International Conference on Transportation and Development

Transcription

1 International Conference on Tranportation and Development Modeling Freight Tranportation a a Sytem-of-Sytem to Determine Adoption of Emerging Vehicle Technologie A. Guerrero de la Peña 1 ; N. Davendralingam 2 ; A. K. Raz 3 ; V. Sujan 4 ; D. DeLaurenti 5 ; G. Shaver 6 ; and N. Jain 7 Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. 1 School of Mechanical Engineering Purdue Univ. Wet Lafayette IN guerrer2@purdue.edu 2 School of Aeronautic and Atronautic Purdue Univ. Wet Lafayette IN School of Aeronautic and Atronautic Purdue Univ. Wet Lafayette IN Cummin Inc. Columbu IN School of Aeronautic and Atronautic Purdue Univ. Wet Lafayette IN School of Mechanical Engineering Purdue Univ. Wet Lafayette IN School of Mechanical Engineering Purdue Univ. Wet Lafayette IN ABSTRACT The U.S. freight tranportation ytem i a complex agglomeration of interacting ytem that include line-haul and urban delivery vehicle inter and intra-city highway and upport infratructure. In order to project the evolution of the ytem and the market penetration of emerging freight vehicle technologie it i important to model the aforementioned interconnection public adoption preference and operational and policy contraint that impact it. In thi paper we propoe a ytem-of-ytem engineering approach to define the cope of influential mechanim and abtract an appropriate model of the U.S. freight tranportation ytem with focu on a line-haul cenario. Implementation over a multi-city network i poed a a contrained mixed-integer linear program. The allocation and operation of three vehicle architecture conventional dieel dieel platooning and battery electric are optimized over a multi-city network to minimize the fleet-wide total cot of ownerhip over a twenty-year time horizon. We examine the effect of projected change in energy cot freight demand and hourof-ervice regulation on the annual market hare evolution of thee technologie. INTRODUCTION Motivation and problem definition: More than 90% of Heavy Duty Cla 8 vehicle ued for line-haul operation are till dieelized and are a ignificant producer of carbon emiion in the U.S (National Reearch Council 2010) (U.S. Department of Tranportation 2016). Over the lat few decade tringent regulation and improved vehicle technologie have been introduced to reduce national fuel conumption and CO2 emiion. Technologie that increae the aerodynamic efficiency of the vehicle reduce rolling reitance or increae engine efficiency have already been widely adopted (North American Council for Freight Efficiency 2016). Through thee technologie vehicle fuel efficiency ha een an average increae of 12% in the U.S. (NACFE 2016). However thee technologie only offer incremental improvement in the fuel economy of conventional Dieel vehicle. New and revolutionary architecture including alternative fuel vehicle hybridization full electrification and increaed level of autonomy have been propoed to further decreae national fuel conumption and vehicle emiion (Vimmertedt et al. 2015). A tool that can project the adoption path of thee emerging technologie would enable technology manufacturer and policy maker to make deciion regarding technology innovation introduction to market and economic incentive that will International Conference on Tranportation and Development 2018

