As the world grapples with climate change, the transition to sustainable transportation has become imperative. In China, the rapid adoption of electric cars is a cornerstone of the national strategy to achieve carbon peak by 2030 and carbon neutrality by 2060. The China EV market has seen exponential growth, with millions of electric cars on the roads, driven by government incentives and rising environmental awareness. However, the widespread use of electric cars presents challenges, particularly in managing their charging loads to minimize indirect carbon emissions and grid instability. Indirect carbon emissions refer to the CO2 released during electricity generation to power these vehicles, which can be substantial if charging occurs during peak hours when grid carbon intensity is high. This paper explores a time-of-use electricity pricing strategy tailored for electric cars in China, aiming to reduce these emissions and balance grid loads through optimized pricing models.
The proliferation of electric cars in China has led to concerns about their impact on the power grid. Uncoordinated charging of electric cars can exacerbate peak loads, increasing the risk of grid failures and higher carbon emissions. For instance, when electric cars charge during evening hours, they often align with residential electricity demand, creating a “peak-on-peak” effect. This not only strains the grid but also results in higher indirect carbon emissions because power plants may rely on fossil fuels to meet the surge. To address this, I propose a time-of-use pricing model that incentivizes electric car owners to charge during off-peak hours or when renewable energy sources are abundant. This approach leverages economic signals to shift charging behavior, thereby reducing the carbon footprint of electric cars and enhancing grid reliability. The China EV sector stands to benefit significantly from such strategies, as they align with national carbon reduction goals and promote sustainable urban mobility.
In this study, I employ a multi-objective optimization framework to design time-of-use electricity prices for electric cars. The model incorporates dynamic carbon emission factors, which vary hourly based on the grid’s energy mix, to accurately quantify indirect emissions. By simulating uncontrolled charging using Monte Carlo methods and then applying a genetic algorithm for optimization, I derive pricing schemes that minimize both carbon emissions and grid peak-valley differences. The results demonstrate that strategic pricing can effectively guide electric car charging patterns, leading to substantial environmental and operational benefits. This research contributes to the growing body of knowledge on smart charging for electric cars, emphasizing the importance of integrating carbon considerations into pricing mechanisms for the China EV market.
Simulating Electric Car Charging Demand Using Monte Carlo Methods
To understand the baseline charging behavior of electric cars, I first simulate uncontrolled charging scenarios using Monte Carlo methods. This approach models the stochastic nature of electric car usage, including variables such as charging start time, initial state of charge (SOC), and daily driving distance. The charging start time for electric cars is assumed to follow a normal distribution, based on transportation survey data, with a probability density function given by:
$$ f_s(t_s) = \begin{cases}
\frac{1}{\sigma_{t_s} \sqrt{2\pi}} \exp\left[-\frac{(t_s – \mu_{t_s})^2}{2\sigma_{t_s}^2}\right], & \mu_{t_s} – 12 < t_s \leq 24 \\
\frac{1}{\sigma_{t_s} \sqrt{2\pi}} \exp\left[-\frac{(t_s + 24 – \mu_{t_s})^2}{2\sigma_{t_s}^2}\right], & 0 < t_s \leq \mu_{t_s} – 12
\end{cases} $$
where \( t_s \) is the charging start time in hours, \( \mu_{t_s} = 17.6 \), and \( \sigma_{t_s} = 3.4 \). This reflects typical patterns where electric cars are charged after daily commutes. The daily driving distance for electric cars follows a log-normal distribution, with the probability density function:
$$ f_d(x) = \frac{1}{x \sigma_d \sqrt{2\pi}} \exp\left[-\frac{(\ln x – \mu_d)^2}{2\sigma_d^2}\right] $$
where \( x \) is the daily distance in kilometers, \( \mu_d = 3.2 \), and \( \sigma_d = 0.88 \). The required charging time \( T_c \) for an electric car is calculated based on the energy consumed per kilometer, charging power, and efficiency:
$$ T_c = \frac{W \cdot x}{P_c \cdot \eta_c} $$
Here, \( W = 0.15 \) kWh/km is the unit energy consumption, \( P_c = 7 \) kW is the charging power for slow charging, and \( \eta_c = 0.9 \) is the charging efficiency. For a fleet of \( N = 1000 \) electric cars, the total charging power at time \( t \) is the sum of individual charging powers:
$$ P_{\text{sum}}(t) = \sum_{i=1}^{N} P_{c_i} \cdot x_i^t $$
where \( x_i^t \) is a binary variable indicating whether car \( i \) is charging at time \( t \). The Monte Carlo simulation involves generating random samples for each electric car’s parameters and aggregating the loads over 24 hours. The resulting load profile for uncontrolled charging shows significant peaks during evening hours, coinciding with high grid demand and carbon intensity. This simulation provides a baseline for evaluating the impact of time-of-use pricing on electric car charging behavior in the China EV context.
