Climate Implications of Truck Platooning Adoption: Insights from System Dynamics Modeling

Abstract

Freight transportation is a significant contributor to greenhouse gas (GHG) emissions in the US. As an emerging technology, truck platooning leverages vehicle-to-vehicle communications to enable trucks to travel in convoys with close proximity, which reduces air drag and consequently truck fuel use and GHG emissions. However, uncertainties remain about how this emerging technology may be adopted and its climate impacts. To this end, this paper investigates the role of truck platooning adoption in mitigating the climate impact of trucking from a system perspective. Considering the dynamic nature of truck platooning adoption, system dynamics (SD) models based on stock and flow diagrams are developed to estimate the potential reduction in fuel use and CO₂ emissions in the US trucking sector when truck platooning technology becomes available. The results show that adopting platooning could save 292 million metric tons of CO₂ emissions in 180 months after the initial introduction of the technology in the US truck sector. The analysis provides insights for accelerating truck platooning adoption while enhancing its environmental impact.

1. Introduction

The US freight transportation system is essential to the movement of goods and the national economy, carrying 55.2 million tons of freight per day in 2019 and being projected to grow by 1.4% annually from 2022 to 2050 [1,2]. Among all freight modes, trucking carries the largest share in terms of both tonnage and value, and continues to grow faster than other modes due to its speed, flexibility, and broad service coverage across both urban and remote regions [3,4,5]. However, this dominant role also brings substantial environmental impacts: trucking generated 1060 million metric tons of CO₂-equivalent GHG emissions in 2021 and accounted for 60.6% of transportation-related CO₂ emissions in the US [6]. Medium- and heavy-duty trucks are especially significant, producing 23% of the total emissions despite representing only 4% of US vehicles [7], highlighting the urgent need for cleaner technologies in the trucking industry.

Truck platooning has recently emerged as a promising technology in which two or more trucks travel in a convoy with short headways enabled by advanced driver-assistance systems [8]. This convoy, called a platoon, typically has a human-driven lead truck, while the following trucks automatically respond to the lead vehicle, although human drivers are still required in all trucks [9]. In this paper, trucks capable of platooning are referred to as platoonable trucks. By reducing aerodynamic drag, platooning can lower fuel consumption and thus CO₂ emissions, with savings of up to 20% demonstrated in theoretical, simulation, and real-world studies [10,11,12]. Owing to this potential, truck platooning is expected to see substantial growth, with the market projected to expand from $37.6 million in 2021 to $2.7 billion by 2030 [13].

As an emerging technology, truck platooning faces significant uncertainties regarding its performance, adoption rates, and ultimate impact on transportation and the environment [14]. While it promises reduced fuel consumption and CO₂ emissions, its actual climate impact depends heavily on dynamic adoption trends and operational conditions within the US trucking industry [15,16]. Crucially, these uncertainties introduce the risk of unintended consequences that could offset platooning’s initial benefits. For example, operational inefficiencies on congested routes could negate fuel savings through increased idling or routing detours. Furthermore, the technology’s perceived convenience might induce an overall increase in the Vehicle Miles Traveled (VMT) by driving a mode shift from rail to road freight [17], potentially increasing road freight transport by up to 18% [18]. Because these variables—fuel consumption, VMT, and mode shift—interact continuously, isolating their individual effects is insufficient; thus, evaluating the true climate impact of truck platooning requires a system thinking approach to comprehensively capture the complex, time-delayed behavior of the trucking industry [19].

Motivated by the above, the objective of this research is twofold. First, this research seeks to comprehensively understand the dynamics of the trucking industry as it pertains to adopting truck platooning. We employ a system dynamics (SD) approach, which is well known for its effectiveness in constructing a stock and flow model to comprehensively characterize the structure of complex systems. Two SD models are developed. The first one represents the US trucking system’s functioning without platooning, while the second SD model depicts the system’s functioning with truck platooning. In the first SD model, we solely consider the trucking system. In this context, the trucking system is defined as a system contains trucking fleets operating on the US road network, their skilled drivers, and their interactions. However, in the second SD model, we analyze a System-of-Systems (SoS) that comprises two components: the trucking system and the truck platooning system, as well as their interaction. The truck platooning system refers to a system that includes platoonable trucks and the associated rules and regulations implemented by relevant entities or authorities. Furthermore, the second SD model incorporates the decision making of truck drivers through the assessment of three technology adoption scenarios based on the technology’s benefit-to-cost ratio.

Building on the developed SD models, the second objective of this research is to investigate the potential impact of truck platooning on GHG emissions in the US’s trucking system. Given that CO₂ emissions account for 96% of total transportation-related GHG emissions [20] and data availability, we focus on quantifying and forecasting CO₂ emissions that result from the implementation of truck platooning over time. The modeling results offer valuable insights to freight stakeholders, helping them make informed decisions and develop relevant policies toward a greener and more sustainable trucking system and the overall freight transportation system through the adoption of truck platooning.

2. Literature Review

In this section, we first provide an overview of the existing works on SD modeling, with an emphasis on the applications of SD to analyze CO₂ emissions from road transportation. Then, among all the benefits of truck platooning, we focus on one which can lead to CO₂ emission reduction and review related studies.

2.1. System Dynamics and Its Application in Sustainable ROAD Transportation

Many of the challenges we currently face, such as air pollution and climate change, are unforeseen consequences of decisions made in the past. Although decision makers may have the best intentions to address these issues, a lack of dynamic, system thinking leads to policies not always achieving their intended goals and even inadvertently causing new problems. As such, a system thinking and modeling approach becomes indispensable for understanding the behavior of complex systems over time, especially for emerging technologies [19]. SD is a powerful modeling tool that has been widely employed across various fields over the past three decades [21]. By utilizing SD, transportation researchers can effectively explore and address complex problems, aiding in the formulation of sustainable road transportation strategies.

