Batteries vs PSH: Technical comparison
Summary
Lithium-ion battery storage (BESS) and pumped storage hydropower (PSH) have genuinely complementary rather than competing technical profiles. BESS excels at millisecond response, short discharge durations (1–4 hours), flexible siting, and high cycle frequency. PSH excels at long discharge durations (6–24+ hours), negligible capacity degradation, century-scale asset life, native synchronous inertia, and near-zero self-discharge. The key area of convergence is grid-forming inverter technology, which is narrowing the gap on inertia and black-start services historically exclusive to synchronous machines like PSH.
This page is part of the research tree for How do batteries and pumped storage hydro compare as electricity storage technologies for renewable-intensive systems like the UK's?.
1. Round-trip efficiency (AC-to-AC)
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Fleet average (EIA 2019) | 82% | 79% |
| Modern systems (central estimate) | 88–92% (LFP) | 80–85% (high-head, modern) |
| Documented range | 84–92% (AC system level) | 70–87% |
| Cell-level only (BESS) | 90–96% (DC, not comparable) | — |
The dominant metric for commercial procurement and grid comparison is AC-to-AC round-trip efficiency (RTE), measured at the Point of Connection per IEC 62933-2-1. The EIA reported U.S. fleet averages of 82% (BESS) and 79% (PSH) for 2019 — the most comparable apples-to-apples figures available, though both technologies have improved since.
BESS losses are distributed across: battery cells and their cooling, the power conversion system (PCS/inverter), transformers, and auxiliary systems. Merus Power notes that "actual losses in commercial operation are about 30–40% of the energy losses forecast by peak RTE" because systems typically operate at partial load.
PSH efficiency varies with head, flow rate, and technology type. Variable-speed (adjustable-speed) machines recover part-load efficiency and approach the upper range. Low-head PSH faces greater relative friction losses; high-head projects (>400 m) typically achieve the best efficiencies.
Contested area: DC cell-level figures (90–96% for BESS) are commonly cited in marketing materials but are not comparable to AC-to-AC system figures. The gap between "best achievable" and fleet average is significant for both technologies.
Sources: EIA (2020) — batteries and PSH return ~80% of electricity; Merus Power — RTE as performance guarantee; NREL ATB 2024 — Battery Storage (uses 85%); NREL ATB 2024 — PSH (80%, range 70–87%)
2. Typical discharge duration
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Dominant current deployments | 1–4 hours | 6–24+ hours |
| Emerging capability | Up to 8–10 hours | Up to weeks (very large reservoirs) |
| UK example | Most GB batteries: 1–2 hours | Dinorwig: ~5 hours; Coire Glas (planned): ~20 hours |
Duration is the organising variable for comparing these two technologies. As of the early 2020s, more than 90% of new U.S. grid-scale battery capacity has a duration of 4 hours or less. The fleet-weighted average duration grew from ~46 minutes in 2015 to ~1.5 hours in 2019. The 4-hour duration dominates because it aligns with peak-shaving windows and because lithium-ion capital costs scale linearly with energy capacity — extending from 2 to 8 hours roughly quadruples the energy-side capital cost.
PSH duration is set by the volume of the upper reservoir relative to the turbine capacity and is determined at design by site topography. Coire Glas (1,500 MW / 30 GWh, Scotland, consented 2024) would deliver approximately 20 hours of storage. Bath County (USA, 3 GW) has approximately 11 hours. Fengning (China, 3.6 GW / 40–60 GWh) yields ~11–17 hours.
Sources: EIA — Duration of utility-scale batteries depends on use; NREL — Moving Beyond 4-Hour Li-Ion Batteries (2023); NREL ATB 2024 — PSH; SSE Renewables — Coire Glas
3. Response time
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Rated power output from idle | 10–100 milliseconds | <70 seconds (cold start to full generation, per GE Vernova) |
| From spinning reserve/standby | ~10–100 ms | <15 seconds |
| Full pumping from standstill | N/A | ~5 minutes |
| Pumping mode regulation (fixed-speed) | Continuous modulation | None — must run at full load or off |
| Pumping mode regulation (variable-speed) | Continuous modulation | ±20–30% modulation possible |
BESS responds in milliseconds — orders of magnitude faster than any thermal or hydro plant. Grid-forming BESS begins injecting/absorbing power proportional to the rate-of-change-of-frequency within the first cycle (20 ms at 50 Hz). This response time is consistent regardless of prior operating state.
PSH response time is mode-dependent. In spinning reserve or synchronous condenser mode, output can be adjusted in under 15 seconds. From standstill, GE Vernova states "less than 70 seconds to switch from idle to full load for units up to 400 MW." The NHA cites "approximately 90 seconds."
