PaperlayerView exact revision on arXiv

Source-linked AI summary

Water as a Battery: Using Hydropower Flexibility to Support the Grid

Chinmay Morankar, Abdullah Al Mehadi

arXiv:2610.34089v1eess.SYcs.SY

TL;DR

Growing data-center demand makes reliably serving round-the-clock loads with variable renewables a challenge. The paper optimizes renewable and battery capacities alongside hydropower, finding that The Dalles case required no battery storage while Conowingo considered about 130 MW of PV for partial grid support.

Abstract

from arXiv · show

This study presents a two-layer optimization framework to determine the optimal sizing of PV, wind, and battery storage co-located with existing hydroelectric plants to support rapidly increasing data center loads while minimizing cost, grid reliance, and transmission upgrades. The outer layer uses Particle Swarm Optimization (PSO) to select capacity combinations, while the inner layer performs rolling-horizon dispatch simulations to model hourly energy balance among PV, wind, hydro, and BESS, considering operational constraints. Two case studies were analyzed. At The Dalles Dam (Oregon), favorable wind resources and land availability enabled deployment of 111.2 MW PV and 117.9 MW wind, with no BESS required. Hydro flexibility allowed the system to meet the load with minimal grid support. In contrast, Conowingo Dam (Maryland) faced land constraints and limited renewables to 130 MW PV, leading to higher grid dependence. Co-locating hydro with renewables significantly enhances reliability and cost effectiveness by leveraging hydro's dispatchability to balance daily and seasonal variable generation from PV + Wind, reducing the need for storage or costly grid imports. This analysis provides a framework to meet the 99.9% availability targets of this new load type on the grid with sustainable energy resources.

C. MORANKAR, A. AL MEHADI* WSP USA, Inc. USA … INTRODUCTION

The introduction motivates hydropower-backed renewables for data centers’ continuous loads and presents a two-layer framework to size and dispatch co-located resources. It frames hydropower flexibility as a way to balance variable generation and reduce reliance on storage or grid imports.

KEYWORDS

  • The stated scope spans data-center availability and hybrid energy systems combining hydroelectricity, PV, wind, and BESS.These are the section’s listed keywords.Source: KEYWORDS · p. 1, INTRODUCTION · p. 2

INTRODUCTION

  • Variable wind and solar could provide 54–71% of global electricity by 2050, but their intermittency creates supply-demand imbalances without storage or flexible generation.The introduction identifies fast-balancing capacity and the ability to store excess energy as reliability needs.Source: INTRODUCTION · pp. 1–2
  • U.S. data centers consumed roughly 176 TWh in 2023 (≈4.4% of electricity use), and annual consumption could reach 1,050 TWh (12%) by 2030.About 56% of U.S. data-center electricity still comes from fossil fuels, while major operators are exploring reliable 100% carbon-free power.Source: INTRODUCTION · p. 2
  • Co-located hydro can balance variable PV and wind by storing excess renewable energy as reservoir water and releasing it when generation wanes or demand spikes.This can reduce the need for BESS or fossil backup and raise existing hydro-substation capacity factor from 40–50% to 80%–90%.Source: INTRODUCTION · p. 2
  • The two-layer framework uses Particle Swarm Optimization to select PV, wind, and BESS capacities, then simulates hourly rolling-horizon dispatch subject to hydro operating constraints.It aims to meet a data center load at minimum cost and with minimal grid draw.Source: INTRODUCTION · p. 2
  • The framework is applied to The Dalles Dam in Oregon, with favorable wind and available land, and Conowingo Dam in Maryland, where land and resource limits constrain on-site renewables.The passage introduces these as contrasting case studies and does not report their results here.Source: INTRODUCTION · p. 2

METHODOLOGY

The study uses a two-layer optimization framework to size PV, wind, and battery storage alongside an existing hydroelectric plant. Particle Swarm Optimization selects capacities, while rolling-horizon dispatch evaluates short-term operation and total cost.

Outer Layer: Particle Swarm Optimization

  • The two-layer framework combines Particle Swarm Optimization with rolling-horizon dispatch for short-term planning of PV, wind, and battery storage capacities.It is designed for a low-carbon power system integrated with an existing hydroelectric plant.Source: METHODOLOGY · p. 4
  • The PSO layer searches combinations of PV capacity, wind capacity, and BESS energy capacity.Each particle represents these three capacity variables.Source: METHODOLOGY · p. 4
  • The optimization objective is to minimize total cost as evaluated by the dispatch simulation.Source: METHODOLOGY · p. 4

Inner Layer: Rolling-Horizon Dispatch

The inner layer simulates hourly contributions from PV, wind, hydro, and battery storage to meet load. Its dispatch rules prioritize renewables, use hydro and battery discharge when needed, and rely on grid power only when transformer capacity is exceeded.