SD modeling has been widely used to study sustainable road transportation in the past. Kim (1998) [22] develops an SD model for highway management to achieve sustainable development for the Commonwealth of Virginia. The urban transportation of Dalian city, China, is modeled by Wang et al. (2008) [23]. They propose policies based on the control variable, vehicle ownership, to mitigate NO₂ emission. An SD model is proposed by Fong et al. (2009) [24] to predict the future CO₂ emission patterns in the urban development of Malaysia and global warming potential. Barisa and Rosa (2018) [25] propose an SD model for analysing the mitigation of CO₂ emissions in the road transportation system in Latvia. Their model aims to predict the CO₂ emissions generated from the transportation sub-system by considering changes in social, economic, and technical aspects. Esfandabadi et al. (2020) [26] apply SD modeling to systematize the interconnections between carsharing and its environmental effects. Using an SD model, Wen and Wang (2022) [27] investigate how carbon neutrality and carbon peak would be achieved in Beijing’s public transportation considering the adoption of relevant policies.

2.2. Emissions Reduction Using Truck Platooning

Platooning helps reduce truck emissions in one major way. Platooning reduces fuel use due to the decrease in aerodynamic drag, especially at high speeds (since the aerodynamic drag is proportional to the second power of speed, platooning can be more effective with regard to energy saving at higher speed [28]) [15,29]. The extent of air drag reduction during platooning is estimated based mainly on (1) theoretical studies using computational fluid dynamics [30,31,32,33,34] and (2) experimental wind tunnel and track testing [35,36,37,38]. In these estimations, parameters such as trucks’ shape and size, cruising speed, the number of trucks in a platoon, and inter-vehicular spacing within the platoon are considered.

Despite the considerable efforts made within the literature to investigate the environmental impacts of truck platooning diffusion, a significant gap remains in understanding these effects through the lens of a system dynamics framework. The absence of such investigations limits the comprehensive understanding of truck platooning as a dynamic, interconnected system, thereby impeding the identification of critical components and chains of interdependencies influencing its environmental consequences. Addressing this gap is imperative as it can provide decision makers with a holistic perspective, facilitating the observation of how changes in various system parameters impact CO₂ emissions, and ultimately aiding in the formulation of effective policies to optimize system efficiency and environmental outcomes.

Another gap in the existing literature relates to the lack of assessment regarding the sensitivity of CO₂ emission production to different economic and technological attributes of truck platooning technology. Quantifying the sensitivity of the CO₂ emissions of the technology to various technological and economic factors (such as fuel saving efficiency, production cost, fuel price, etc.) holds immense significance for both technology enablers and policy makers. Through such analysis, it becomes feasible to identify the most influential features of the technology in relation to emissions. Consequently, this knowledge empowers stakeholders to allocate resources strategically, focusing on the most promising and cost-effective technology enhancement strategies, which can lead to substantial reductions in CO₂ emissions within the trucking system.

While Aboulkacem and Combes (2023) [39] provide a valuable foundation by examining the market uptake of truck platooning from a micro-economic standpoint, their study does not address the environmental consequences of adoption within a system dynamics framework and further assumes that conventional trucks cannot be retrofitted into platoonable vehicles. More recently, Choobchian et al. (2024) [40] investigate truck platooning diffusion using a system dynamics approach and explicitly consider the role of a matching platform, demonstrating its potential to accelerate adoption and improve fuel and labor savings in the US trucking industry. However, their analysis is primarily centered on platform-enabled coordination and the associated operational gains. The present study departs from this focus by developing a climate-oriented system dynamics model of truck platooning adoption in the US trucking sector, explicitly quantifying its effects on fuel use and CO₂ emissions over time. Moreover, the model incorporates the possibility of retrofitting conventional trucks into platoonable trucks and examines the sensitivity of environmental outcomes to major economic and technological factors, such as fuel price, technology fuel efficiency, and technology cost. In this way, the study extends the literature beyond adoption dynamics and operational benefits toward a direct evaluation of the environmental implications of truck platooning diffusion.

2.3. Contributions of This Study

To address the identified gaps in the literature, this study attempts to make two contributions. Firstly, a novel SD model is developed to capture the diffusion of truck platooning technology within the trucking industry. This model investigates the influence of various factors, such as the economic efficiency of the technology, and the rate of return of the economy, on the diffusion of truck platooning and its resulting CO₂ emissions. Given the complex interdependency among these factors, the SD approach proves to be a suitable strategy for comprehending the complexities inherent in the entire diffusion process. Furthermore, an extensive sensitivity analysis is conducted to assess the impacts of different parameters in the model, including the fuel price, technology fuel efficiency, and technology price, on CO₂ emissions. The results highlight significant variations in the CO₂ emissions attributed to the various elements of the technology. These findings offer valuable insights to technology enablers, empowering them to devise the most effective technological alignments to curtail CO₂ emissions effectively.

3. System Definition

The truck platooning system will operate as an embedded part of the trucking system. Since truck platooning system has operational and managerial independence from the trucking system, we use the paradigm of SoS [41] for the study. In the next subsection, we define the truck platooning system, the existing trucking system, and the system boundaries through a function-oriented perspective.

3.1. Truck Platooning System

Truck platooning is an innovative system that is revolutionizing the trucking industry in the US. It offers several benefits, including enhanced fuel efficiency and reduced driver wage and shortage [28]. It serves as an excellent illustration of a SoS that aligns precisely with the criteria proposed by [41]. The fundamental building blocks of this SoS are clearly the platoonable trucks themselves, which act as the key constituent systems (in the context of an SoS definition, constituent systems refer to the individual systems or components that make up the larger, complex system [42]). These trucks dynamically form platoons, facilitating efficient collaboration and coordination as they operate over time. However, the widespread implementation of truck platooning faces challenges such as regulatory frameworks, infrastructure requirements, and public acceptance. Nonetheless, as technology continues to advance and stakeholders work towards addressing these challenges, truck platooning holds great potential for transforming the US freight market by improving efficiency and reducing overall emissions [28].

3.2. Existing Trucking System

The trucking system is highly favored for its flexibility, swift delivery, and door-to-door service in transporting various commodities. It contains a fleet of trucks, skilled drivers, and intricate logistic systems. Despite its advantages, the trucking system encounters challenges, including CO₂ emissions, rising operational costs, and driver shortages and aging [43]. The expected driver shortage in the future may expedite the adoption of truck platooning, especially in countries with high labor costs, due to potential health benefits for drivers, improved well-being, and the ability to multitask [28,44,45,46]. However, the decision between platoonable and traditional trucks remains complex, as it requires the careful consideration of various factors.