2 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. maximize adoption and reult in targeted reduction in fuel conumption and CO2 level. Gap in literature: The cot attractivene of emerging freight tranportation technologie and the enuing reduction in emiion given future adoption i an iue that ha been addreed in everal tudie. Lammert et al. (2014) preent fuel conumption reduction of two Cla 8 Dieel tractor-trailer vehicle operating in platooning mode at varying gap ditance teady-tate peed and gro vehicle weight during a erie of track tet. Their reult demontrate combined fuel aving of up to 6.4% howing an attractive return on invetment with repect to operational fuel cot. Zhao et al. (2013) evaluate variou powertrain architecture including Dieel conventional hybrid electric and natural ga over day-long hort-haul and long-haul imulated drive cycle. The author derive the CO2 emiion reduction potential given adoption of thee technologie and preent the break-even fuel cot for economic attractivene. Fulton and Miller (2015) explore deep market penetration cenario of different low-carbon vehicle technologie uch a alternative fuel architecture and electric vehicle and the reulting capability to meet 80% reduction of CO2 emiion in the U.S. by Their reult are baed on hitorical and projected data for freight demand trucking tonnage and mile hare in the U.S. Finally Schafer and Jacoby (2006) preent rate of adoption for different peronal vehicle and Heavy Duty Cla 8 Dieel truck technologie under CO2 emiion contraint and penalty cot. Depite introducing interconnection between economic factor CO2 emiion policie and vehicle performance to their analyi Schafer and Jacoby extrapolate trucking metric uch a average annual mile and coneuently truck fuel conumption from hitorical data. Effect on the evolution of freight tranportation metric caued by policy change and introduction of new technologie may not be effectively captured by projection of hitorical trend. Studie like (Lammert et al. 2014) and (Zhao et al. 2013) determine economic attractivene baed only on vehicle efficiencie over an iolated drive cycle without conidering external factor that will alo affect technology adoption. Fulton and Miller (2015) project national reduction in CO2 emiion provided aumed technology adoption cenario while Schafer and Jacoby (2006) do o by extrapolating hitorical trucking data. Thee tudie do not provide a holitic treatment of how fleet adopt and operate vehicle with different technologie over the tranportation network. The coupling of vehicle efficiencie and freight tranportation ytem conideration make technology adoption a multi-faceted problem. Thee interconnection mut be imulated to effectively project the operational cot technology adoption path and reulting emiion outcome. Achieving thi goal reuire a imulation framework with an ability to include factor pertaining to multiple independent ytem factor that cannot be oberved in a ingle drive cycle uch a fleet management and operation policie and network characteritic that ultimately affect economic attractivene of emerging technologie. Contribution: In thi paper we propoe a formulation to etimate fleet-wide operational and purchaing cot a a function of vehicle technology election and allocation over a regional freight tranportation network in order to project future adoption of emerging technologie. A Sytem-of-Sytem (SoS) engineering methodology i ued to define the cope of conideration neceary to model adoption of freight tranportation technologie a well a the appropriate level of abtraction for imulation. Reducing the complexity of vehicle performance modeling enable introduction of fleet management factor policie and infratructure availability conideration that influence the attractivene of fleet tranportation technologie. Moreover thi approach produce a holitic analyi of technology adoption trend in contrat to traditional method of platform-centric analyi. Outline: Thi paper i organized a follow. The next ection include a brief overview of International Conference on Tranportation and Development 2018