The fixed-speed PSH pumping mode limitation is operationally significant: during pumping, a fixed-speed unit must either run at full designed power or shut down — it cannot modulate. This constrains its ability to provide frequency response as a load. Variable-speed units overcome this by allowing ±20–30% power modulation in pumping mode.
Sources: GE Vernova — PSH (<70 s from idle); NHA — ~90 s; MDPI PSH Review — <15 s in standby; Sungrow — 19-second black start demonstration; NHA — variable-speed PSH response
4. Cycle life and design life
| Li-ion BESS (NMC) | Li-ion BESS (LFP) | Pumped Storage Hydro | |
|---|---|---|---|
| Cycle life (to 80% SoH) | ~1,500–2,000 cycles | ~3,000–6,000+ cycles | Effectively unlimited |
| Design life (warranted/modelled) | 10–15 years | 15–20 years | Dam: ~100 years; electromechanical: ~40+ years |
| NREL ATB design life assumption | — | 15 years | 80 years (LCA base case) |
LFP (Lithium Iron Phosphate) has become the dominant grid-scale battery chemistry globally since ~2022, displacing NMC (Nickel Manganese Cobalt) for most new utility-scale projects. Its superior cycle life makes it the standard for grid applications.
PSH has no equivalent to electrochemical cycle degradation. The civil infrastructure (dams, reservoirs, tunnels) is designed for 80–100 years. Many existing plants have operated for 50+ years with turbine refurbishments. Fengning and Bath County are expected to operate well into the 22nd century.
Contested area: BESS "design life" can be extended through module replacement and active thermal management, blurring the boundary between 15-year and 20-year economic lives. Some operators plan for two battery replacement cycles within a 30-year project financing horizon.
Sources: NREL ATB 2024 — Battery (15-year design life); NREL ATB 2024 — PSH (80-year LCA base case); Springer — Life-cycle impacts of PSH (50–150 years); NREL LCA — Closed-loop PSH (80-year base case); ScienceDirect — Grid duty cycle comparison by chemistry
5. Degradation
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Annual capacity fade | ~1–3%/year (LFP, well-managed); up to 7%/year (NMC, adverse conditions) | Negligible — no equivalent mechanism |
| Best available empirical figure | 1.37%/year (utility-scale LFP, Italy, ScienceDirect 2023) | N/A |
| Per-cycle degradation | LFP: ~0.01–0.02%/cycle; NMC: ~0.05–0.1%/cycle | N/A |
BESS degradation is driven by two mechanisms: calendar aging (time and temperature-dependent) and cycle aging (use-dependent). Both mechanisms apply simultaneously. Operating strategy has a major effect: restricting the operating window to 10–90% state-of-charge (SoC) and avoiding elevated temperatures can substantially reduce degradation.
The 1.37%/year figure from a peer-reviewed study of a real utility-scale LFP system in Southern Italy (356 equivalent full cycles over 3 years) is the best available empirical primary source for a well-managed modern system. Higher figures (3–7%/year) likely reflect NMC systems, high-frequency regulation duty cycles, hot climates, or suboptimal management.
Contested area: Degradation figures vary enormously across sources (1–7%/year range). The National Center for Energy Analytics has cited higher figures to argue against battery deployment; the ScienceDirect empirical study of modern LFP represents a more optimistic real-world outcome. Chemistry (LFP vs NMC), climate, duty cycle, and management strategy all significantly affect outcomes.
Sources: ScienceDirect — Utility-scale LFP aging (1.37%/year); Nature Energy — Multi-year home storage degradation (2–3%/year, heavier cycling); National Center for Energy Analytics (3–7%/year); Thunder Said Energy — degradation causes
6. Power-to-energy ratio flexibility
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Can power (MW) and energy (MWh) be sized independently? | Yes — largely decoupled at design stage | Limited — coupled by civil/hydraulic design |
| Post-construction flexibility | Moderate (adding modules; adjusting inverter loading) | Very limited — requires major civil works |
| Minimum practical duration | <15 minutes | ~6 hours (site-determined) |
BESS duration is the ratio MWh/MW, set at design time but with meaningful flexibility. Adding battery racks extends duration; oversizing the inverter reduces the ratio. This means a BESS can be optimised for specific revenue streams (e.g., a 15-minute frequency response asset vs. a 4-hour energy arbitrage asset) from a common platform.
In PSH, power is set by turbine-generator unit sizes; energy is set by upper reservoir volume and head — both fixed by civil engineering at construction. Altering either dimension post-construction requires major new works. Some expansion is possible (Cruachan in Scotland is being expanded from 440 MW to 600 MW by Drax), but this is the exception rather than the rule.