The decision-making process of truck drivers involves evaluating the quantifiable benefits and drawbacks of adopting truck platooning technology. Two primary decisions arise: comparing platooning benefits when purchasing new trucks and evaluating the advantages and costs of adopting platoonable trucks or retrofitting existing ones with advanced technologies like lidar and sensors [46]. This assessment involves considering factors such as improved fuel efficiency, reduced emissions, and the potential cost savings resulting from decreased labor requirements [28]. By carefully weighing these considerations, truck drivers can determine the optimal approach to integrate platooning technology while balancing economic viability and operational feasibility.

3.3. System Boundary

SD models should be developed within a closed-system boundary, where all system interactions occur [47]. Although illustrating the boundary of a complex system may present challenges, organizing components within the SoS can be achieved effectively. This study recognizes two main systems that have an impact on the truck platooning industry. In addition to the truck platooning system, the trucking system is a system that contains different truck companies, infrastructure, and drivers. This paper focuses on examining the relationship between the truck platooning system and trucking system.

The system boundary adopted in this study is defined to focus on the endogenous interactions between the trucking system and the truck platooning adoption system, particularly those associated with technology transition, operational fuel savings, and resulting CO₂ emissions. Government and regulatory processes, infrastructure readiness, and other broader institutional interactions are treated as exogenous to the present model. This modeling choice is made to preserve tractability and to develop a parsimonious first-order representation of the principal feedback mechanisms governing platooning diffusion and its environmental implications. Nevertheless, these excluded factors may affect real-world adoption dynamics [48]. For example, regulatory approval, public incentives, infrastructure constraints, and communication system readiness can influence both the speed of technological uptake and the scale of achievable emission reductions. Accordingly, the results of this study should be interpreted as conditional on the external policy and infrastructure context represented implicitly in the model’s assumptions, rather than as universally transferable across all settings. In this sense, the present model provides a foundational system-level assessment, while future research should extend the boundary to explicitly incorporate regulatory, institutional, and infrastructure-related mechanisms.

4. Model Development

The problem addressed in this paper, which is often referred to as the reference mode in the SD literature [49], relates to the increasing trend of GHG emissions resulting from growing energy consumption, specifically of diesel, in the truck freight transportation system. According to EPA [6], the annual CO₂ emissions released from the transportation sector have shown an upward trend over the past three decades. As the reference mode changes over time, driven by the continued growth of freight demand, the reference mode exhibits dynamic behavior. Additionally, the complexity of the reference mode arises from a multitude of interdependent factors that influence the reference mode. We consider two model scenarios to capture the dynamics of the reference mode: one with truck platooning and the other without.

4.1. Casual Loop Diagram (CLD)

The theoretical framework of the platooning model scenario is conceptualized in a qualitative causal loop diagram, which illustrates the interactive relationships among the main factors within the SoS. Figure 1 displays this diagram, which depicts the relationship between the variables and two feedback loops. A link marked + indicates a positive relation where an increase (decrease) in the causal variable leads to an increase (decrease) in the effect variable. A link marked − indicates a negative relation where an increase (decrease) in the causal variable leads to a decrease (increase) in the effect variable. In addition, the loops are explained as follows:

• PT Fleet ⇒ Platooning Opportunities ⇒ Actual Platooning ⇒ Labor Saving ⇒ Incentive to Purchase/Transition Platoonable Trucks ⇒ PT Fleet (positive feedback loop).
• PT Fleet ⇒ Platooning Opportunities ⇒ Actual Platooning ⇒ Fuel Saving ⇒ Incentive to Purchase/Transition Platoonable Trucks ⇒ PT Fleet (positive feedback loop).

In Figure 1, the “R” signs indicate reinforcing impacts, characterized by a positive feedback loop, meaning that a change in one variable will lead to an amplified change of itself through the feedback loop.

4.2. Stock and Flow Diagram (SFD)

A CLD is transformed into a quantitative SFD, which provides an algebraic representation of models based on the causal loops identified [50]. SFDs consist of stocks (levels) and flows (rates), which serve as the fundamental elements of system dynamics models [51]. These components describe the state of the system under investigation and serve as the foundation for making decisions and taking actions [19]. Changes in stocks can solely occur through their corresponding flows, which indicate the quantities added to (inflow) or removed from (outflow) a stock as time progresses. Indeed, in an SFD, stocks are fundamental to generating behavior in a system; flows cause stocks to change. Figure 2 shows an SFD for the non-platooning scenario. In the figure, the stock “CT Fleet”, which stands for “conventional truck fleet”, has a one-stock two-flow structure, in which CT fleet is filled by purchases and drained by retirements. The total CO₂ emissions from the CT fleet are calculated based on the amount of fuel consumed by the CT fleet and the appropriate emission factor.

Table 1. Variables, units, and definitions of non-platooning model scenario.

Variable Name	Type	Unit	Definition
CT Fleet	level	vehicle	CT Fleet Initial + ${\sum }_{t=0}^{179}$ (CT Purchase$\left(t\right)$ − CT Retirement$\left(t\right)$)
Number of Drivers	auxiliary	person	Number of Drivers in the US (Time)
CT Purchase	auxiliary	vehicle per month	Drivers Without Trucks × Person to Truck Converter
Drivers Without Truck	auxiliary	person	Number of Drivers − CT Fleet × Truck to Person Converter
CT Retirement	rate	vehicle per month	CT Fleet/Expected Truck Lifetime
Fuel Consumption	rate	gallon per month	CT Fleet × Fuel Consumption per Mile × Mile Trip per Month per Truck
Total Fuel Consumption	level	gallon	${\sum }_{t=0}^{179}$ (Fuel Consumption$\left(t\right)$)
CO2 Emission	rate	kilograms of CO2 per month	Total Fuel Consumption × Emission Factor
Total CO2 Emissions	level	kilograms	${\sum }_{t=0}^{179}$ (CO2 Emission$\left(t\right)$)

details the employment of variables, and equations in the non-platooning SFD. Note that gray variables which are in angle brackets in the SFD, as well as throughout this paper, are referred to as “shadow variables” (shadow variables are valuable in modeling complex systems with interdependencies, enabling modelers to simplify complex equations for better clarity and manageability). On the other hand, Figure 3 shows the SFD for the platooning scenario. This SFD contains three submodels related to (1) platooning technology adoption, (2) drivers decision, and (3) CO₂ emissions.

