⚡ electrical engineering

Techno-Economic Analysis of Shared Mobile Storage for Demand Charge Reduction

This paper proposes a high-fidelity mixed-integer linear programming framework with a marginal-value-based heuristic to demonstrate that shared electric vehicle fleets can effectively reduce demand charges and achieve techno-economic viability under realistic logistical constraints, as validated by real-world data from San Francisco.

Original authors: B Hari Kiran Reddy, Ge Chen, Junjie Qin

Published 2026-06-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: B Hari Kiran Reddy, Ge Chen, Junjie Qin

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you own a large factory or office building. Your electricity bill has two parts: one for the total amount of power you use, and a second, often massive fee called a "demand charge." This fee is based on your single highest spike in power usage during a short window (like 15 minutes). It's like a restaurant charging you not just for the food you ate, but for the single moment you asked for the most expensive dish, regardless of how much you ate the rest of the month.

Traditionally, to fix this, businesses buy giant, stationary batteries to sit on their property and "soak up" these spikes. But these batteries are expensive to buy, take up space, and can only help the building they are attached to.

This paper proposes a smarter, more flexible idea: The "Rideshare" for Electricity.

The Core Concept: Mobile Power Banks

Instead of every building buying its own giant battery, imagine a fleet of electric vehicles (EVs) that act as mobile power banks. These cars have special chargers that let them not only take electricity from the grid but also give electricity back to a building.

Here is how the "Rideshare" works:

The Driver: A fleet operator owns several EVs and employs drivers.
The Mission: When a building is about to hit its "spike" in power usage, a driver is dispatched to that location.
The Swap: The EV arrives, plugs in, and discharges its battery into the building just long enough to flatten that spike.
The Move: Once the spike is over, the EV drives to the next building that needs help, or returns to a central hub to recharge.

The Hidden Costs (The "Traffic" and "Wear" Problem)

The authors realized that previous studies were too optimistic. They treated EVs like magic wands that could teleport. In reality, moving a car costs money and energy. The paper introduces a "High-Fidelity" (very realistic) model that accounts for three messy realities:

The Commute: The car has to drive to the job. This uses battery power (transit energy) and costs money.
The Driver: Someone has to drive the car. The paper calculates the actual wages of the driver, which turns out to be the biggest cost.
The Wear and Tear: Every time the battery is used, it gets a little older. The paper calculates the cost of this "battery depreciation."

The Solution: A Smart Dispatcher

To make this work, the researchers created a complex math model (a "Mixed-Integer Linear Program") to figure out the perfect schedule. It's like a super-intelligent GPS that decides:

Which car should go to which building?
When should it leave the charging station?
When should it arrive to help?
Should it charge now or later?

Because solving this perfectly for a whole city is incredibly slow and difficult for computers, they also invented a "Marginal-Value Heuristic."

The Analogy: Think of this like a smart delivery driver who doesn't try to plan the perfect route for the whole week in advance. Instead, at every moment, they ask: "Which customer is paying the most right now? Which of my cars has done the least work today? Let's send that car to that customer."
This method is fast and gets results that are almost as good as the perfect solution, but it runs in minutes instead of hours.

What They Found (The Results)

Using real data from San Francisco, the team ran simulations to see if this business model actually makes money.

It Works, But Labor is Key: The system can save businesses a lot of money on their demand charges. However, the driver's salary is the biggest hurdle. If drivers are too expensive, the business loses money. The study suggests that for this to be profitable, the savings from the electricity bill must be high enough to cover the driver's wages.
The "Tiered" Strategy is Best: Not all buildings are the same. Some are small (Schedule B-10), and some are huge (Schedule B-19).
- The researchers found that the best setup is a Tiered approach: Give the small buildings standard chargers and the huge buildings super-fast chargers. This mix saves more money than giving everyone the same type of charger.
Seasonal Differences: The business is much more profitable in the summer than in the winter. This is because electricity demand spikes are higher and more expensive during hot summer afternoons.
- Winter: A fleet of 3 cars could save about $30,000 a month.
- Summer: A fleet of 6 cars could save nearly $100,000 a month.
Diminishing Returns: Adding more cars helps, but only up to a point. The first few cars fix the biggest, most expensive spikes. The tenth car only fixes tiny, cheap spikes, so it might not be worth the extra driver's salary.

The Bottom Line

This paper proves that using shared electric vehicles as mobile batteries is a technically and economically viable idea, but it's not a magic bullet. It requires careful planning to ensure the money saved on electricity bills is greater than the cost of the drivers, the car wear-and-tear, and the fuel used to drive the cars to the job sites.

The authors conclude that while stationary batteries are rigid and expensive, a "fleet of mobile batteries" can be a highly profitable tool for businesses, provided the operational costs (especially labor) are managed correctly.

