Whoosh: Journal of Transportation and Logistics

Decarbonizing Motorcycle-Based Last-Mile Delivery in Jakarta: A Jakarta-Calibrated Scenario Simulation of Routing, Consolidation, and Fleet Electrification

2026-08-09T14:57:54+00:00

Urban last-mile delivery is increasingly shaped by e-commerce growth, rapid-delivery expectations, and motorcycle-based courier operations. Jakarta is a relevant setting because motorcycles dominate road traffic and two-wheelers are widely used for last-mile delivery. This study develops a Jakarta-calibrated scenario simulation to estimate operational carbon dioxide (CO2) emissions under alternative combinations of route optimization, micro-hub consolidation, pickup-point delivery, and electric-motorcycle adoption. The model represents an illustrative 500-courier fleet only as a scaling device; daily courier distance is calibrated at 70 km/day, the midpoint of the 60-80 km/day range reported for Indonesian two-wheeler last-mile couriers, and 300 operating days/year is treated as an analytical normalization rather than an observed company schedule. Baseline emissions are therefore also reported per courier-year. Scenario VKT reductions are literature-informed inputs rather than outputs of an independently solved Vehicle Routing Problem or GIS model. To avoid double counting in the integrated scenario, individual distance-reduction effects are compounded and constrained by an explicit 40% VKT-reduction cap. Electric-motorcycle electricity use is set at 0.025 kWh/km, with a 10% charging-loss allowance, and electricity emissions are calculated using a Jakarta-relevant Java-Madura-Bali (JAMALI) grid factor of 0.87 kg CO2/kWh. Under the calibrated baseline, the illustrative fleet travels 10.50 million km/year, consumes 300,000 L of gasoline, and emits 693.0 tCO2/year (1.386 tCO2 per courier-year). Conditional on the scenario inputs, route optimization, micro-hub batching, and pickup-point consolidation correspond to 12.0%, 18.0%, and 25.0% lower emissions, respectively. A route-optimization plus 40% electric-motorcycle scenario produces an estimated 38.9% reduction, while the integrated scenario produces an estimated 62.9% reduction and 70.8% lower energy expenditure. These values should be interpreted as scenario-contingent estimates, not observed treatment effects. The analysis indicates that reducing avoidable vehicle-kilometers before electrifying the remaining travel is a promising decarbonization sequence for motorcycle-dominant urban delivery systems.

Managing Product Returns at the Source: Associations between SCOR Processes and Return Performance - Multimethod Evidence from the Gaziantep Carpet Manufacturing Industry

2026-08-23T14:32:32+00:00

Product returns are not merely a post-sale reverse-logistics activity; they can also represent the visible outcome of deficiencies in planning, sourcing, production, delivery, and supply chain governance. This study examines associations between the Plan, Source, Make, Deliver, and Enable processes in the SCOR Model 12.0 and a perception-based Return construct and identifies processes that merit managerial attention for return prevention. A multimethod design is employed. A total of 119 valid questionnaires collected from carpet manufacturing firms operating in Gaziantep, Türkiye, were analyzed via exploratory factor analysis, correlation analysis, and multiple linear regression. Evaluations of 28 SCOR metrics obtained from 10 supply chain managers were then weighted via the entropy method, and semistructured interviews with four stakeholders connected to one large carpet manufacturer were used for contextual interpretation. Because the survey is cross-sectional and all the constructs were measured from the same respondents at one measurement occasion, the regression coefficients are interpreted as associations rather than causal effects, and common-method variance cannot be ruled out. The main regression model accounts for 75.7% of the variance in the Return construct (R² = .757; adjusted R² = .747; p < .001). Plan is the strongest predictor (β = -.559; p < .001), followed by Enable (β = -.253; p = .004), Make (β = -.200; p = .017), and Deliver (β = -.170; p = .021); Source is not statistically significant (β = -.091; p = .115). Exploratory item-level regressions identify supply chain data and information management (β = -.668; p < .001) and engineer-to-order production (β = -.546; p < .001) as the strongest nominal associations, but these item-level p values are unadjusted for multiple testing and are treated as hypothesis-generating. Within the 10-manager panel, entropy-derived weights rank Plan first, and an indicator-count-normalized sensitivity check also leaves Plan first, although the ordering of intermediate processes changes. Overall, the findings are consistent with a preventive, upstream approach to returns centered on planning, information governance, customized production, and fulfillment control.

Artificial Intelligence Readiness and Carbon Productivity in Freight Transport: Evidence from OECD Countries

2026-08-23T14:32:16+00:00

This study examines whether national artificial intelligence (AI) readiness is associated with a freight-oriented transport carbon productivity proxy in OECD countries. Using a strongly balanced panel of 28 countries for 2020-2022, the dependent variable relates road-and-rail freight transport work, measured in million tonne-kilometers, to national transport-related CO2 emissions. Government AI Readiness is the main explanatory variable, with GDP per capita, renewable energy use, internet penetration, urbanization, and trade openness included as controls. Pooled OLS, fixed-effects (FE), and random-effects (RE) models are estimated with country-clustered standard errors. Correlated random effects/Mundlak, wild cluster bootstrap, and exploratory interaction analyses are added to assess estimator choice, small-cluster inference, and possible complementarity with renewable energy and internet connectivity. The available 2020-2022 data do not identify a statistically significant contemporaneous association between AI readiness and the carbon-productivity proxy. This conclusion is unchanged under within-between decomposition and wild cluster bootstrap inference, while the interaction terms are also statistically insignificant. These null estimates should not be interpreted as evidence of no effect because the short panel, limited within-country variation, construct and outcome measurement limitations, and potentially lagged effects constrain statistical identification.