4.2.1. Technology Adoption Submodel

The platooning technology adoption submodel, shown in Figure 3, has two stocks and five flows. In this submodel, each CT fleet is filled by purchases and drained by retirements and transitions to a platoonable truck fleet. Moreover, the stock “PT Fleet”, which stands for “platoonable truck fleet” is filled by purchase and the transition from a conventional fleet and drained by retirements. Indeed, in the technology adoption submodel, drivers face two options. Firstly, drivers without a truck may decide to purchase either a conventional truck or a platoonable truck based on the potential benefits associated with each option. Thus, drivers carefully weigh both the benefits and costs of each type of truck before making their decision.

Secondly, drivers with a conventional truck have the option to transform their vehicles into new platoonable trucks. This can be achieved by retrofitting an existing conventional truck with advanced technologies, such as radar sensors, lidar systems, and communication devices. Despite the cost incurred in transforming to platoonable trucks, owning such a truck can lead to benefits such as labor and fuel savings. Then, by converting potential future benefits and costs values of purchasing a platoonable truck to the present values (calculating present value (Pr) by using the present worth formula $Pr=\frac{A(-1+{(1+i)}^n)}{i{(1+i)}^n}$, in which A is the monthly benefit of having a platoonable truck, i is the monthly interest rate, and n is the average monthly lifetime of platoonale truck [52]), and dividing this amount by the present investment (cost of transitioning a conventional truck to a platoonable truck) that drivers need to pay currently, they can obtain the benefit over the cost of purchasing a platoonable truck and decide to buy one or stand with their current traditional truck.

Table 2. Variables, units, and definitions of platooning model scenario.

Submodel	Variable Name	Type	Unit	Definition
Technology adoption	CT Fleet	level	vehicle	CT Fleet Initial + ${\sum }_{t=0}^{179}$ (CT Purchase Rate(t) − CT Retirement Rate(t) − CT Transition to PT(t))
	CT Flee Initial	auxiliary	vehicle	Number of Drivers × Person to Truck Converter − PT Fleet Initial
	CT Purchase	rate	vehicle per month	Drivers without Truck × Person to Truck Converter × (1 – Probability Function(B over C))
	Drivers Without Truck	auxiliary	person	Person Converter − PT Fleet × Truck to Person Converter
	CT Retirement	rate	vehicle per month	CT Fleet/Expected Truck Lifetime
	CT to PT Transition	rate	vehicle per month	$exp(\mathrm{Transition}\mathrm{B}\mathrm{over}\mathrm{C}-1)/(1+exp(\mathrm{Transition}\mathrm{B}\mathrm{over}\mathrm{C}-1))\times \mathrm{CT}\mathrm{Fleet}$
	PT Fleet	level	vehicle	PT Fleet Initial + ${\sum }_{t=0}^{179}$ (PT Purchase(t) + CT Transition to PT Cost(t) − PT Retirement(t))
	PT Purchase	rate	vehicle per month	Drivers Without Truck × Person to Truck Convector× (Probability Function(B over C))
	PT Retirement	rate	vehicle per month	PT Fleet/Expected Truck Lifetime
CO2 emission	Fuel Saving	rate	gallon per truck per month	Avg Fuel Saving × PT Fleet/Fuel Price Dollar per Gallon
	Total Fuel Saving	level	gallon	${\sum }_{t=0}^{179}$ (Fuel Saving Rate(t))
	Fuel Consumption	rate	gallon per month	CT Fleet × Fuel Consumption per Mile × Mile Trip per Month per Truck + (Fuel Consumption per Mile) × Mile Trip per Month per Truck × (PT Fleet) − (Avg Fuel Saving × PT Fleet/Fuel Price Dollar per Gallon)
	Total Fuel Consumption	level	gallon	${\sum }_{t=0}^{179}$ (Fuel Consumption Rate(t))
	CO2 Emission Rate	rate	kilograms of CO2 released per month	Fuel Consumption Rate × Emission Factor
	Total CO2 Emissions	level	kilograms	${\sum }_{t=0}^{179}$ (CO2 Emission Rate(t))
Drivers Decision	Avg Fuel Saving	auxiliary	dollar per truck per month	Fuel Price Dollar per Gallon × Fuel Consumption per Mile × Mile Trip per Month per Truck × Fuel Saving per Platooning Mile per Truck × Platooning Opportunities
	Avg Labor Saving	auxiliary	dollar per truck per month	Mile Trip per Month per Truck × Dollar Value Break per Platooning Mile × Platooning Opportunities
	PT B over C	auxiliary	gallon per truck per month	(Avg Fuel Saving + Avg Labor Saving − Avg Ongoing Opportunistic Cost per Month) × (−1 + pow(1 + Interest Rate, Expected Truck Lifetime))/Interest Rate × pow(1 + Interest Rate, Expected Truck Lifetime)/(PT Cost − CT Cost)
	Platooning Opportunities	auxiliary	-	Probability of Proximity Trip Mile × Probability Path Consistency × Probability Successful Platooning × Probability Platooning Compatibility × Proportion of Road platoonability
	Probability Platooning Compatibility	auxiliary	-	PT Fleet/(PT Fleet + CT Fleet)
	Avg Ongoing Opportunistic Cost per Month	auxiliary	dollar per month per vehicle	Ongoing Opportunistic Coordination Cost per Mile × Mile Trip per Month per Truck
	Transition B over C	auxiliary	-	(Avg Fuel Saving + Avg Labor Saving − Avg Ongoing Opportunistic Cost per Month) × (−1 + pow(1 + Interest Rate, Expected Truck Lifetime))/Interest Rate × pow(1 + Interest Rate, Expected Truck Lifetime)/(CT to PT Transition Cost)

demonstrates variables, units, values, and definitions of adoption submodel.