Sources: Gridcog — Storage duration explainer; NREL — Grid-Scale BESS FAQ (2019); EIA — Duration depends on use; NREL ATB 2024 — PSH site characterisation
7. Self-discharge and parasitic losses
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Idle energy loss rate | ~2–5%/month (cell self-discharge + BMS power draw) | Near zero (potential energy is stable) |
| LFP cell self-discharge | <2%/month at room temperature | N/A |
| Physical mechanism | Electrochemical reactions (SEI evolution, cathode oxidation, micro-shorts) + BMS auxiliary power | Minor evaporation/seepage from open reservoirs; near zero for closed-loop systems |
Self-discharge is a significant consideration for long-duration storage scenarios. For a BESS holding charge for a week, approximately 1–2% of stored energy may be lost before discharge begins. For a month, this reaches 5–10%. This makes BESS poorly suited for seasonal storage without regular cycling.
PSH holds stored energy as gravitational potential of water in the upper reservoir. Evaporation from open-loop reservoir surfaces represents a minor water (and therefore energy) loss, particularly compensated by rainfall in the UK's wet climate. Closed-loop PSH (both reservoirs purpose-built, not connected to a river) minimises this further.
PSH units in synchronous condenser mode (spinning with no hydraulic flow) consume auxiliary power for cooling, lubrication, and control — but this is a service provision choice, not a self-discharge loss.
Sources: EL-CELL — self-discharge explainer; Anern — LFP self-discharge (<2%/month); DOE — Open vs closed-loop PSH comparison
8. Scalability
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Regulatory minimum (utility-scale) | 1 MW (EIA definition) | N/A |
| Practical commercial minimum | ~5–10 MW | ~100 MW (economic viability threshold) |
| Typical grid-scale range | 10 MW – 1+ GW | 100 MW – 3+ GW |
| Largest demonstrated single project | ~1 GW / 7.8 GWh (Saudi Arabia, 2025) | 3,600 MW / 40–60 GWh (Fengning, China) |
| UK pipeline largest project | ~1 GW (Trafford Power, in development) | 1,500 MW / 30 GWh (Coire Glas, consented) |
BESS is highly modular — standard containerised units of 2–5 MW are stacked to any required capacity, with no inherent upper limit imposed by the technology itself. Land availability, grid connection capacity, and supply chain are the practical constraints.
PSH economics strongly favour large scale. The fixed costs of civil infrastructure (permitting, dam construction, tunnels, access roads, grid connection) are poorly distributed below ~100 MW. The NREL ATB characterises viable U.S. PSH sites at 162–3,589 MW per site, reflecting real topographic constraints. Very small PSH (micro-PSH using abandoned mines or purpose-built small reservoirs) in the 5–30 MW range has been studied but rarely built commercially.
Sources: EIA — Battery storage capacity 2024; NREL ATB 2024 — PSH (162–3,589 MW range); PV Magazine — Saudi 7.8 GWh BESS; Tandfonline — Small PSH case studies
9. Synchronous inertia
This parameter has growing importance as coal and gas generators retire from the UK grid. Inertia is the tendency of large spinning masses to resist sudden changes in frequency — a key grid stability property.
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Native synchronous inertia | None (inverter-connected) | Yes — fixed-speed units are synchronous machines (H ~4 s) |
| Synthetic/virtual inertia | Yes — via grid-forming (GFM) inverters; H ~1–6 s (software-tuned) | Variable-speed units: virtual inertia via converter control |
| Synchronous condenser mode | No | Yes — can spin with no hydraulic flow to provide inertia + reactive power 24/7 |
| Key distinction | Requires available energy; control-system dependent; can be "switched off" | Passive physical property when online; cannot be disabled; state-of-charge independent |
PSH's rotating machinery (fixed-speed turbine-generators) is electromagnetically coupled to the grid when operating, providing the same physical inertia mechanism as coal, gas, and nuclear generators. This inertia is passive — it exists as a physical property whenever the machine is spinning, regardless of grid conditions, and does not consume stored energy from the reservoir.
PSH units can operate as synchronous condensers — spinning with no hydraulic flow — to provide inertia and reactive power support independently of storage state. This is a significant operational advantage: inertia can be provided 24/7 at near-zero marginal cost.
Grid-forming BESS can emulate inertial behaviour through software control (synthetic inertia), delivering power proportional to rate-of-change-of-frequency. The Hornsdale Power Reserve (150 MW, South Australia) began providing ~2,000 MWs of equivalent inertia commercially in 2022 — the first large-scale deployment. Australia now has 74% of its battery storage pipeline committed to GFM capability.
Critical distinction: Synthetic inertia from GFM-BESS requires: (a) the battery to have available stored energy; (b) the control system to function correctly; (c) the inverter to remain online during the fault event. Real synchronous inertia from PSH requires none of these conditions — it is an automatic physical response. This distinction matters for grid security design, particularly for rare but severe fault scenarios.
