Dynamic Menu-Based Pricing for Electric Vehicle Charging with Vehicle-to-Grid Integration

This paper proposes a dynamic menu-based pricing mechanism for EV charging stations that incentivizes vehicle-to-grid discharging by offering trade-offs between energy prices and discharge levels, resulting in significant profit increases for operators and cost reductions for EV owners compared to existing tariff models.

The Gist

Imagine a busy parking lot as a giant, high-tech battery farm. Instead of just holding cars, this lot is filled with electric vehicles (EVs) that can do two things: eat electricity (charge up) or spit electricity back out (discharge) to help the neighborhood power grid.

The problem is that most parking lots today treat EVs like dumb appliances. They just charge them up whenever the owner plugs them in, often during expensive "rush hour" times, which strains the power grid. Worse, they don't really encourage drivers to give energy back to the grid when it's needed most.

This paper proposes a clever new system called Dynamic Menu-Based Pricing. Think of it as a smart restaurant menu for your car's battery.

The Problem: The "One-Size-Fits-All" Trap

Currently, most charging stations use a "Time-of-Use" tariff. It's like a restaurant that says, "Breakfast is $5, Lunch is $10, Dinner is $15." It's simple, but it doesn't account for your specific hunger or how much energy the grid actually needs right this second. It also rarely pays you to give food back to the kitchen (discharging).

The Solution: The "Build-Your-Own-Burger" Menu

The authors suggest that every time a car pulls into the parking lot, the operator (the restaurant manager) generates a personalized menu just for that car.

Here is how the interaction works, using a simple analogy:

1. The Arrival: You pull up to the charger. You tell the system: "I need my battery at 80% when I leave in 3 hours."
2. The Menu: The system looks at the current electricity prices (which change every 30 minutes, like stock prices) and the grid's needs. It then presents you with a menu of options, like:
   • Option A: "We'll charge you $10. You just charge up. No discharging."
   • Option B: "We'll charge you $8. You charge up, but you let us take 5 kWh out of your battery during the evening rush hour."
   • Option C: "We'll charge you $6. You charge up, but you let us take 10 kWh out during the peak rush."
3. The Choice: You (the driver) pick the option that saves you the most money or fits your needs best.
4. The Magic: The system calculates these prices instantly. If you choose Option C, the operator knows exactly how much to pay the power grid and how much to charge you to make a profit, while ensuring the grid doesn't blow a fuse.

Why This is a Game-Changer

The paper runs a massive simulation (like a video game) using real data from Australia's power grid. Here is what they found, translated into plain English:

• The Operator Wins Big: The parking lot owner makes 30% more profit compared to doing nothing, and 22–29% more than using standard pricing. They are essentially acting as a smart middleman, buying low and selling high at the perfect moments.
• The Driver Wins Too: Even though the operator makes more money, the drivers actually pay 9–18% less for their charging. Why? Because the "menu" finds the most efficient way to use the grid, passing those savings to you.
• The Grid Gets a Hero: The system encourages cars to give energy back to the grid (V2G) 87% to 235% more than other methods. It's like having thousands of tiny power plants (your cars) ready to help when the sun goes down or the wind stops blowing.

The "Secret Sauce": How They Did It

The math behind this is complex (involving "bilevel optimization," which sounds scary but is just a fancy way of saying "The Manager sets the rules, and the Customer picks the best deal, and we solve for the perfect balance").

The authors turned this complex math into a super-fast computer program.

• Speed: When a new car arrives, the system calculates the perfect menu and prices in less than one second.
• Scalability: It can handle 100 cars in under 40 seconds, or even 250 cars in under 3 minutes. This means it can run in real-time on a normal laptop without needing a supercomputer.

The Bottom Line

This paper proves that we don't need to force people to do things for the grid. Instead, we can create a smart marketplace where:

1. The parking lot operator acts like a savvy trader.
2. The driver acts like a smart shopper.
3. The grid gets the energy it needs exactly when it needs it.

It turns every parked electric car from a "drain on the system" into a flexible, money-making asset for both the owner and the operator, all while keeping the lights on for everyone else. It's a win-win-win scenario.

Technical Summary

1. Problem Statement

The rapid adoption of Electric Vehicles (EVs) creates significant stress on power distribution networks, particularly during peak charging hours. While Vehicle-to-Grid (V2G) technology offers a solution by allowing EVs to discharge energy back to the grid (balancing supply and demand), its practical deployment is hindered by:

• Lack of Incentives: Current pricing schemes (e.g., Time-of-Use tariffs) offer little financial motivation for users to discharge.
• Rigid Infrastructure: Existing mechanisms often assume static tariffs, one-way energy flow, or ignore real-time grid constraints (feeder capacity).
• User Uncertainty: Users are concerned about battery degradation and lack transparency in pricing.

The core problem addressed is how a parking lot operator can design a pricing mechanism that dynamically coordinates EV charging and discharging in real-time. The goal is to maximize operator profit and grid support while minimizing costs for EV owners, all while respecting technical constraints (battery limits, feeder capacity) and user preferences.