4.2.2. Driver Decision Submodel

The truck drivers’ decision submodel, which can be seen in Figure 3, involves evaluating the benefit-over-cost ratio of either transitioning their current truck into a platoonable truck (“B over C Transition”) or persisting with their conventional truck. Furthermore, drivers who do not currently own trucks assess the benefit-over-cost ratio of purchasing a platoonable truck (“PT B over C”). These decisions are based on the potential benefits they may receive and the associated costs they may incur. There are two sources of benefit generated by platooning: fuel savings and labor savings for the follower trucks in the platoon. These savings are determined by the fuel price, labor wage, and the platooning opportunities available. The platooning opportunity is determined by the number of platooning trucks in the system and the network characteristics, which include the probability of path consistency among trucks, proximity to other trucks on the road, and the probability of being on a platoonable road in the network. Additionally, adopting a platoonable truck incurs two types of costs: the initial fixed cost associated with adoption (installation costs or the price difference between a platoonable truck and a conventional one) and the ongoing costs of platooning operations (assumed to be a constant portion of each trip). Table 2 shows the variables, units, values, and definitions of the driver decision process submodel.

In the present model, the driver decision-making mechanism is represented by a benefit-to-cost ratio derived from expert interviews with 29 transportation researchers. As detailed in Section 4.3, this ratio serves as an approximation of adoption incentives. This formulation is intended to capture the core economic logic that higher expected net benefits increase the likelihood of adopting truck platooning technology. The use of this representation is particularly appropriate in the absence of large-scale survey-calibrated behavioral data or revealed preference evidence on truck owners’ adoption decisions in this context. However, this specification may be a simplification of real-world decision making. In practice, truck owners may choose not to adopt even when the expected benefits exceed the expected costs, due to uncertainty, risk aversion, limited access to capital, organizational inertia, a lack of familiarity with the technology, or concerns regarding operational reliability. Accordingly, the current model should be interpreted as reflecting a first-order economic decision rule rather than being a complete behavioral model. This simplifying assumption may lead to a more responsive adoption pattern than would occur in practice, since the model does not explicitly represent delayed adoption or a “do nothing” behavioral alternative. Therefore, the estimated adoption dynamics should be interpreted with this limitation in mind.

4.2.3. CO₂ Emission Submodel

The CO₂ emission submodel is demonstrated in Figure 3. In this submodel, the total CO₂ emissions stock represents the amount of emissions released into the air, taking into account the interactions between the CV Fleet and PT Fleet stocks and reflecting the influence of the drivers’ decision-making process on this stock. The factors that affect this stock include the number of platoonable trucks and the percentage of fuel saving that each truck can achieve through platooning. The percentage of fuel saving resulting from platooning has been reported by various studies to range from 3% to 15% [53,54,55]. However, to maintain a conservative approach, we adopt a 5% fuel-saving assumption for trucks in a platoon within our model. It is worth noting that we assume that the emission factor is equal to 10.19 kg per gallon of diesel consumed in our models [56]. Table 2 shows further details on the variables, units, values, and definitions of the emission submodel.

4.3. Parameter Identification and Assumptions

In this section we present the input parameters are obtained based on data from the US road network, since our research focuses on the entire US road freight transportation network. We describe the numerical values given to the model’s parameters, based on data in journal publications [57,58] and information from additional sources such as leading trucking companies like Volvo [59], Scania [60], and Daimler [61]. Additionally, the authors’ engineering estimations have also been taken into account. The parameters values that are used in the non-platooning model and platooning submodels can be seen in the

Table 3. Parameters of platooning model scenario.

Parameter Name	Type	Unit	Value	Reference
PT Fleet Initial	constant	vehicle	1000	Assumption 1
CT Fleet Initial	constant	vehicle	13,859,181	[62,63]
Platoonable Truck Price	constant	dollar per vehicle	150,000	Assumption 2
Conventional Truck Price	constant	dollar per vehicle	135,000	[59,60,61]
Interest Rate (%)	constant	per year	5	[64]
CT to PT Transition Cost	constant	dollar per vehicle	11,000	Assumption 3
Fuel Price	constant	dollar per gallon	4	[65]
Fuel Saving per Platooning Mile per Truck (%)	constant	–	5	[53]
Probability Proximity Trip Mile a	constant	–	0.5	Assumption 4
Probability Path Consistency b	constant	–	0.5	Assumption 4
Probability Road Platoonability c	constant	–	0.6	[66]
Probability Successful Platooning d	constant	–	0.75	Assumption 4
Fuel Consumption per Truck	constant	gallon per mile	0.05	[67]
Ongoing Opportunity Coordinating Cost	constant	dollar	0.005	[40]
Model timescale	constant	month	180	Assumption 5
Hour Cost of Trucking	constant	dollar	10	[68]
Average Truck Speed	constant	mile per hour	50	[69]
Expected Truck Lifetime	constant	months	180	[70]
Emission Factor	constant	kilograms of CO2 released per gallon of diesel consumed	10.19	[56]

^a This parameter represents the probability of two platoonable trucks being in close proximity to each other. ^b This parameter shows the probability of two platoonable trucks having consistent paths while encountering each other during a trip. ^c This parameter indicates the probability of two platoonable trucks meeting on a road with platooning capabilities; approximately 60% of roads are deemed platoonable [66]. This determines the proportion of miles where trucks can sustain a speed of 50 mph or higher for more than 15 min. ^d This parameter explains the probability of successful platooning by considering two platoonable trucks that are in close proximity, have a consistent path, and are on a platoonable road.

.

Given the absence of empirical data on large-scale truck platooning, five model parameters are necessarily specified based on the informed assumptions in Table 3. The first assumption concerns the initial fleet of platoonable trucks, which is assumed to be 1000, reflecting the current lack of large-scale implementation. In SD modeling, a non-zero initial stock is mathematically required to activate the system’s feedback loops. At under 0.01% of the total US fleet, this negligible seed value accurately represents early real-world pilot programs and safely initializes the simulation without artificially inflating long-term emission projections. The second assumption relates to the additional cost associated with platoonable trucks. While the cost of purchasing a conventional truck is reported as $135,000 in the literature [71], a platooable truck is assumed to cost $150,000. The third assumption specifies a transition cost of $11,000 for retrofitting a conventional truck with platooning technology. As a key model parameter, this value is subjected to sensitivity analysis in the results section to evaluate its impact on system outcomes. Assumption four concerns operational matching probabilities. Parameters such as vehicle proximity and path consistency are set to a neutral baseline of 0.5, while the probability of successful platooning execution is set heuristically at 0.75. In the absence of empirical matching data from centralized coordination platforms, these neutral baseline values help prevent the overestimation of coordination efficiency. Operational matching probabilities will be further investigated in future research to empirically calibrate coordination efficiency.