3 International Conference on Tranportation and Development the Sytem of Sytem (SoS) methodology applied to a regional line-haul freight tranportation ytem. The linear program implemented for optimization of operational and purchaing cot of a regional fleet i decribed in the Model Formulation ection. Technology adoption projection for a ingle fleet operating over a mall network of regional highway are preented in the Simulation ection followed by concluding tatement. SYSTEM OF SYSTEMS FORMULATION Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. A Sytem of Sytem i a complex ytem that conit of a collection of entitie which collaborate for a uniue purpoe while retaining operational and managerial independence. An SoS will alo how evolutionary dynamic that i their internal tructure will change over time a the contituent ytem and network evolve. The U.S. line-haul freight tranportation ytem how a prevalence of thee trait it i compoed of interconnected ytem of vehicle inter and intra-city highway and upport infratructure organized at multiple level and evolving over time. In order to model the evolution of thi ytem and determine adoption of emerging freight vehicle technologie it i important to identify relevant holitic factor and mechanim and tranlate them ytematically into an actionable model. Recognizing the need for a holitic methodology DeLaurenti (2005) ha developed a Sytem-of-Sytem modeling and analyi framework that can be ued to view the tranportation ytem within a SoS context. Thi SoS framework ha three main phae the definition phae the abtraction phae and the implementation phae a dicued below. Definition Phae: The definition phae eek to etablih a tructural map of the SoS in thi cae the regional freight tranportation ytem in term of hierarchie and categorie. Thi phae ue a contruct known a a ROPE table hown in Table 1 which erve a a problem coping platform for ubeuent analyi. The column of the ROPE table correpond to the Reource Operation Policy and Economic dimenion of the SoS pace. Each row repreent the hierarchical level of the SoS. In thi cae the alpha level of the ROPE table correpond to the dicrete technologie that are implemented in line haul vehicle. Higher level beta and above reflect aggregation of element from lower level entitie; for example a combination of technologie at the alpha level would contitute a vehicle unit at the beta level and a combination of beta-level vehicle would contitute a fleet at the gamma-level. The ROPE table enable the SoS engineer/deigner to eek out variou factor acro ROPE categorie that are expected to play a role in the projected evolution of the SoS and it future compoition of vehicle technologie. The entrie hown in Table 1 are repreentative of a U.S. regional Line-Haul ytem. For example when building the ROPE table for line-haul tranportation driver hour of ervice limit weight peed and emiion retriction are identified a influential U.S. policy conideration at different level of the SoS. Thi implie that evaluation of powertrain or vehicle technology performance alone detached from the ret of the SoS level i not enough to oberve evolution of the ytem. Abtraction Phae: The abtraction phae include decriptor for the entrie of the ROPE table and conideration for takeholder behavior incentive tructure and relevant model ued to decribe them. In the context of our line-haul problem the total cot of ownerhip i a utility that fleet owner eek to optimize when conidering vehicle purchae. An examination of the ROPE table reveal that the total cot of ownerhip (lited under Economic at the gammalevel) i influenced by more factor than the cot of emerging vehicle technologie alone. The goal of the abtraction phae then i to develop repreentation that will decribe the total cot of ownerhip and incorporate the influence of uch relationhip o an appropriate trategy for International Conference on Tranportation and Development 2018

4 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. implementation can be brought to bear in the next phae. Thi problem abtraction i provided in the form of a paper model i.e. a decription of the big picture dynamic a hown in Figure 1. The paper model facilitate the tranition to the implementation phae by defining the organization and interconnection of the overall entitie decribed in the ROPE matrix. Implementation Phae: The implementation phae eek to realize olution approache for the problem formulation obtained in the abtraction phae. In thi work we cope the profiteeking behavior of a repreentative line-haul fleet in the abtraction phae. We then model thi behavior in the context of a cot minimization optimization problem and write the relevant mathematical expreion baed on the variable and decriptor etablihed in the abtraction phae. The mathematical formulation and implementation are decribed in the following ection. Alpha Beta Gamma Delta Table 1. Regional Line-Haul Freight Tranportation Sytem ROPE matrix Reource Operation Economic Policy Dieel engine Powertrain fuel Cot of fuel Emiion Battery Electric conumption Cot of energy retriction ICE Cot of fuel energy Conventional conumed vehicle Cot of purchae ICE + Platooning Cot of driver/hour BEV Ton-mi/gal efficiency Average day operation Vehicle life cycle Cargo load/capacity Mile driven baed on elected route Operate at contant peed over route Operator hour 2-veh platooning Vehicle range Fleet ditribution Fleet ize Vehicle in ingle Vehicle replacement regional fleet cycle and year of ervice 4-city network: City 1-City 2- City 3- City 4 MODEL FORMULATION limit Total Freight demand between citie by weight Traffic condition: vehicle on road road denity capacity travel time Single direct route between citie Total cot of ownerhip deciion metric per year Cot of fuel Cot of energy b. weight limit Driver hour of ervice limit Speed limit Regional emiion reduction Thi tudy focue on line-haul truck operation over a regional inter-city network a defined in the ROPE matrix and abtraction model (Table 1 and Figure 1 repectively). The computer model can be ued to imulate the deciion proce of fleet to purchae thoe vehicle architecture that are economically attractive to them given their operational and purchae cot. Our intent i to replicate the Total Cot of Ownerhip (TCO) minimization behavior of fleet International Conference on Tranportation and Development 2018

