From Packet to Power Switching: Digital Direct Load Scheduling

Mahnoosh Alizadeh, Anna Scaglione, Robert J. ThomasIEEE JSAC 2012 · DOI

TL;DR

This paper introduces Digital Direct Load Scheduling (DDLS), a novel approach to manage electricity demand by digitizing and scheduling individual appliance requests. DDLS aims to improve grid efficiency, facilitate renewable energy integration, and offer customers financial incentives by allowing them to voluntarily cede control of their appliance usage times to a neighborhood aggregator.

What Problem Does This Paper Address?

The current power grid has excellent control over electricity generation but very limited control over demand. This imbalance makes it difficult to integrate volatile renewable energy sources, as they require a stable and predictable load. Consumers are largely shielded from price signals, leading to inefficient energy consumption patterns that often favor fossil fuel generation. The paper addresses the need for a mechanism to make electricity demand more flexible and controllable, enabling better integration of renewables and improving overall grid efficiency and market operations.

Methods

The core of the proposed solution is the Digital Direct Load Scheduling (DDLS) model. This system works by:

  • Digitizing Load Requests: Individual appliance requests for energy are digitized and represented by a "charging code" ($C_i$), which captures parameters like charging rate and required capacity.
  • Queue-based Scheduling: These digitized requests are then quantized into $Q$ classes of service, forming queues. Each queue is managed by a neighborhood controller or aggregator.
  • Optimized Dispatch: The aggregator optimizes the start times of these "deferrable loads" (appliances like EVs, dishwashers, etc.) to shape the aggregate neighborhood load profile. This optimization considers the spot price of energy and aims to align demand with available generation, particularly from renewables.
  • Cellular Architecture: The system is envisioned with a hierarchical architecture: Home Energy Management Systems (HEMS) at the residence level collect and anonymize requests, sending them to a Community Energy Management System (CEMS) for scheduling. A Green Energy Management System (GEMS) interfaces with the electricity market.
  • Delay Cost Modeling: The inconvenience to customers is modeled as a "Delay Cost Increment" (DCI), which is minimized by the scheduler. This allows for a trade-off between customer reward and grid efficiency.
  • Model Predictive Control (MPC): For real-time scheduling, an MPC approach is used, considering a look-ahead horizon to make foresighted decisions.

Datasets

Not discussed in the paper. The paper focuses on modeling and simulation rather than empirical data from real-world deployments.

Key Insights and Ideas

  • Digitizing demand is key: Treating individual appliance energy requests as "packets" that can be scheduled digitally unlocks new levels of demand-side management.
  • Voluntary participation is crucial: The DDLS model is designed as a voluntary program, incentivizing customers through financial rewards for their flexibility.
  • Neighborhood aggregation offers benefits: Aggregating demand requests at a neighborhood level provides economies of scale and allows for more effective load shaping than individual appliance control alone.
  • Delay cost as a proxy for inconvenience: Modeling customer inconvenience as a delay cost provides a quantifiable metric for optimization, balancing grid needs with user satisfaction.
  • Decentralized control is scalable: A cellular architecture with local schedulers (HEMS) and a central aggregator (CEMS) offers a scalable approach to managing a large number of deferrable loads.
  • Price signals alone are insufficient: The paper highlights that relying solely on price signals for individual customer optimization can lead to undesirable rebound peaks and grid instability, underscoring the need for coordinated scheduling.

Key Results

The paper presents simulation results demonstrating the effectiveness of DDLS:

  • Load Profile Shaping: DDLS allows the aggregate load to closely follow a desired generation profile (ZIC power profile), significantly improving upon uncontrolled load behavior.
  • Cost Reduction: A relative reduction of 41% in operational costs was observed in simulations compared to an uncontrolled load scenario when customers participate in DDLS.
  • Distributed Control Benefits: Using distributed schedulers (20 schedulers) resulted in a 21% saving compared to a single centralized scheduler, indicating that a more localized control can be more efficient.
  • Minimal Customer Delay: The average delay experienced by customers participating in DDLS was shown to be minimal, suggesting that the program can be implemented with little inconvenience.

Limitations and Future Work

  • Reactive Power Ignored: The current model primarily focuses on active power requests and assumes reactive power requirements can be easily incorporated as a constraint.
  • Deterministic Arrivals in Scenario I: Scenario I assumes deterministic arrival processes, which is a simplification.
  • Stochastic Arrivals in Scenario II: Scenario II addresses stochastic arrivals but relies on approximations and state aggregation for tractability.
  • Future Work: The authors suggest several avenues for future research:
  • Developing architectures to integrate these "reservoirs of storable demand" into the electricity market.
  • Further studying the statistical information required for scheduling, such as arrival patterns and renewable availability.
  • Integrating the flexibility of thermostatically controlled appliances into neighborhood DSM programs.
  • Detailed analysis of the market interface for aggregators.

Summary updated

BibTeX
@article{alizadeh2012from,
  author = {Mahnoosh Alizadeh and Anna Scaglione and Robert J. Thomas},
  title = {{From Packet to Power Switching: Digital Direct Load Scheduling}},
  journal = {IEEE JSAC},
  year = {2012},
  doi = {10.1109/jsac.2012.120702},
}
Figure