2. Methodology

The authors propose a Dynamic Menu-Based Pricing Mechanism formulated as a Bilevel Optimization Problem, which is subsequently reformulated into a single-level Mixed-Integer Linear Program (MILP) for computational efficiency.

A. System Model

• Setting: A parking lot with bidirectional chargers connected to a distribution grid with a specific feeder capacity limit ($Y_{max}$).
• Process: EVs arrive sequentially. Upon arrival, an EV declares its energy requirements (initial/final State of Charge) and parking duration.
• The Menu: The operator generates a personalized menu of options ($D$) for the arriving EV. Each option $d$ specifies a total amount of energy the EV is allowed to discharge back to the grid during its stay.
• Real-Time Context: The operator uses real-time wholesale electricity prices (buy/sell) and forecasts for the remainder of the day to calculate costs.

B. Optimization Framework

The interaction is modeled as a Stackelberg game:

1. Upper Level (Operator/Leader): Maximizes profit by determining the optimal price ($\pi_{n,d}$) for each menu option. The price is based on the marginal cost of admitting the new EV under that specific discharge option, plus a non-negative markup. The operator optimizes the charging/discharging schedules for all active EVs to minimize grid settlement costs while respecting feeder limits.
2. Lower Level (EV/Follower): The EV selects the option from the menu that maximizes its utility.
   • Utility Function: $U = (\text{Value of Energy Gained}) - (\text{Price Paid}) - (\text{Battery Degradation Cost})$.
   • The EV only accepts an option if the utility is non-negative.

C. Mathematical Reformulation

To solve the bilevel problem efficiently in real-time:

• The lower-level problem (EV choice) is relaxed into a linear program.
• Karush-Kuhn-Tucker (KKT) conditions of the lower level are derived.
• These conditions are transformed into indicator constraints to convert the bilevel problem into a single-level MILP.
• Rolling Horizon: The optimization is re-run every time a new EV arrives, updating schedules for the current set of active vehicles without altering commitments made to previously accepted vehicles.

3. Key Contributions

1. Dynamic Rolling-Horizon Pricing: A novel approach that updates the price menu and schedules in real-time upon every EV arrival, tailoring incentives to current system conditions and individual user needs.
2. Integrated V2G and Network Constraints: Unlike prior work, this model simultaneously handles:
   • Real-time wholesale price fluctuations (buying and selling).
   • Bidirectional power flow (V2G).
   • Feeder capacity limits (grid constraints).
   • Per-EV personalized menus.
3. Computational Efficiency: By reformulating the bilevel problem into a single MILP using indicator constraints, the system achieves real-time performance.
   • Performance: Solves a full-day instance with 100 EVs in <40 seconds and 250 EVs in <200 seconds. Individual EV pricing occurs in <1 second.
4. Comprehensive Validation: Extensive testing using 12 months of Australian Energy Market Operator (AEMO) data, comparing against three tuned baseline schemes and a charge-only scenario.

4. Results

The proposed mechanism was compared against three baselines:

1. Adjusted Real-time Pricing: Dynamic charging prices but fixed/discounted discharging.
2. Fixed Flat Tariff: Static prices all day.
3. Hybrid Pricing: Dynamic charging, fixed discharging.

Key Performance Metrics (Average Improvements):

• Operator Profit: Increased by 27.01% compared to the best baseline.
• EV Payments: Reduced by 14.69% for EV owners.
• V2G Contribution: Energy exported to the grid increased by 161.5%.
• Specific Comparisons:
  • vs. Fixed Flat Tariff: Profit +22.91%, EV Payments -18.47%.
  • vs. Adjusted Real-time: Profit +29.61%, EV Payments -13.7%.
  • vs. Charge-only (No V2G): Profit +29.61%, EV Payments -17.01%.

Sensitivity Analysis:

• Menu Granularity: Benefits plateau after 6–8 discharge options; a small set of options is sufficient.
• Grid Capacity: The mechanism remains effective even under tight feeder constraints.
• Price Uncertainty: The system is robust to wholesale price forecast errors (up to 50% noise), with profit deviations remaining low (median profit stability).

5. Significance

This paper provides a practical, computationally efficient, and economically viable framework for real-time EV charging management with V2G integration.

• Economic Viability: It proves that V2G can be profitable for operators without penalizing users, addressing the "chicken-and-egg" problem of V2G adoption.
• Grid Stability: By incentivizing discharging during peak demand and charging during off-peak times, the mechanism actively supports grid stability and renewable integration.
• Scalability: The sub-second computation time per vehicle demonstrates that this approach is ready for deployment in large-scale urban parking facilities without requiring specialized hardware, relying only on standard optimization solvers and existing market data.

In summary, the study bridges the gap between theoretical bilevel optimization and real-world application, offering a robust solution to coordinate the complex interactions between EV owners, parking operators, and the power grid.