Technical Summary: Techno-Economic Analysis of Shared Mobile Storage for Demand Charge Reduction

Problem Statement
Commercial and industrial (C&I) users often face significant electricity costs driven by demand charges, which are based on the maximum real power demand averaged over short intervals. While stationary battery energy storage systems (BESS) are a conventional solution, they suffer from high upfront costs and spatial rigidity. Shared electric vehicle (EV) fleets offer a "mobile storage" alternative that can serve multiple users across different locations. However, existing literature often relies on idealized models that neglect critical logistical and operational constraints. Specifically, prior studies frequently overlook the spatio-temporal coupling of energy consumption (transit overheads), the labor costs of EV drivers, and the complex trade-offs involving battery degradation and charging infrastructure capabilities. The core problem addressed is determining the true net economic value of shared EV fleets for demand charge reduction once these granular, practical factors are rigorously accounted for.

Methodology
The authors propose a high-fidelity fleet management framework that integrates operational and economic factors often omitted in previous mobile storage studies.

Mathematical Formulation: The dispatch problem is formulated as a Mixed-Integer Linear Program (MILP). The objective is to jointly minimize the total demand charges for C&I users and the operational costs of the EV fleet. The model explicitly captures:
- Spatio-temporal coupling: Energy consumption during transit between the central charging station and user locations.
- Labor costs: Modeled as a fixed cost per driver based on working hours.
- Battery degradation: Represented through usage-based depreciation (cost per kWh cycled) and age-based depreciation.
- Charging infrastructure: Costs and capacities of both AC Level-2 and DC fast chargers at user sites and the central station.
- Constraints: The formulation includes constraints for battery state-of-charge (SOC), charging/discharging efficiencies, transit times (assumed to take one interval), and the requirement that an EV must be in transit for one interval before it can begin discharging at a user site.
Algorithm Development: Due to the computational complexity of the MILP, which grows exponentially with the number of users and EVs, the authors develop a marginal-value-based heuristic algorithm.
- The algorithm iteratively identifies the "most-valued" user (highest marginal demand charge reduction per service) and the "least-used" available EV.
- It dispatches the selected EV to the user if the marginal demand charge savings exceed the incremental operational costs (charging and depreciation).
- This approach achieves near-optimal performance with significantly reduced computation time compared to solving the full MILP, making it scalable for daily fleet operations.
Case Study Setup: The framework is validated using real-world smart meter data from 730 business buildings in San Francisco, California, served by PG&E. The study considers two user tariff schedules (B-10 and B-19) and evaluates three infrastructure configurations: All-AC, All-DC, and a Tiered setup (mixing AC for smaller users and DC for larger users).

Key Contributions
The paper makes three primary contributions:

Unified MILP Formulation: It presents a model that explicitly integrates transit energy, labor costs, and battery degradation into a single dispatch objective, ensuring operational feasibility and accurate economic assessment.
Efficient Heuristic: It proposes a marginal-value-based heuristic algorithm that solves the complex dispatch problem in polynomial time, achieving near-optimal results while handling the computational burden of large fleets.
Comprehensive Techno-Economic Assessment: Through a case study, it identifies labor costs as the dominant factor affecting revenue and validates a Tiered charger deployment strategy. This strategy, which mixes DC fast chargers for large users and AC chargers for smaller users, is shown to yield superior net savings compared to uniform infrastructure configurations.

Results
Using data from San Francisco, the analysis reveals several key findings:

Economic Viability: A modest number of EVs can achieve significant demand charge savings sufficient to recover ownership and operational expenses.
Seasonal Variation: Net savings are substantially higher in summer (averaging ~~$34,111 for a single EV) compared to winter (~~$18,025), driven by higher demand charge rates and peak loads during summer months.
Infrastructure Impact: The Tiered infrastructure setup consistently yields the highest net savings. While All-DC setups offer higher savings in summer, the high amortized cost of DC chargers makes them less viable for smaller users (Schedule B-10). The Tiered approach balances these factors effectively.
Fleet Sizing: There is a point of diminishing returns. For the Tiered setup, the optimal fleet size is found to be three EVs in winter and six EVs in summer. Beyond these sizes, the marginal demand charge reduction per additional EV decreases while labor costs continue to rise linearly, eventually eroding net savings.
Cost Drivers: Labor costs are identified as the dominant factor affecting profitability. The study calculates a "critical labor cost" (approx. $120–$ 210/hr depending on season) beyond which the business model becomes unprofitable. Current estimated labor costs ($19/hr) are well below this threshold.
Uncertainty Sensitivity: The model is sensitive to load forecast uncertainty. Net savings decline monotonically as forecast error increases. However, the business model remains viable under moderate uncertainty levels (up to 10–15% error), particularly if autonomous operation (eliminating labor costs) is assumed.

Significance and Claims
The paper claims that its primary significance lies in bridging the gap between theoretical mobile storage potential and practical business viability. By explicitly modeling "real-world" frictions—specifically transit energy, driver labor, and battery wear—the authors demonstrate that shared mobile storage is not just theoretically possible but economically attractive under specific conditions.

The authors modestly note that their heuristic is a "myopic" approach without a universal theoretical approximation guarantee, yet it demonstrates near-optimal performance empirically. They emphasize that the optimal fleet sizes and savings figures are case-specific, dependent on local tariff structures, load characteristics, and cost assumptions. The work provides a realistic baseline for evaluating the commercial viability of shared mobile energy storage, moving beyond idealized assumptions to offer a framework that accounts for the complex interplay of logistics, energy, and economics. Future work is suggested to incorporate stochastic modeling for travel time variations, EV availability, and charger reliability.