| Parameter | Symbol | Value | Description |
|---|---|---|---|
| Charging Start Time Mean | \( \mu_{t_s} \) | 17.6 h | Average time when charging begins |
| Charging Start Time Std Dev | \( \sigma_{t_s} \) | 3.4 h | Standard deviation of start time |
| Daily Distance Mean | \( \mu_d \) | 3.2 km | Log-normal mean for distance |
| Daily Distance Std Dev | \( \sigma_d \) | 0.88 km | Log-normal standard deviation |
| Charging Power | \( P_c \) | 7 kW | Slow charging power for electric cars |
| Charging Efficiency | \( \eta_c \) | 0.9 | Efficiency of charging process |
The uncontrolled charging load profile for electric cars, derived from the Monte Carlo simulation, highlights the need for intervention. Peak loads occur between 18:00 and 22:00, aligning with residential electricity use. This not only increases the grid’s peak-valley difference but also raises indirect carbon emissions due to higher carbon intensity during these hours. For the China EV market, addressing this issue is crucial to maximizing the environmental benefits of electric cars. By shifting charging to periods with lower carbon intensity, such as overnight or when renewable generation is high, we can significantly reduce the overall carbon footprint of electric car usage.
Modeling Time-of-Use Electricity Pricing with Carbon Emission Considerations
The core of this research is the development of a time-of-use electricity pricing model that incorporates carbon emission factors. The model aims to optimize pricing for electric cars to achieve two primary objectives: minimizing indirect carbon emissions and reducing the grid’s peak-valley load difference. Indirect carbon emissions from electric car charging are calculated using dynamic carbon emission factors, which vary by hour based on the grid’s energy mix. Unlike average emission factors, dynamic factors provide temporal precision, allowing for more accurate carbon accounting. For example, in regions like Jiangsu, China, the carbon emission factor per kWh can range from 568 gCO2/kWh to over 610 gCO2/kWh throughout the day, as shown in Table 2.
| Time Period | Carbon Emission Factor (gCO2/kWh) |
|---|---|
| 00:00 – 01:00 | 568.73 |
| 01:00 – 02:00 | 580.68 |
| 02:00 – 03:00 | 591.52 |
| 03:00 – 04:00 | 588.05 |
| 04:00 – 05:00 | 600.44 |
| 05:00 – 06:00 | 599.52 |
| 06:00 – 07:00 | 611.03 |
| 07:00 – 08:00 | 611.41 |
| 08:00 – 09:00 | 599.60 |
| 09:00 – 10:00 | 601.42 |
| 10:00 – 11:00 | 580.03 |
| 11:00 – 12:00 | 579.84 |
| 12:00 – 13:00 | 568.67 |
| 13:00 – 14:00 | 581.34 |
| 14:00 – 15:00 | 590.38 |
| 15:00 – 16:00 | 601.97 |
| 16:00 – 17:00 | 591.37 |
| 17:00 – 18:00 | 601.14 |
| 18:00 – 19:00 | 609.54 |
| 19:00 – 20:00 | 609.34 |
| 20:00 – 21:00 | 598.65 |
| 21:00 – 22:00 | 591.12 |
| 22:00 – 23:00 | 591.83 |
| 23:00 – 00:00 | 579.56 |
The first objective function \( f_1 \) minimizes the total indirect carbon emissions from electric car charging over 24 hours:
$$ f_1 = \min \sum_{t=1}^{T} (e_t \cdot P_{t,EV} \cdot \Delta t) $$
where \( e_t \) is the dynamic carbon emission factor at time \( t \), \( P_{t,EV} \) is the charging load of electric cars at time \( t \), and \( \Delta t \) is the time interval (e.g., 1 hour). The second objective function \( f_2 \) minimizes the peak-valley difference of the grid load, which includes both the base load and the electric car charging load:
$$ f_2 = \min \frac{1}{T-1} \sum_{t=1}^{T} (P_{t,0} + P_{t,EV} – \bar{P}_s)^2 $$
Here, \( P_{t,0} \) is the base grid load at time \( t \), and \( \bar{P}_s \) is the average total load over the period. This function aims to flatten the load curve, reducing stress on the grid and improving stability for the China EV infrastructure.
