A Multi-Criterion Approach to Smart EV Charging with CO 2 Emissions and Cost Minimization

Giuseppe C. Calafiore, Luca Ambrosino, Khai Manh Nguyen · 2025

Read A Multi-Criterion Approach to Smart EV Charging with CO 2 Emissions and Cost Minimization by Giuseppe C. Calafiore, Luca Ambrosino, Khai Manh Nguyen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-06131.txt.

Introduction

Giuseppe Calafiore and co-authors charge electric vehicles on a fossil-heavy grid while watching both the bill and the carbon. Smart charging can shift demand, but its climate value depends on when the grid burns what. The team couples a simple grid dispatch model, with hydro, thermal, and renewable sources, to a charging scheduler.

Hourly carbon intensity from the dispatch feeds a carbon-cost signal, combined with a local time-of-use tariff. Sweeping one weight traces the trade-off frontier between cost and emissions. In a three hundred vehicle scenario based on Vietnam data, their scheduler kept the full bill saving of a cost-only optimizer while cutting charging emissions. A sensitivity study on the hydro budget showed the scheduler stayed cheaper and cleaner than first-come charging throughout.

Catalog record

CREATORGiuseppe C. Calafiore, Luca Ambrosino, Khai Manh Nguyen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-06131

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Multi-Criterion Approach to Smart EV Charging with CO 2 Emissions and Cost Minimization Thanks: L. Ambrosino and G. C. Calafiore are with the Department of Electronics and Telecommunications, Politecnico di Torino, Italy. K. M. Nguyen, M. B. Vu and L. El Ghaoui are with VinUniversity, Hanoi, Vietnam. R. Zorgati is with EDF Lab Paris-Saclay, France. Giuseppe C. Calafiore, Luca Ambrosino, Khai Manh Nguyen Abstract We study carbon-aware smart charging in a fossil-dominated grid by coupling a simplified hydro-thermal-renewable dispatch model with a tractable linear charging scheduler. The case study is informed by Vietnam’s regional data. Thermal units remain dominant, renewables are time-varying, and hydropower is modeled through a single reservoir budget. From the day-ahead dispatch we derive hourly carbon intensity and a corresponding carbon-cost signal; these are combined with a local time-of-use tariff in the EV charging problem. The resulting weighted-sum linear program is multi-objective: by sweeping the trade-off coefficient, we recover the supported Pareto frontier between electricity cost and charging-associated emissions. In a 300-EV public-charging scenario with a 0.8 MW feeder cap, the proposed carbon-aware scheduler preserves the 19.8% bill reduction of a cost-only optimizer while lowering charging-associated emissions by 7.3%; a more carbon-focused tuning still remains 12.6% cheaper and 9.3% cleaner than a FIFO baseline. A hydro-sensitivity study shows that changing the reservoir budget by ± \pm 20% moves the mean grid carbon intensity from 360 to 466 g/kWh, yet the carbon-aware scheduler remains consistently cheaper and cleaner than FIFO. The dispatch and charging LPs solve in few milliseconds on a standard desktop computer, showing that the framework is lightweight enough for repeated day-ahead studies. I Introduction Electric vehicles can strongly reduce transport emissions, but the grid-side benefit depends on when charging takes place and on the carbon content of the electricity actually used [ 1 ] . This is particularly relevant in systems where coal and gas still provide a large share of daily generation. In such contexts, a charging strategy that reacts only to electricity prices may shift demand toward hours that are cheap for the operator but not necessarily clean for the environment. The two research threads behind this paper are well established. On the supply side, unit-commitment and dispatch formulations with emission-aware objectives are standard tools for studying cost–carbon trade-offs in hydro-thermal and renewable-rich systems [ 2 , 3 ] . On the demand side, smart EV charging is often posed as a linear or convex scheduling problem that reallocates flexible charging load across time [ 4 , 5 , 6 ] . A parallel literature shows that the environmental impact of charging depends strongly on marginal or time-varying supply conditions: Ontario and Great Britain studies quantify large timing effects [ 7 , 8 ] , while optimization-based formulations schedule EV demand directly against marginal-emissions or real-time carbon-intensity signals [ 9 , 10 ] . The gap we address is practical rather than theoretical: public charging operators often need a lightweight workflow that turns system-level generation information into a carbon-aware charging signal without solving one