehicle routing and scheduling problem with heterogeneous fleet including reverse logistics in the form of collecting returned goods along with weighted earliness and tardiness costs. To find near-optimal solutions simulated annealing (SA) and a genetic algorithm (GA) were suggested and evaluated with respect to two considered criteria, solutions quality, and 35 computational times. Goeke and Schneider (2015) optimize the routing of a mixed vehicle fleet consisting of electric commercial vehicles (ECVs) and conventional internal combustion commercial vehicles (ICCVs). In contrast to the existing routing models, the presented model utilizes a realistic energy consumption model that incorporates speed, gradient and cargo load distribution. An Adaptive Large Neighborhood Search algorithm that is enhanced by a local search for intensification was developed. Lebeau et al. (2015) formulated a fleet size and mix vehicle routing problem with time windows for electric vehicles integrated with the energy consumption model, so that variable range of electric vehicles can be considered. The authors investigate different vehicle’s sizes with either electric propulsion or internal combustion engine vehicles. Those vehicles vary in costs, payload, energy consumption and many more. In their study the authors considered different aspects of the problem, such as the fast charge of electric vehicles at the depot, time window and range constraints. Mouhrim et al. (2018) presented a mixed integer linear programming to model a vehicle routing problem with a mixed fleet of electric and conventional vehicles with time window and capacity constraints, and two limitations; the conventional vehicles are limited with a fixed quantity of greenhouse gas emissions, and the electric vehicles are limited with range. Schiffer et al. (2021) developed a new methodology to analyze the comparative competitiveness between conventional and electric vehicles, by combining the calculations of total cost of ownership with a rich location-routing model. The authors presented an integrated model that takes into account strategic network design and operational routing decisions over multiple periods. The presented model in this thesis is an extension to the one presented by (Lee et al. 2008). The authors present a vehicle mix fleet problem with a single depot denote by (0) and a set of N customers, 𝑑𝑖𝑗the distance between customers, k type of vehicles, 𝐺𝑘 is the capacity of vehicle from type k, 𝑞𝑖 is the customer’s demand, two decision variable are presented: 𝑋𝑖𝑗𝑘, 𝑌𝑖𝑗. When the vehicle travel over the arc between i, j, 𝑋𝑖𝑗𝑘 equal 1 otherwise zero, 𝑌𝑖𝑗 present the vehicle load from customer i to j. The objective function is to minimize the operation costs for a set of heterogeneous vehicles: 36 ∑ ∑ ∑ 𝑜𝑘 𝑁 𝑗=0 𝑁 𝑖=0 𝐾 𝑘=1 . 𝑑𝑖𝑗. 𝑋𝑖𝑗𝑘 (2.20) S.t: ∑ ∑ 𝑋𝑖𝑗𝑘 𝑁 𝑖=0 𝐾 𝑘=1 = 1, ∀𝑗 ∈ {1,2, … , 𝑁} (2.21) ∑ ∑ 𝑋𝑖𝑗𝑘 𝑁 𝑗=0 𝐾 𝑘=1 = 1, ∀𝑖 ∈ {1,2, … , 𝑁} (2.22) ∑ 𝑋𝑖𝑗𝑘 𝑁 𝑖=0 = ∑ 𝑋𝑗𝑙𝑘 𝑁 𝑙=0 , ∀𝑗 ∈ {1,2, … , 𝑁}, ∀𝑘 ∈ {1,2, … , 𝐾} (2.23) ∑ 𝑌𝑖𝑗 𝑁 𝑖=0 − ∑ 𝑌𝑗𝑙 𝑁 𝑙=0 = 𝑞 𝑗 , ∀𝑗 ∈ {1,2, … , 𝑁} (2.24) ∑ 𝑌𝑖0 𝑁 𝑖=1 = 0 (2.25) ∑ 𝑌0𝑗 𝑁 𝑗=1 = ∑ 𝑞 𝑖 𝑁 𝑖=1 (2.26) 𝑌𝑖𝑗 ≤ ∑ 𝐺𝑘 𝐾 𝑘=1 . 𝑋𝑖𝑗𝑘, ∀𝑖, 𝑗 ∈ {1,2, … , 𝑁}, 𝑖 ≠ 𝑗 (2.27) 𝑌𝑖𝑗 ≥ 0, 𝑌𝑖𝑖 = 0, 𝑋𝑖𝑗𝑘 ∈ {0,1} (2.28) Constraints (2.21, 2.22) ensure that each customer is served once by one vehicle from any type. Constraint (2.23) implies that the vehicle entering each customer, should leave that customer. Constraint (2.24) ensures that the demand of each customer has to be satisfied. Constraint (2.25) guarantees that all vehicles from all 37 types should return back to the depot empty. Constraint (2.26) shows that the sum of all deliveries in all vehicles is equal to the sum of all customers’ demand. Constraint (2.27) is the capacity constraints, where the vehicle load from customer i to customer j should not exceed the vehicle’s capacity. And the last constraint refers to the non- negativity of the decision variable 𝑌𝑖𝑗 and that 𝑋𝑖𝑗𝑘 is a binary decision variable. 2.4.2 Fleet replacement model Previous studies have discussed different aspects o