5 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. owner for Heavy Duty Cla 8 trucking highway operation given vehicle highway performance fleet management conideration infratructure availability and external influence uch a cot of energy and regional freight demand. A ytem optimal traffic allocation approach i ued to etimate the operational cot while new vehicle purchae cot and turnover ale revenue are included a metric for vehicle acuiition. Fleet management policie and vehicle operational conideration are formulated a contraint. The linearity of the reulting model objective function and contraint make the optimization problem a Mixed-Integer Linear Program (MILP) to which highly efficient and matured mean of olution are available. Figure 1. Regional Line-Haul Freight Tranportation Sytem Paper Model. Problem Formulation: A mixed-integer linear programming (MILP) formulation i choen to repreent a large regional fleet with a focu on the line-haul highway operation conideration identified in the ROPE matrix. The objective function repreent vehicle purchaing criteria a a function of etimated economic attractivene of different technologie operated over the network route. The deciion variable are defined a x repreenting the number of n vehicle of type that originate at per year x n vehicle n of type originating at traveling on link h (ij) and y the cargo link flow originating at h. Objective Function: The objective function repreent the Total Cot of Ownerhip (TCO) criteria commonly ued by fleet owner in order to elect vehicle for purchae. On average fuel conumption repair and maintenance of a vehicle and driver wage incur the highet percentage of total operational cot on a per km bai over a vehicle lifecycle (Torrey and Murray 2016)(NACFE 2016)(Tranportation Reearch Board 2011). Fleet owner compute thee cot component a a deciion-making criteria for purchae of new technologie over conventional one. The TCO objective i compried of thee operational cot the cot aociated with vehicle technology reliability and the cot of purchae and turnover ale revenue. Fleet commonly purchae vehicle on a yearly bai and therefore thi deciion-making proce i exercied annually throughout the period of tudy. The objective function i defined a the total fleet operational and purchaing cot uch that: Jk Cop Cpur. Here the ubcript k indicate that the total cot of ownerhip i computed every year where k 120 for the cae tudy preented. To implify the notation the ubcript k i dropped throughout the formulation with the exception of cae where reuired to indicate the ue of value from previou year. The cot n International Conference on Tranportation and Development 2018

6 International Conference on Tranportation and Development of operation include the cot of energy conumed Ce driver Cdr maintenance CM and revenue loe CR uch that Cop Cec Cdr CM CR. Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. Table 2. Model Parametrization Beta Level-Vehicle Architecture Parametrization Operational ξ Efficiency function of vehicle type gal km R Driving range of vehicle type km W Capacity for vehicle type unit of weight ton Eco2 CO2 emiion rate for vehicle architecture g km B Reliability of vehicle architecture %trip year Economic $ C M Cot of maintenance per mile km C Cot of purchae for vehicle type $ p C Reale value for vehicle type r $ Gamma Level-Fleet Management Parametrization lmin l max hicle turnover range year γ Projected vehicle life-cycle period year C driver river wage B eet budget for vehicle purchae $ f C delay c jection of revenue lo due to delay for cargo type c $ km Delta Level-Network Parametrization h b i Cargo demand from origin h to detination i ton d Length (ditance) of link (ij) km h o Hour of ervice limit hour C e Cot for energy conumed by vehicle type $ $ hr Energy conumption cot vary depending on the vehicle technology ued. Cot of energy i defined a C x d C where x xn repreent the flow of vehicle of ec i j A type over highway link (ij) regardle of their origin and A i the et of city-node in the n International Conference on Tranportation and Development 2018