Sources: AEMO — Quantifying Synthetic Inertia from GFM-BESS (2024); ARENA — Large-scale battery storage as inertia substitute; PV Magazine — Hornsdale inertia services (2022); NHA — PSH grid services; IRENA — Innovative PSH operation (2020); MDPI — Inertia optimisation with PSH (H=4 s)
10. Black-start capability
| Li-ion BESS | Pumped Storage Hydro | |
|---|---|---|
| Can restart grid after total blackout? | Yes — grid-forming BESS only; demonstrated commercially | Yes — well-established; decades of operational track record |
| Fastest demonstrated black start | 19 seconds (Sungrow 30 MW GFM, 2024) | Site-dependent; typically minutes to first generation |
| Qualification | Requires grid-forming inverter (additional cost/complexity) | Standard capability for all generating units |
| External power required? | No (BESS is self-powered from battery) | No (turbine driven by stored water head) |
Traditional BESS using grid-following inverters cannot provide black-start service — they require an existing grid voltage/frequency reference. Grid-forming BESS can establish an independent voltage and frequency reference, enabling black-start. Sungrow demonstrated a 19-second black start at a 30 MW test facility in 2024, verified by TÜV Rheinland.
PSH has provided black-start service for decades. In generation mode, a PSH unit self-starts by opening its inlet valves, using the potential energy of stored water — requiring no external power. It then provides cranking power to restart nearby conventional generators.
Sources: Sungrow 19-second black start (2024); FFD Power — black start explainer; GE Vernova — PSH black start; IEEE — Variable-speed PSH for black start (2024); IHA factsheet; Vattenfall — PSH Swiss Army knife
Summary comparison table
| Parameter | Li-ion BESS | Pumped Storage Hydro |
|---|---|---|
| Round-trip efficiency (AC-to-AC) | 84–92% (fleet avg ~82%; modern LFP ~88–92%) | 70–87% (fleet avg ~79%; central estimate 80%) |
| Typical discharge duration | 1–4 hours (dominant); up to 10 hours (emerging) | 6–24+ hours (site-dependent) |
| Response time | 10–100 milliseconds | <15 s (spinning reserve); <70 s (cold start to generation) |
| Cycle life | NMC: ~1,500–2,000 cycles; LFP: ~3,000–6,000+ cycles | Effectively unlimited |
| Design life | 10–20 years (NREL ATB: 15 years) | 40–100+ years (dam ~100 yrs; electromechanical ~40 yrs) |
| Annual capacity degradation | ~1–3%/year (LFP, well-managed); up to 7%/year (NMC, adverse) | Negligible — no equivalent mechanism |
| Power-to-energy sizing | Largely decoupled — flexible at design stage | Coupled by civil/hydraulic design; limited post-construction flexibility |
| Self-discharge (idle) | ~2–5%/month (cell + BMS) | Near zero — potential energy is stable |
| Minimum grid-scale size | 1 MW (EIA threshold); 5–10 MW typical commercial minimum | ~3 MW technically; ~100 MW for economic viability |
| Maximum demonstrated size | ~1 GW / 7.8 GWh (Saudi Arabia, single or multi-site) | 3,600 MW / 40–60 GWh (Fengning, China) |
| Synchronous inertia | None natively; GFM inverters provide synthetic inertia (H ~1–6 s) | Yes natively (H ~4 s); synchronous condenser mode available |
| Black-start capability | Yes — GFM-BESS only; demonstrated at 30 MW in 19 s | Yes — well-established; decades of track record |
Key technical caveats
- Efficiency measurement boundary matters. DC cell-level BESS figures (90–96%) are commonly cited in marketing but are not comparable to AC-to-AC system figures (84–92%). PSH efficiency also varies with head and technology type. Fleet averages (82% BESS, 79% PSH per EIA 2019) are the most comparable single figures.
- BESS degradation varies enormously. The 1.37%/year empirical figure for a well-managed modern LFP system is more optimistic than the 3–7%/year range sometimes cited; the difference reflects chemistry (LFP vs NMC), climate, duty cycle, and management strategy.
- PSH response time is mode-dependent. From spinning reserve: seconds. From cold start: <70–90 seconds. The cold-start figure is relevant for dispatch from idle; the spinning-reserve figure is relevant for frequency response.
- Synchronous vs. synthetic inertia is qualitatively different. PSH's passive physics-based inertia cannot be fully replicated by software-controlled synthetic inertia, though GFM technology is advancing rapidly. The Hornsdale demonstration (2022) showed commercial viability, and the distinction is narrowing in practice.
- PSH minimum viable scale is economically, not technically, determined. While 3 MW is technically feasible, projects below ~100 MW face unfavourable unit economics compared to alternatives.