To model the response of electric car owners to price changes, I use a demand price elasticity matrix. This matrix captures how charging demand shifts between time periods in response to variations in electricity prices. The elasticity matrix \( E \) is defined as:
$$ E = \begin{bmatrix}
\epsilon_{11} & \epsilon_{12} & \cdots & \epsilon_{1n} \\
\epsilon_{21} & \epsilon_{22} & \cdots & \epsilon_{2n} \\
\vdots & \vdots & \ddots & \vdots \\
\epsilon_{n1} & \epsilon_{n2} & \cdots & \epsilon_{nn}
\end{bmatrix} $$
where \( \epsilon_{kk} \) is the self-elasticity, measuring the change in demand in period \( k \) due to a price change in the same period, and \( \epsilon_{km} \) is the cross-elasticity, measuring the demand change in period \( k \) due to a price change in period \( m \). The elasticity coefficients are calculated as:
$$ \epsilon_{kk} = \frac{\Delta P_k / P_k}{\Delta C_k / C_k} $$
$$ \epsilon_{km} = \frac{\Delta P_k / P_k}{\Delta C_m / C_m} $$
where \( \Delta P_k \) and \( P_k \) are the change and initial demand in period \( k \), and \( \Delta C_k \) and \( C_k \) are the change and initial price. After implementing time-of-use pricing, the new demand for electric car charging can be expressed as:
$$ \begin{bmatrix}
P’_1 \\
P’_2 \\
\vdots \\
P’_n
\end{bmatrix} = \frac{1}{n} \begin{bmatrix}
P_1 & 0 & \cdots & 0 \\
0 & P_2 & \cdots & 0 \\
\vdots & \vdots & \ddots & \vdots \\
0 & 0 & \cdots & P_n
\end{bmatrix} \cdot E \cdot \begin{bmatrix}
\Delta C_1 / C_1 \\
\Delta C_2 / C_2 \\
\vdots \\
\Delta C_n / C_n
\end{bmatrix} + \begin{bmatrix}
P_1 \\
P_2 \\
\vdots \\
P_n
\end{bmatrix} $$
This equation allows us to predict how charging loads for electric cars will shift under different pricing schemes, enabling the optimization of prices to achieve the desired outcomes.
Constraints and Optimization Algorithm
The time-of-use pricing model for electric cars includes several constraints to ensure practicality and feasibility. First, the total grid load, including electric car charging, must not exceed the grid’s capacity:
$$ 0 \leq P_{t,0} + P_{t,EV} \leq P_{g,\max} $$
where \( P_{g,\max} \) is the maximum allowable load. Second, the number of electric cars charging simultaneously is limited by the available charging infrastructure:
$$ 0 \leq K_t \leq K_{t,\max} $$
where \( K_t \) is the number of cars charging at time \( t \), and \( K_{t,\max} \) is the maximum capacity. Third, the state of charge (SOC) of each electric car must meet the user’s expected level by the end of charging:
$$ \text{SOC}_{i,\text{desire}} \leq \text{SOC}_{i,\text{end}} \leq \text{SOC}_{i,\max} $$
This ensures that electric cars have sufficient charge for their next trip. Finally, the electricity prices must remain within feasible bounds to balance user affordability and utility profitability:
$$ C_{\min} \leq C_t \leq C_{\max} $$
For example, in this study, the base price is set at 1 CNY/kWh, with a minimum of 0.25 CNY/kWh and a maximum of 2 CNY/kWh for electric car charging in the China EV market.
To solve this multi-objective optimization problem, I employ the Non-dominated Sorting Genetic Algorithm II (NSGA-II). This algorithm is well-suited for handling multiple objectives and constraints, as it efficiently explores the Pareto front of non-dominated solutions. The NSGA-II process involves initializing a population of candidate solutions (i.e., sets of time-of-use prices), evaluating their fitness based on the objective functions, and iteratively applying selection, crossover, and mutation operators to evolve better solutions. The algorithm’s fast non-dominated sorting and crowding distance mechanisms ensure diversity in the solution set, allowing decision-makers to choose the most appropriate pricing strategy for electric cars. The optimization aims to find prices that strike a balance between minimizing carbon emissions and grid peak-valley differences, tailored to the dynamics of the China EV ecosystem.