7 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. network. The cot of energy conumed per mile C u C e i a function of the vehicle efficiency which depend on average vehicle peed. Operational cot are computed a a function of the etimated total number of trip in an average operational day. In order to etimate lifecycle cot the cot of energy conumed over an average day i multiplied by γ the number of year in a vehicle lifecycle period. The driver cot C dr are computed on a per km bai given the total ditance traveled by fleet vehicle on an average operational day. Similar to the energy conumption cot the driver cot are weighted over the vehicle expected lifecycle: C x d C. The dr driver i j A technology type and age of a vehicle can affect it maintenance and repair cot and i commonly ued a a metric by fleet owner to identify the appropriate turnover age of their vehicle. Maintenance cot are defined a C x d C. Reliability of cargo delivery may be M. M i j A affected by vehicle or component break-down. Here we aume technology reliability iue reult in time delay T d and cheduling of a econd vehicle for completion of delivery i not neceary. In that manner reliability cot are modeled a loe in revenue due to the incurred delay and are a function of both vehicle and cargo type: C x B T C. R d delay c i j A The cot of purchae conider the cot of buying new vehicle and the revenue generated by elling ued one: Cpurch Cnv Cr. Fleet will purchae new vehicle 1) to replace thoe beyond their economic lifecycle or 2) to increae fleet volume due to an increae in freight demand. Here we aume that all vehicle are purchaed new uch that C x C. The nv new p variable x new i introduced to repreent the vehicle of technology type newly adopted in the current year of projection k. Thi mean there are additional vehicle of type originating at node that were not allocated in previou year. The variable x new i given by x x where x xn. Furthermore new n k k1 x i poitive only if new vehicle are allocated to origin and zero otherwie. Fleet will ell older vehicle when they are near the end of their economic life the age at which maintenance and repair cot increae and efficiency i no longer optimal. At thi point fleet may replace older vehicle with a newer purchae. The revenue obtained from a ale i computed a C x c and then implemented a an offet to the purchaing r r r budget of the current year. The turnover period lmin l max during which a vehicle approache the end of it economic life and i conidered for replacement varie by fleet. Here we aume that a line-haul fleet ha a fixed range for vehicle turnover age. In contrat to new vehicle purchaed the variable x r i given by x x k 1 k and i introduced to repreent vehicle old by the fleet. The value i poitive if vehicle of type allocated to origin during the current year of projection are le than in the previou year and zero otherwie. Computation of x r mut take precedence over new vehicle purchae a older vehicle may be old and replaced with newer vehicle of the International Conference on Tranportation and Development 2018

8 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. ame technology. In ummary the TCO i computed a follow: J x d C C C B T C x C x C (1) driver M d delay c new p r r i ja Contraint: The vehicle demand over the network i defined a a function of cargo demand h b i between hi city pair. Vehicle link flow will be optimized in order to atify cargo demand vehicle flow balance entering and leaving node and capacity contraint a given by Euation (2a)-(2d). Hour of ervice limit h o a hown in E. (2e) will alo have an effect on the amount of vehicle trip taken within the time contraint and therefore the number of vehicle needed to be allocated over the network. An intermediate binary variable x n i introduced and aigned a value of 1 if the nth vehicle of type i ued. Thi ait in the computation of total number of vehicle of type purchaed and allocated to city uch that xn M x and n Mx n x j n for all i where M i a ufficiently large number. Table 3. Optimization Contraint Contraint Expreion Contraint Expreion h h h h (2a) y ji y bi (2d) xn ii 0 yii 0 j h (2b) y x W j n h n (2c) xn ji xn 0 i j j (2e) x t h n r o n New vehicle purchae are contrained by a uer-defined fleet budget which i offet by the revenue created from vehicle ale uch that x newc p x rcr Bf i j. A market penetration contraint x new Qavailable Q i alo added to repreent the availability of vehicle technologie entering the market. The parameter Qavailable can be calibrated to limit the rate of penetration of newer technologie with lower production rate a exiting technologie are phaed out. Vehicle reale i alo contrained uch that x t l x x t l x t l. Vehicle older than the maximum new yk max r new yk max new yk min allowable age will be old while vehicle within the turnover range may be conidered for replacement. Some technologie may have a ignificant impact on the operation of vehicle over the network. For two-vehicle platooning operation vehicle with thi technology mut travel in pair over any link (ij) in order to gain the aociated efficiency benefit regardle of their origin or detination. Therefore an intermediate integer variable P i introduced to enforce thi contraint uch that xn 2P for any architecture with platooning capability. n We implement the method and euation introduced by Zheng et al. (2017) to determine feaibility of travel e for electric vehicle originating at traveling on link (ij) given the vehicle range and location of charging tation. A ummarized in (Zheng et al. 2017) the j International Conference on Tranportation and Development 2018