large, tightly coupled market-clearing problem. We therefore adopt a three-step pipeline. First, a simplified hourly dispatch model computes a least-cost Vietnam-inspired hydro-thermal-renewable mix over one representative day. Second, the dispatch is converted into hourly carbon intensity and carbon cost. Third, these quantities are combined with a local time-of-use (TOU) tariff in a linear charging scheduler. The contribution is not a new class of unit-commitment models; rather, it is a compact and reproducible way to merge (i) thermal, renewable and hydro resources, (ii) a locally consistent tariff instead of foreign wholesale prices, (iii) an optimization-based cost-only baseline besides FIFO, (iv) an explicit Pareto analysis, and (v) computational-burden reporting. The decoupled architecture is appropriate when the charging station behaves as a price taker . In the experiments, the station peak is 0.8 MW whereas the representative system peak is 35.5 GW, hence the charging hub accounts for only 2.25\times 10^{-5} of system peak demand. For larger aggregations, the coupling would need to be iterated or formulated as a bilevel/fixed-point problem; we return to this in the discussion. II Vietnam-Inspired Dispatch Layer II-A Generation mix and representative profiles The supply-side model integrates thermal sources, renewables, hydropower, imports, and associated CO 2 emissions. Table I reports the source shares used to scale installed capacity from official Vietnam Electricity (EVN, the state-owned national electric utility of Vietnam) data, together with dispatch costs and life-cycle emission factors from the sources used in the original study [ 11 , 16 ] . Monetary quantities are handled in \mathrm{VND/MWh} . TABLE I: Source data used in the dispatch LP. Source Cap. share [%] Cost [ 10^{6} VND/MWh] CO 2 [g/kWh] Coal 33.2 2.100 820 Gas 8.9 1.428 490 Fuel 1.4 3.000 740 PV 18.55 2.046 48 Wind 8.25 2.086 12 Hydro 28.4 1.128 24 Import 1.2 2.200 300 The demand trajectory is a representative 24-hour system profile spanning 26.1–35.5 GW. Photovoltaic (PV) availability is modeled through a daylight bell shape, while wind follows a milder diurnal profile consistent with Vietnam-scale wind modeling based on reanalysis-driven virtual wind farms [ 12 ] . Hydropower is aggregated into a single energy-equivalent reservoir, following the usual virtual-reservoir simplification employed in system studies and motivated by the importance of hydro flexibility in Vietnam [ 13 ] . This aggregation keeps the hydro discussion in the paper while avoiding plant-level binaries that are not needed for the charging application. Thermal units (coal, gas, fuel, oil) are assumed fully available up to their aggregate capacity, so the thermal layer plays the role of the dispatchable backbone. Fuel oil therefore appears as a costly emergency option rather than a routinely used source. Renewable production is curtailable and no stationary battery is included, which means that midday PV surplus cannot be shifted to the evening peak. The hydro parameters are set to s_{0}=110 GWh, \underline{s}=30 GWh, \overline{s}=160 GWh, with a flat inflow of 7 GWh/h; these values keep hydro energy in the same order of magnitude as its annual contribution while still leaving enough freedom to arbitrage between midday and evening demand. The representative-day retail electricity price profile in Figure 1 is constructed from the Vietnam Electricity business-customer retail tariff under Decision No. 1279/QD-BCT dated 9 May 2025, using the published off-peak, standard, and peak rates together with EVN’s time-of-use period definitions; the official half-hour boundaries are mapped to our hourly discretization, [ 15 , 17 ] . For the weekday representative day considered here, off-peak hours are taken overnight, peak hours occur in the late morning and early evening, and the remaining hours are standard-price periods, consistent with EVN’s published TOU schedule [ 17 ] . Fig. 1: Representative-day retail electricity price, based on the EVN business-customer time-of-use tariff. Fig. 2: Representative demand and renewable availability (top), and optimal hourly dispatch mix (bottom). II-B Hydro-thermal-renewable dispatch model Let p_{t,j} denote the power produced at hour t\in\{1,\ldots,24\} by source j\in\mathcal{G}:=\{ coal, gas, fuel, PV, wind, import, hydro \} . For thermal units and imports, 0\leq p_{t,j}\leq\bar{p}_{t,j} with time-invariant upper bounds. For PV and wind, \bar{p}_{t,j} is hour-dependent. Hydropower output is denoted h_{t}:=p_{t,\mathrm{hydro}} , and the reservoir state s_{t} is measured in energy-equivalent MWh. Problem ( 1 ) below is the day-ahead least-cost dispatch model:

Read the complete text