Inner Layer: Rolling-Horizon Dispatch

  • At each hourly time step, the simulator computes PV, wind, hydro, and BESS contributions toward meeting the load profile.The hourly-resolution dispatch simulator evaluates these resources against the load profile.Source: Inner Layer: Rolling-Horizon Dispatch · p. 4
  • When PV and wind meet or exceed load, renewables serve demand, hydro is deferred, and surplus energy charges the battery.Otherwise, hydro—including stored water—is dispatched, with battery discharge if needed.Source: Inner Layer: Rolling-Horizon Dispatch · p. 4
  • Grid power is used only when maximum transformer capacity is exceeded, while the simulator updates battery SOC and hydro storage and enforces ramp-rate and SOC constraints.The dispatch flowchart identifies Csaved(t) as saved hydro power greater than zero.Source: Inner Layer: Rolling-Horizon Dispatch · p. 4

Cost Model

The cost model minimizes total energy cost, accounting for PV, wind, and battery costs alongside grid-import, degradation, ramping, and unmet-demand penalties.

Cost Model

  • EC is the objective function, calculated as EPV · CostPV + EBESS · CostBESS + EWind · CostWind.EPV and EBESS are total PV and BESS output energy over the simulation horizon; PV and wind unit costs are in $/MW, while BESS unit costs are in $/MWh and include capital, replacement, and O&M costs.Source: Cost Model · p. 4
  • Grid-import costs vary over time, battery degradation costs scale with throughput, and CostRamp penalizes rapid changes in BESS output.Unmet demand incurs a large penalty (e.g., 107) to discourage infeasible solutions.Source: Cost Model · p. 5

PV and Wind Generation

PV and wind generation profiles were derived from SAM simulations using location-specific TMY weather data, then scaled by installed capacities during optimization.

PV and Wind Generation

  • SAM simulated renewable output from site-specific TMY weather and system parameters, with hourly PV and wind profiles extracted for a representative day.The simulations use physical system configurations and atmospheric inputs, including irradiance, temperature, and wind speed.Source: PV and Wind Generation · p. 5
  • During simulation, normalized SAM profiles were scaled using installed PV and wind capacities, with PV inverter and wind turbine electrical efficiencies represented.The profiles, sPV(t) and sWind(t), describe per-unit output; GPV(t) and GWind(t) are net AC power outputs in MW.Source: PV and Wind Generation · p. 5

SIMULATION ENVIRONMENT · PV System

The study modeled hourly PV generation at The Dalles and Conowingo Dams using SAM, with site-specific TMY weather data, as input to the optimization framework.

PV System

  • A 1 MW PV system with a 1.2 DC-to-AC ratio was modeled at The Dalles and Conowingo Dam using NREL’s SAM.The configuration was selected to align with typical utility-scale deployment standards.Source: PV System · p. 6
  • Site-specific NSRDB Typical Meteorological Year weather files provided realistic irradiance and temperature conditions for the PV simulations.The modeled system captured direct and albedo irradiance effects, relevant to bifacial technologies.Source: PV System · p. 6
  • SAM’s resulting hourly AC power output served as the base PV generation profile for the optimization framework.Source: PV System · p. 6

Wind System

The wind-system input was derived from a simulated single-turbine profile for The Dalles Dam and normalized for the optimization framework.

Wind System

  • A 100-meter-hub-height turbine used SAM wind-resource data for The Dalles Dam to generate the hourly AC wind profile, which was normalized by its peak output.The study kept turbine parameters simple; the resulting profile served as the wind input to the optimization framework.Source: Wind System · p. 6

Component Cost Assumptions

The outer-layer PSO sizes PV, wind, and battery storage by minimizing a total-cost objective that includes component costs, degradation, operations, grid costs, and penalties. The assumptions also set a 77 MW peak demand and 1,848 MWh daily energy requirement at each site.

Component Cost Assumptions

  • PV and wind capital costs are modeled as daily costs per installed MW, while BESS capital cost is based on installed energy capacity in MWh.The plant operating period is 20 years (7,300 days); BESS economic life is 15 years (5,475 days).Source: Component Cost Assumptions · p. 6
  • Battery degradation cost is throughput-based, using the degradation rate and energy charged or discharged over each time step.The typical time step is 1 hour.Source: Component Cost Assumptions · pp. 6–7
  • The total investment objective sums PV, wind, and BESS capital costs with battery degradation, O&M, grid, and penalty costs.This total cost is the outer-layer PSO objective for selecting PV, wind, and BESS sizing.Source: Component Cost Assumptions · p. 7
  • 55 MW of data-center IT load at each site, adjusted for 1.4 PUE, yields a 77 MW peak demand and 1,848 MWh daily energy requirement.Source: Component Cost Assumptions · p. 7

Case-1 The Dalles hydro power plant at Oregon

The Dalles case optimized PV and wind capacity under an infrastructure limit; hydro flexibility avoided BESS, and renewable generation kept grid reliance minimal.

Case-1 The Dalles hydro power plant at Oregon

Case-2 The Conowingo hydro power plant at Maryland

Geographic constraints limited realistic PV deployment at Conowingo to approximately 130 MW, so renewables provided only partial support for data-center demand and grid imports remained necessary.

Case-2 The Conowingo hydro power plant at Maryland

Loading 2610.34089v1…