9 International Conference on Tranportation and Development variable L j i incremented by the (ij) link ditance d if node i doe not have charging infratructure. The variable e will then have a value of 1 if the total ditance L i i within the vehicle range limit. The location of charging tation i not optimized in our formulation; it i intead defined a a network parameter. Therefore the tatu Ei of a node a a charging tation i an input to the MILP. Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. Contra int Expreion (3a) Lj Li d M 1 e (3b) (3c) Table 4. Electric Vehicle Contraint Contrai Expreion nt (3d) x Me n L i n R (3e) L 0 L 0 L L ME L L ME i i i i i i Li M 1 Ei (3f) i i E e Ei 0 i ic Traffic Model: Vehicle efficiency will vary with repect to average vehicle peed over a network route. The olution u to Greenhield macrocopic traffic flow model (William n.d.) provide the average traffic peed baed on the number of vehicle f both freight and paenger introduced to the link and the route characteritic uch that: k 2 f u ku 0 v N. The Greenhield euation involve a nonlinear term with repect to f the traffic peed u itelf a function of freight traffic flow. In order to have a linear contraint to facilitate the ue of a linear olver we etimate the number of freight vehicle traveling over d each link (ij) and olve for the peed a-priori. Finally route time can be computed a tr. u Emiion Model: All vehicle architecture elected a part of thi tudy mut be compliant with the CO2 emiion and fuel tandard a pecified by the EPA Greenhoue Ga (GHG) Phae 2 releae (US EPA 2016). However computation of cumulative regional emiion can be ueful to determine policy utainability and future carbon reduction given the diplacement of foil fuel emiion by greener technologie. It can alo ait in the determination of neceary cot incentive for cutomer in order to achieve a deirable outcome. The regional emiion ER are calculated a a cumulative regional value given by ER x de co2. The vehicle i ja g emiion output E co2 in km provide the ma of CO2 produced by the energy conumed per kilometer. In the cae of Dieel engine thi i proportional to the gallon of fuel ued wherea electric vehicle emiion are a function of the ma of CO2 produced per kwh CO2 conumed: E co2 kwh. The value of CO 2 can vary with repect to the year of projection to kwh repreent a regional hift to cleaner ource for production of energy. International Conference on Tranportation and Development 2018

10 International Conference on Tranportation and Development MODEL SIMULATION Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. A 4-city network i defined to demontrate the capability of the propoed model; the cargo demand and route ditance are hown in Table 6. Three vehicle technologie Dieel Dieel platooning and battery electric are evaluated with the parametrization hown in Table 5. Vehicle capacity and maintenance cot are contant acro all technologie to reduce complexity of model calibration and analyi. To limit computation time only citie 1 and 2 are choen a origin for vehicle allocation and location of charging tation. Thi and the aumed electric vehicle range limit electric vehicle operation to the direct route between citie 1 and 2. The cot of electricity i maintained from year to year at $0.10 per kwh while the cot of Dieel i aumed to vary. The analyi how the purchaing behavior of a mall regional fleet over a 20 year period auming contant cargo demand a vehicle turnover range of 3-5 year and vehicle lifecycle γ of 5 year. Figure 2 and 3 how the effect of the hour of ervice (HOS) policy fleet purchaing budget contraint and cot of Dieel on vehicle adoption projection vehicle mile traveled and reulting CO2 emiion for a ingle fleet. All adoption VMT and emiion value are normalized with repect to the Dieel vehicle output for the year Although long-term Dieel price are difficult to etimate the elected value for imulation hown by the dahed curve in Figure 2 erve to demontrate the effect of fuel cot volatility on technology election. Table 5. Vehicle Architecture Parametrization Peak W Cp Depreciation R Emiion B Eff. Rate Veh. Type (mi/ec) (ton) ($k) (%/year) (mi) kg CO2/EC (% trip delayed) Dieel % Dieel % Platooning BEV % Table 6. Network Parametrization Cargo Demand (ton/day) Route ditance (mi) O/D City 1 City 2 City 3 City 4 O/D City 1 City 2 City 3 City 4 City City City City City City City City We can oberve in all cae hown in Figure 2 that the adoption of platooning vehicle increae in the year 2022 a the cot of Dieel increae. However technology adoption doe not vary between 2022 and 2031 regardle of variation in cot of Dieel. Thi indicate that fleet adoption of more efficient technologie may be delayed following a large but temporary change in the cot of Dieel due to vehicle turnover retriction. On the other hand an increae in International Conference on Tranportation and Development 2018