Case Study: Application in a Chinese Regional Grid
To validate the proposed time-of-use pricing model, I conduct a case study using data from a regional grid in China, focusing on a fleet of 1000 electric cars. The base grid load profile for a typical day shows peaks during daytime and evening hours, with the lowest loads occurring early in the morning. The uncontrolled charging load from electric cars, simulated via Monte Carlo methods, adds significant demand during peak periods, increasing the grid’s peak-valley difference and indirect carbon emissions. The time periods are classified into peak, flat, and valley hours using K-means clustering: peak hours are 12:00-15:00 and 19:00-22:00, valley hours are 06:00-11:00, and flat hours cover the remaining times.
The demand price elasticity matrix used in this case study is derived from historical data on electric car charging behavior:
$$ E = \begin{bmatrix}
-0.623 & 0.3241 & 0.2305 \\
0.3553 & -0.6166 & 0.2216 \\
0.3215 & 0.3038 & -0.6065
\end{bmatrix} $$
This matrix reflects how electric car owners in China respond to price changes across different time periods. For instance, a price increase during peak hours may reduce demand in that period but increase it in flat or valley hours due to cross-elasticity effects.
Using NSGA-II with a population size of 200 and 100 generations, I optimize the time-of-use prices for electric cars. The algorithm parameters include a mutation rate of 0.33 to maintain genetic diversity. The Pareto-optimal solutions are evaluated based on the two objectives, and the best compromise solution is selected using a satisfaction function. For each Pareto solution, the satisfaction \( h_j \) for objective \( j \) is calculated as:
$$ h_j = \begin{cases}
1, & f_j \leq f_{j,\min} \\
\frac{f_{j,\max} – f_j}{f_{j,\max} – f_{j,\min}}, & f_{j,\min} < f_j < f_{j,\max} \\
0, & f_j \geq f_{j,\max}
\end{cases} $$
The overall satisfaction \( h \) is the sum of satisfactions across all objectives. The solution with the highest \( h \) is chosen as the optimal time-of-use pricing scheme for electric cars.
| Time Period | Optimized Price (CNY/kWh) | Change from Base Price |
|---|---|---|
| Peak (12:00-15:00, 19:00-22:00) | 1.33 | +33% |
| Flat (Other hours) | 1.05 | +5% |
| Valley (06:00-11:00) | 0.71 | -29% |
The optimized prices encourage electric car owners to shift charging to valley and flat periods, reducing loads during peak hours. As a result, the total grid load profile becomes flatter, with a lower peak-valley difference. Specifically, the peak load decreases from 3599.03 kW to 3525.92 kW, and the valley load increases from 1434.10 kW to 1539.92 kW, leading to a 3.83% reduction in the peak-valley difference rate. Moreover, the indirect carbon emissions from electric car charging drop significantly. The maximum hourly emissions decrease from 431.76 kg to 413.83 kg, a 4.15% reduction, and the total daily emissions fall from 3706.82 kg to 3650.74 kg, saving over 50 kg of CO2 per day. These results underscore the effectiveness of time-of-use pricing in promoting sustainable charging practices for electric cars in China.
Conclusion and Implications for the China EV Market
In conclusion, this study demonstrates that time-of-use electricity pricing can play a pivotal role in optimizing the charging behavior of electric cars to reduce indirect carbon emissions and enhance grid stability. By integrating dynamic carbon emission factors and demand elasticity into a multi-objective optimization model, I derive pricing strategies that align with the goals of the China EV initiative. The case study results show that strategic price adjustments can lead to significant environmental and operational benefits, including lower carbon emissions and a more balanced grid load. For the China EV market, this approach offers a practical pathway to maximize the positive impact of electric cars on carbon reduction efforts.
However, there are limitations to this research. The study focuses on a specific regional grid and a single typical day, which may not capture seasonal variations or differences between weekdays and weekends. Future work should expand the analysis to include more diverse scenarios and larger datasets, incorporating real-time data from smart grids and electric car telematics. Additionally, the model could be enhanced by considering vehicle-to-grid (V2G) technologies, where electric cars can discharge energy back to the grid, further stabilizing demand and supporting renewable integration. As the China EV market continues to grow, such advanced pricing strategies will be essential for achieving a sustainable and resilient energy ecosystem.
Overall, the findings highlight the importance of collaborative efforts among policymakers, utilities, and consumers in fostering a low-carbon transportation future. By adopting intelligent pricing mechanisms, we can ensure that the rise of electric cars contributes positively to China’s carbon neutrality ambitions, setting a global example for sustainable mobility.