The fifth assumption concerns the model’s timescale. A time horizon of 180 months (15 years) is adopted in this study. This duration reflects the typical service life of commercial heavy-duty trucks, which are designed for long-term operation and often exceed 750,000 miles over their lifetime. For regional operations averaging approximately 50,000 miles annually, this corresponds to about 15 years of use. This assumption is further supported by the U.S. EPA’s heavy-duty standards, which define the useful life of certain heavy-duty engine classes as 15 years [70]. Based on this, the same time horizon (180 months or 15 years) is adopted for our analysis and model timescale. Collectively, these assumptions are intended to reflect transparent input values for a scenario-based assessment, serving to explore system behavior under uncertainty rather than providing exact point predictions. Moving forward, future research should leverage field data collection, enterprise interviews, and survey-based behavioral studies to empirically calibrate these operational probabilities and enhance the model’s precision.

Furthermore, in another part of the platooning model, we require data for VMT and number of single and combination unit trucks for the next 180 month. As the data we aim to obtain is time-based, employing time series data prediction methods is essential to forecast the values until 2035. We use historical data relating to VMT and the number of SU and CU trucks in the US road network from 1970 to 2021 [62,63]. To generate the forecasts, we employ the AutoRegressive Integrated Moving Average (ARIMA) model, which is a widely used method for time series forecasting [72]. Figure 4 and Figure 5 depicts both the historical and forcasted values of VMT and the number of SU and CU trucks in the period of 1970 to 2035. ARIMA is selected over baseline alternative models due to its superior capability in handling the non-stationary, long-term upward trends inherent in the historical US data [73,74].

To verify predictive reliability and prevent overfitting, an out-of-sample validation approach is employed by partitioning the historical dataset (1970–2021) into a training set (80%) and a testing set (20%). A grid search algorithm systematically identified the optimal autoregressive (p), differencing (d), and moving average (q) parameters by minimizing the Akaike Information Criterion (AIC). The validation results demonstrate high predictive accuracy across the modeled variables. The VMT forecasts exhibited good precision, with the SU and CU VMT models yielding Mean Absolute Percentage Errors (MAPE) of 3.01% and 2.22%, respectively. The vehicle count forecasts also demonstrated strong reliability, yielding MAPEs of 6.34% for SU trucks and 9.93% for CU trucks. These error margins are well within the acceptable forecasting tolerances, particularly given the inherent macroeconomic volatility associated with long-term commercial fleet purchasing. Ultimately, the combination of minimized AIC scores and low out-of-sample errors verifies that the selected ARIMA models provide a robust, empirically validated baseline for the subsequent SD simulations.

Moreover, the adoption rate to the different levels of benefit-over-cost ratios related to buying a platoonable truck over the alternative case, i.e., buying a conventional truck, is mathematically modeled using a logistic function defined as $P\left(x\right)=1/(1+e^{-k(x-x_0)})$, where $P\left(x\right)$ is the probability of adoption and x is platoonable trucks’ benefit-to-cost ratio. To address behavioral uncertainty and establish a rigorous baseline, three adoption scenarios (optimistic, expected, and pessimistic) are constructed using data from structured expert interviews. During these interviews, 29 transportation researchers are asked to estimate adoption probabilities at 13 specific platoonable truck benefit-to-cost ratios (ranging from zero to six) for each scenario. Non-linear Least Squares (NLS) is used to fit the logistic function to these elicited data points, yielding smooth, continuous probability curves. The adoption curves of the three scenarios with regard to the benefit-over-cost ratios of the technology are depicted in Figure 6. The resulting inflection points ($x_0$) and growth rates (k) for the optimistic ($x_0=1.20;k=2.67$), expected ($x_0=1.93;k=1.97$), and pessimistic ($x_0=2.73;k=1.67$) scenarios provide a mathematically explicit and strictly reproducible representation of adoption behavior under varying economic incentives.

These scenarios are behavioral representations used to capture a plausible range of driver responses to the platoonable truck benefit-over-cost ratio of platooning technology. In the optimistic scenario, drivers are assumed to adopt platoonable trucks at relatively lower benefit over cost thresholds; in the expected scenario, adoption occurs at moderate thresholds; and in the pessimistic scenario, drivers require stronger economic incentives before adoption occurs. It is worth noting that truck platooning remains an emerging technology and that real-world operational data suitable for large-scale behavioral calibration are not yet available. As a result, although these structured interviews provide a robust expert consensus for analyzing system behavior under uncertainty, future work should refine these foundational curves using large-scale survey-based or field-based behavioral data from freight operators. In addition, by utilizing the standard logistic function (the standard logistic function is represented as $F\left(x\right)=\frac{e^x}{1+e^x}$, where x denotes the decision-making factor and, in this study, represents the benefit-to-cost ratio governing a truck driver’s likelihood to transition to platooning technology [75]), this framework maps the transition benefit-over-cost ratio directly into the probability of switching from a conventional truck to a platoonable truck.

5. Model Implementation and Results

The results present the general output and findings of the models for two considered scenarios: the non-platooning scenario and the platooning scenario. The study compares their outcomes by conducting a sensitivity analysis on the main parameters of the model to see is the variation in CO₂ reduces. The results of the presented models in this paper are obtained by using VENSIM PLE Version 9.4.2. All the experiments are conducted on a personal computer with Intel Core (TM) i7 3630QM 2.40 GHz CPU, 8 GB RAM, and Windows 10 operating system.