11 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. adoption of conventional Dieel vehicle occur from a the price of Dieel remain low. An increae in hour of ervice from 11 to 15 hour in addition to a reduction in fuel cot increae the adoption of conventional Dieel vehicle toward the end of the imulation period a indicated by a comparion between Figure 2a and 2b. However no effect are oberved on vehicle adoption in the firt 15 year. A expected a limited purchaing budget will impact the adoption of platooning vehicle a indicated by comparion of Figure 2a and 2c and 2b and 2d. Again Figure 2b and 2c how the ame adoption trend depite change in budget and hour of ervice yet different VMT and emiion value. Upon further inpection of Figure 3a and 3b we oberve that vehicle utilization varie with the increae in hour of ervice a indicated by the vehicle mile traveled (VMT) per year. Adoption value are euivalent in the firt 15 year (for a fixed number of hour of ervice (HOS) but reulting emiion differ between the two cae from 2017 to 2021 (compare Figure 3a and 3c). Note that emiion are plotted on the right y-axi in Figure 3. A hown in Figure 2d and 3d a change in budget and HOS retriction caue the mot variation in technology adoption throughout the 20 year period. Figure 2. Annual vehicle adoption projection for different budget and HOS value. Cot of Dieel i plotted uing a dahed line. Adoption value are normalized. Figure 2 and 3 how electric vehicle are not adopted in the imulated cenario due to the high purchae cot. Figure 4 and 5 how the impact of decreaing the purchae cot of electric vehicle from $ to $ on adoption and fleet emiion a well a the effect of hour of ervice regulation. For thi cae tudy we aume a regional dependency on coal and ga a ource of electricity and etimate a production rate of 0.4kg of CO2 per kwh conumed by battery electric vehicle. We oberve that following a decreae in cot electric vehicle are adopted between the year a hown in Figure 4a and 4b but adoption drop to zero during the lat 5 year of tudy when the cot of Dieel ha remained low. Figure 4b how that an increae in HOS from 11 to 15 hour reult in increaed adoption of electric vehicle from A a reult CO2 emiion are lower throughout thi 10 year period a oberved in Figure 5b. International Conference on Tranportation and Development 2018

12 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. Figure 3. Annual total VMT projection per vehicle type for different budget and HOS value. Value are normalized. Figure 4. Annual vehicle adoption projection for different HOS value and reduced BEV purchae cot. Adoption value are normalized. Figure 5. Annual total VMT projection per vehicle type for different HOS value and reduced BEV purchae cot. Value hown are normalized. CONCLUSION U.S. freight tranportation i a complex ytem-of-ytem; it i compoed of interconnected ytem including line-haul and urban delivery vehicle inter and intra-city highway and International Conference on Tranportation and Development 2018

