Placeholder

TRANSMISSION OPERATIONS & PLANNING

Large Load Integration

We are improving how engineers’ model new and existing assets, including large loads, (e.g. data centers), developing tools to identify and mitigate modeling errors. 
 

Research and Development Challenges

Across the globe the data center sector is growing rapidly as societies become increasingly reliant on digital connectivity. Data Centers (DCs) are the physical backbone delivering the computing capability necessary to deliver Artificial Intelligence (AI) applications, digital services, and cloud-based storage. The scale and pace of growth in digital demand is driving large growth in electricity demand to power this computing activity. For utilities and grid operators serving energy to the data center sector there are important aspects to consider when assessing the integration of such large single point loads, which include the evolving load profiles of the data center sector, technical characteristics of the equipment within a data center and the potential resultant implications for power quality, system stability and the reliability of power system operation.. A major concern is grid resilience and planning. Integrating large consumers requires extensive coordination to ensure sufficient generation, robust transmission, and adequate contingency measures for loss-of-load events or unplanned shutdowns. Utilities often use sophisticated forecasting, load management, and sometimes install dedicated infrastructure or on-site generation to mitigate the grid impacts of large load.

The long-term goal is to:

  • Develop models for emerging loads like hyperscale and AI data centers, crypto mining facilities, hydrogen electrolysis plants, electric vehicle supply equipment (EVSE). Model development will span across both phasor/positive sequence and electromagnetic transient (EMT) simulation platform.
  • Conduct laboratory testing of new emerging load components like EVSEs and UPSs to ensure that developed models are a reliable representation of the actual device.
  • Provide methods to account for rapidly electrifying sectors in resource adequacy and investigates adequacy risk mitigation strategies in load growth scenarios.
  • Continue leading efforts related to load forecasting, spanning short‑ and long‑term horizons with emphasis on spatial detail (down to zones/substations), realistic post‑connection ramping, and empirically derived operating shapes.

Related Projects

Resource Adequacy Modelling

Resource Adequacy (RA) plays a key role identifying if existing and potential grids could technically and safely handle this load growth. RA is a capacity assessment study to evaluate whether a current or projected resource mix is sufficient to meet demand and energy needs for a particular grid. From a RA point of view, the data center load profile and operational characteristics are very important in assessing risk to resource adequacy. The EPRI team has been investigating the needs for flexibility, the capabilities that may enable it and understand the extent to how it can be delivered.

Modeling data centers for transmission planning

The power quality requirements for these facilities are primarily governed by the requirements set forth by the information technology industry council. Around 90% of data center load consists of switched-mode power supplies, which are power electronic converters feeding the dc server load. While these devices have varying characteristics, it is often unnecessary to model each one separately for bulk system studies. This is because their dynamic responses are similar within the frequency range relevant for planning studies. Modeling all complexities may not significantly impact system stability analysis and can introduce unnecessary complications. Therefore, it is essential to identify and model only the relevant characteristics to understand their impact on system stability. The team is working on improving interconnection processes to leverage flexibility from future data centers. Research remains ongoing on modeling of emerging loads, such as EV charging stations, data centers, and hydrogen electrolyzers.

Load Forecasting

While load forecasting is a well-established field, new challenges have arisen with the advent of large point loads, such as data centers. These facilities, which house vast amounts of computing power and data storage, can significantly impact local and regional power grids. The unpredictable and often substantial energy demands of hyperscale data centers pose unique challenges for traditional load forecasting models. As a result, there are ongoing discussions and research efforts aimed at developing more sophisticated methods to accurately predict the impact of these large point loads on the grid. We are currently developing granular load forecasting methods and maps to aid these types of flexible loads integration

Related Deliverables

Modeling of Emerging Power System Loads

3002032997

Explains load model development.

Large-Load PQ Guidance

3002034267

Distinguishes the effects of large electronic loads into steady‑state and transient categories, showing how continuous operation and fast disturbances can trigger voltage swings, generator trips, and other system issues.

Large Loads in RA

3002033176

Provides methods to account for rapidly electrifying sectors in resource adequacy and investigates adequacy risk mitigation strategies in load growth scenarios.

Application of Day-Ahead Forecasting Framework to Large Loads: Data Centers, EV Charging, and Beyond

3002034070

Addresses that short -term forecasting gaps by applying a large-load forecasting framework to diverse datasets

Area Leads