The simulation results indicate that the adoption of truck platooning technology over 180 months (15 years) for three different scenarios (optimistic, expected, and pessimistic) is projected to follow the patterns illustrated in Figure 7. This figure demonstrates that the time required to achieve the almost 50% market penetration of a platoonable truck fleet– with the adopted parameter values—is approximately 102, 134, and 168 months for the optimistic, expected, and pessimistic scenarios, respectively. This implies that as time progresses and the truck platooning fleet increases, the number of conventional trucks will decrease. These time points are highlighted by the circled intersections in the figure, which mark the moments when the platoonable truck fleet equals the conventional truck fleet in each scenario. Unsurprisingly, the diffusion rate of platooable trucks declines over time, as the pool of drivers who have not yet adopted such vehicles gradually decreases. This finding suggests that platooning technology is likely to play a crucial role in the trucking industry in the near future and has the potential to capture a significant market share from traditional trucks.

Additionally, the average labor-saving cost for these three scenarios is $25, $22, and $20 per month per truck at the end of the model timescale (180 months), respectively (Figure 8). However, as is evident from Figure 9, the fuel-saving values for each scenario are relatively modest, amounting to $3, $2.6, and $2.4 per month per truck, respectively. Moreover, the main factors driving truck drivers’ decisions to invest in platoonable trucks or platooning devices is the benefit-over-cost ratio and transition benefit-over-cost of the decision. Figure 10 illustrates that the benefit-over-cost ratio of purchasing a platoonable truck after 180 months is nearly equal to 3, 1.9, and 1 in the optimistic, expected, and pessimistic scenarios, respectively. This indicates that as time progresses, platoonable trucks offer a more advantageous decision compared to conventional trucks in both the optimistic and expected scenarios. However, in the pessimistic scenario, no clear preference can be observed since the benefit-over-cost ratio of purchasing a platoonable truck is equal to one after 15 years. That is to say, drivers are indifferent to buying a platoonable truck versus a conventional truck based on the pessimistic scenario. Furthermore, the transition benefit-over-cost is equal to 1.8, 1.2, and 0.8 at the end of the model timescale (Figure 11). In other words, as time moves forward, the benefit of purchasing a platoonable truck outweighs the cost of transforming a conventional truck into a platoonable one in the optimistic and expected scenarios.

Figure 12 shows that as the model’s timescale increases, the total fuel saving stock shows an upward trend in all scenarios. Additionally, by considering non-platooning as a baseline model scenario with no fuel saving (zero emission reduction), Figure 13 illustrates the total emission reduction based on three platooning scenarios which follows a growing trend as time progresses. The monotonic trends in Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13 naturally result from the system dynamics framework, as variables like the cumulative emissions and fleet adoption are stock accumulations that build progressively over time. The differentiated impacts of the conditional scenarios (optimistic, expected, and pessimistic) do not cause directional fluctuations; rather, they explicitly dictate the rate (steepness of diffusion) and absolute magnitude (final values) of these growth curves.

Furthermore, as a main finding of this paper, Figure 14 shows the comparison of the total CO₂ emission stock in the platooning and non-platooning scenarios. Despite our conservative assumption that only 5% fuel savings will result from platooning, Figure 14 indicates a substantial difference in the CO₂ emissions: nearly 292 million metric tons (MMT) after 180 months. To put this into context, the CO₂ reduction achieved through truck platooning is equivalent to the GHG emissions from 64,978,841 gasoline-powered passenger vehicles (passenger vehicles refer to two-axle four-tire vehicles, including passenger cars, vans, pickup trucks, and sport/utility vehicles; additionally, the average VMT and fuel efficiency of vehicles are approximately 11,520 miles per year and 22.9 miles per gallon [76] and, furthermore, the amount of carbon dioxide emitted per gallon of motor gasoline burned is $8.89\times {10}^{-3}$ metric tons [77]). Moreover, a significant portion of the energy consumed in homes is derived from the combustion of fossil fuels such as natural gas, propane, and oil. When these fuels are burned for electricity generation or to provide heating for homes, they release GHG, primarily CO₂, into the atmosphere. As a result, for further comparison, the amount of CO₂ reduction achieved by truck platooning can be expressed as the equivalent energy saved, which in this case would be the energy required to power 36,801,817 houses for a year (This calculation assumes that each home in the US emits 7.93 metric tons of CO₂ annually, comprising 5.139 metric tons of CO₂ from electricity, 2.29 metric tons from natural gas, 0.23 metric tons from propane, and 0.27 metric tons from fuel oil [78]. These values are based on the [78] report in which each home consumes almost 11,880 kWh of delivered electricity, 41,590 cubic feet of natural gas, 42 gallons of propane, and 25.6 gallons of oil. To convert electricity, natural gas, propane, and oil to equivalent released CO₂ emission, we use the following converters: 884.2 lbs of CO₂ per megawatt-hour [79], 0.0550 kg of CO₂ per cubic foot [80], 235.0 kg of CO₂ per barrel [81], and 426.1 kg of CO₂ per barrel [81], respectively).

To further contextualize the relative magnitude of this impact, this 292 MMT reduction represents an approximately 3.8% decrease in the total accumulative CO₂ emissions from the US trucking sector over the 15-year simulation period compared to the reference baseline. This system-level reduction aligns well with previous macro-level assessments. For instance, while isolated field tests often report localized fuel savings of 4% to 10% for following trucks, system-wide studies like Muratori et al. (2017) [66] suggest that the actual fleet-wide savings are heavily constrained by network platoonability and market penetration timelines. Our dynamic model mathematically bridges this gap. By accounting for the gradual diffusion of the technology and the probabilistic nature of finding a platooning partner on the road network, our results confirm that even a conservative assumption of 5% operational fuel savings translates to a significant reduction in aggregate national emissions. From a policy perspective, this demonstrates that truck platooning does not need to operate flawlessly at the maximum theoretical efficiency to deliver meaningful climate mitigation over the next decade.