13 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. upport infratructure. In thi paper we ued the SoS engineering methodology to define abtract and imulate the U.S. freight tranportation ytem with a focu on line-haul cenario. We propoed a contrained mixed-integer linear program to optimize the allocation of three vehicle technologie conventional Dieel Dieel platooning and battery electric over a multicity network with repect to minimization of total cot of ownerhip over a twenty-year time horizon. The effect of energy cot freight demand and hour-of-ervice regulation were evaluated to determine annual market hare evolution of thee technologie. The reult demontrated the enitivity of future adoption trend to change in exogenou factor identified during the SoS definition phae fuel cot fleet budget and vehicle turnover conideration and hour of ervice policie. Future work can focu on increaing the fidelity of implementation and extend the formulation to imulate vehicle allocation of everal fleet over a larger inter-city region. Thi would enable the uer to oberve regional adoption trend of emerging technologie provided a range of fleet management conideration and repreentative ditribution network. ACKNOWLEDGEMENTS The author thank Cummin Inc. for the upport provided during the development of thi reearch. REFERENCES DeLaurenti D. (2005 January). Undertanding Tranportation a a Sytem-of-Sytem Deign Problem. American Intitute of Aeronautic and Atronautic. Fulton L. and Miller M. (2015 June). Strategie for Tranitioning to Low-Carbon Emiion Truck in the United State.pdf. National Center for Sutainable Tranportation. Retrieved Lammert M. P. Duran A. Diez J. Burton K. and Nicholon A. (2014 September). Effect of Platooning on Fuel Conumption of Cla 8 Vehicle Over a Range of Speed Following Ditance and Ma. SAE International Journal of Commercial Vehicle 7(2) Retrieved doi: / National Reearch Council. (2010). Technologie and Approache to Reducing the Fuel Conumption of Medium- and Heavy-Duty Vehicle. Wahington DC.: National Academie Pre. Retrieved doi: /12845 North American Council for Freight Efficiency. (2016) Annual Fleet Fuel Study (Tech. Rep.). Retrieved Schafer A. and Jacoby H. D. (2006 June). Vehicle technology under CO2 contraint: a general euilibrium analyi. Energy Policy 34(9) Retrieved doi: /j.enpol Torrey W. F. and Murray D. (n.d.). An Analyi of the Operational Cot of Trucking: 2016 Update. Retrieved Operational-Cot-of-Trucking pdf Tranportation Reearch Board National Cooperative Freight Reearch Program and Tranportation Reearch Board. (2011). Impact of Public Policy on the Freight Tranportation Sytem. Wahington DC.: National Academie Pre. Retrieved doi: /14453 U.S. Department of Tranportation. (2016 January). Freight Fact and Figure U.S. Department of Tranportation BTS. Retrieved from bt/ite/rita.dot.gov.bt/file/data and tatitic/by ubject/freight/ freight fact 2015 International Conference on Tranportation and Development 2018

14 International Conference on Tranportation and Development Downloaded from acelibrary.org by Purdue Univerity Librarie on 07/31/18. Copyright ASCE. For peronal ue only; all right reerved. U.S. EPA. (2016 September). Final Rule for Greenhoue Ga Emiion and Fuel Efficiency Standard for Medium- and Heavy-Duty Engine and Vehicle - Phae 2 [Policie and Guidance]. Retrieved Vimmertedt L. Brown A. Newe E. Markel T. Schroeder A. Zhang Y. Johnon S. (2015). Tranformative Reduction of Tranportation Greenhoue Ga Emiion. Opportunitie for Change in Technologie and Sytem (Tech. Rep.). National Renewable Energy Laboratory (NREL) Golden CO US. Retrieved William J. C. (n.d.). Revied Monograph on Traffic Flow Theory. Retrieved from Zhao H. Burke A. and Zhu L. (2013). Analyi of Cla 8 hybrid-electric truck technologie uing dieel LNG electricity and hydrogen a the fuel for variou application. In Electric Vehicle Sympoium and Exhibition (EVS27) 2013 World (pp. 1 16). IEEE. Retrieved Zheng H. He X. Li Y. and Peeta S. (2017 June). Traffic Euilibrium and Charging Facility Location for Electric Vehicle. Network and Spatial Economic 17(2) Retrieved doi: / z International Conference on Tranportation and Development 2018