Sensitivity Analysis

In this subsection, we investigate the sensitivity of CO₂ emissions to several parameters of the model during the diffusion of truck platooning technology. First, Figure 15 reports the sensitivity of the CO₂ emission reduction to fuel price for three scenarios of technological adoption. The vertical axis in the figure is equal to the amount of CO₂ savings after 15 years. As shown in the figure, the changes in the amount of CO₂ saving is more significant, with a higher fuel price for all the mentioned scenarios. For instance, when the fuel price is equal to $10 per gallon, the CO₂ savings after 15 years amount to 13.4, 12.7, and 12.6 MMT for the optimistic, expected, and pessimistic scenarios, respectively. Interestingly, even when the fuel price tends towards zero, the amount of CO₂ emission reduction due to truck platooning remains substantial, exceeding 10.5 MMT for all the scenarios. This is attributed to the additional benefits of the technology for truck drivers beyond fuel savings, like labor savings. Furthermore, within the simulated range of $0 to $10 per gallon, the amount of CO₂ emission reduction exhibits a quasi-linear positive trend with increasing fuel price. It should be noted, however, that the fundamental system behavior is non-linear. Because technological adoption is bounded by the finite size of the commercial fleet and governed by a logistic function, the marginal increase in CO₂ emission reduction will eventually diminish at extreme fuel prices as market penetration reaches saturation. Nevertheless, within practical, real-world pricing scenarios, this finding suggests that raising fuel prices in the trucking industry can significantly contribute to reducing CO₂ emissions by encouraging the more rapid diffusion of truck platooning technology.

Figure 16 illustrates the sensitivity of CO₂ emission reduction to the transition cost from a conventional fleet to a platoonable fleet. The results indicate a rapid decrease in CO₂ emission reduction, declining from approximately 12 MMT to around 5 MMT, as the transition cost increases from $1500 to $10,000 for the optimistic, expected, and pessimistic scenarios of technological adoption. Furthermore, the figure shows that the changes in CO₂ emission reduction within the range of transition costs from $10,000 to $25,000 are not significantly impactful. Observing the trend of change, it becomes evident that the decrease in CO₂ emission reduction follows an exponential pattern in response to the transition costs. This behavior can be attributed to the influence of the economy’s interest rate. Figure 17 shows the sensitivity of the CO₂ emission reduction to fuel efficiency of the platooning technology. Note that with the current technology, the fuel saving is around 5% for the leading truck and around 10% for the following trucks [82]. The figure suggests that with one increasing the fuel saving of the technology, the CO₂ emission reduction would be increased by around 3 MMT for the investigated technological adoption scenarios. Finally, Figure 18 shows the sensitivity of CO₂ emission reduction to the interest rate of the economy. It shows an almost linear pattern of decreasing the CO₂ emission reduction with increasing the annual interest rate of the economy.

6. Conclusions

The imminent adoption of truck platooning technology in the US freight sector presents a critical opportunity to mitigate greenhouse gas emissions, yet its actual climate impact depends heavily on complex, dynamic adoption factors. To address this, the present study leverages a SD framework to comprehensively model the diffusion of platooning technology and its environmental impacts over a 180-month period. By developing causal loop and stock and flow diagrams, we contrast a baseline non-platooning system with a platooning scenario that explicitly incorporates truck drivers’ economic decision making based on the benefit-to-cost ratio of technology adoption. Evaluating the system under optimistic, expected, and pessimistic adoption scenarios, our simulation reveals that truck platooning achieves a substantial reduction of approximately 292 MMT of CO₂ emissions. Furthermore, extensive sensitivity analyses on fuel price, transition costs, technology efficiency, and interest rates provide a deeper understanding of the system’s tipping points, establishing a clear foundation for targeted policy interventions.

The findings from the system dynamics model and subsequent sensitivity analyses offer three key insights for policymakers and freight stakeholders seeking to accelerate the diffusion of truck platooning and maximize its environmental benefits. While these recommendations are derived from a generalized macro-level model, they highlight the critical levers that can influence technology adoption:

• Targeted subsidies for transition costs: The model demonstrates that CO₂ emission reductions decrease rapidly as the cost to transition a conventional truck to a platoonable one increases from $1500 to $10,000. However, for costs exceeding $10,000, the decrease in emissions becomes more gradual. Policymakers should consider implementing financial incentives, such as tax credits or direct grants, designed specifically to reduce the effective out-of-pocket transition costs for fleet operators to below the $10,000 threshold.
• Fuel pricing and carbon mechanisms: Sensitivity analysis reveals a strong positive correlation between fuel price and emission reduction within practical pricing bounds. Although this relationship is ultimately non-linear and will diminish as market penetration approaches saturation, raising effective fuel prices in the trucking industry can still significantly contribute to reducing CO₂ emissions by encouraging the more rapid diffusion of truck platooning technology. Implementing carbon taxes or cap-and-trade systems could serve as a powerful indirect policy tool to accelerate this market penetration.
• Favorable financing structures: The model shows an almost linear pattern of decreasing CO₂ emission reduction with an increasing annual interest rate in the economy. State or federal programs that provide low-interest loans for fleet modernization could enhance the benefit-over-cost ratio for prospective adopters, facilitating faster uptake of the technology.

We acknowledge that the present study presents just the beginning of understandings of the climate impact of truck platooning. There exist four limitations which may warrant further investigations. First, the analysis pertains to a generalized network system, with no differentiation of the nuances across local and interstate road segments as well as truck speed variations on these segments. Future studies may consider the integration of some agent-based simulations to better reflect route and traffic details on the road network studied. Second, our current work does not explore the impact of truck spacing while platooning and of platoon size on fuel saving. These aspects warrant further investigation to comprehensively understand their contributions to the overall fuel efficiency of truck platooning. Third, the sensitivity analysis in this study adopts a one-factor-at-a-time approach, which isolates individual variable effects but does not capture interactions among multiple factors. Future research should apply multidimensional scenario analyses to examine the joint sensitivities of parameters such as fuel price, transition cost, and fuel-saving efficiency. Fourth, while this research provides a foundational understanding of the climate impacts of truck platooning, the policy recommendations derived herein should be viewed within the scope of a generalized network model. Specifically, the results do not yet account for speed variations across different road types or the road capacity effects that may arise from denser traffic flow. Furthermore, variables such as the potential shift toward electric platoonable trucks could significantly alter the magnitude of CO₂ mitigation. Consequently, while our findings support the promotion of platooning through fuel pricing and transition subsidies, the actual real-world efficacy of these policies may vary based on local infrastructure and specific fleet configurations. These factors represent vital directions for future research to refine the precision of sustainable freight policies.