Batteries, pumped hydro and long-duration storage

Energy arbitrage earns discharge receipts and incurs charging costs. Charging 100 MWh at $30/MWh and exporting 85 MWh at $100/MWh gives $8,500 − $3,000 = $5,500 of gross energy margin. It is not net profit. A discharge-based cycling allowance of $8/MWh would reduce this example by $680 before other fees and costs.

The energy-only break-even sale price is charging price divided by round-trip efficiency when both are measured consistently. At $40/MWh and 80% efficiency, one discharged MWh requires 1.25 purchased MWh, costing $50. Add a $10 cost per discharged MWh and the break-even sale price becomes $60/MWh. Negative charging prices change the sign of the purchase cost; they do not remove physical constraints.

A battery can potentially provide energy and other services, but each commitment uses power headroom, response capability or energy inventory. If a 100 MW inverter is already exporting 80 MW, its simple upward active-power headroom is 20 MW before additional electrical limits. Promising two simultaneous 20 MW increases would not be feasible under that envelope.

A professional revenue model therefore co-optimises compatible obligations. Model activation, reserve duration, availability, recovery charging and shared-connection limits. Use current Australian registration, technical and market requirements for the actual service. Do not infer eligibility or payable revenue merely from fast equipment response.

Usable energy, power capability and efficiency can evolve differently. If deliverable energy declines from 200 to 180 MWh with power held at 100 MW, duration falls from two hours to 1.8 hours. Temperature, state-of-charge windows, calendar ageing and cycling patterns can affect the operating plan. Warranty conditions must be read as specific contractual conditions, not generic industry guarantees.

Treat augmentation or replacement as dated cash flows and operational events. A per-cycle or per-MWh degradation allowance is a modelling simplification; it should not conceal an inconsistent physical lifetime model. Compare project alternatives using net operating cash flow, capital, replacement, residual value and risk rather than gross market receipts alone.

Optimising against realised future prices produces a perfect-foresight benchmark. It can reveal physical upper bounds under a chosen model, but it is not an executable trading strategy. A realistic test uses the forecasts and information available at the time each decision would have been made, including forecast errors and operational overrides.

Keep initial and terminal state of charge consistent, or assign a defensible terminal inventory value. Otherwise a model can empty a full battery at the end and present inventory liquidation as recurring profit. Stress narrower spreads, outages, efficiency loss, service saturation, connection restrictions and forecast bias. Report downside and sensitivity alongside a central estimate.

Deliver a reproducible interval model with an input sheet, operating constraints, energy balance, charging and discharge cash flows, service commitments and an assumptions register. Explain the difference between gross revenue, gross margin, contribution after variable costs and project cash flow. Reconcile energy and cash independently.

End with a decision: which configuration or operating policy is preferable, under what price and technical assumptions, and what would reverse the choice? Link the work to storage fundamentals (Topic 5), system services (Topic 16), risk (Topic 23) and network connection limits. All prices in this lesson are synthetic teaching inputs, not market forecasts.

Pumped storage shifts energy by moving water between elevations. In a simplified constant-head calculation, gravitational energy is density × gravity × head × usable water volume. With water density 1,000 kg/m³, gravity 9.81 m/s², 300 m head and 1,000,000 m³ usable volume, stored potential energy is 2.943 × 10¹² joules, or 817.5 MWh. At 90% generation efficiency it delivers 735.75 MWh before other losses.

Power is a separate constraint. At 300 m head, 20 m³/s flow and 90% efficiency, generation is about 52.97 MW. Actual head changes with reservoir levels; hydraulic losses, environmental limits, machinery ranges and water availability matter. Pumping input exceeds later delivered output. Do not count recycled pumped water as a new primary energy source.

• US Department of Energy: pumped-storage hydropower

A project described as long-duration still needs a stated MW and usable delivered MWh. An eight-hour 100 MW requirement is 800 MWh delivered. If the whole electricity-to-storage-to-electricity route has 40% efficiency, replenishing that delivered energy needs 2,000 MWh of input under the simplified cycle boundary. Duration alone says nothing about recharge time or cost.

Compare pumped hydro, electrochemical storage, thermal routes and fuels only after specifying the useful output. Heat storage serving an industrial process is not directly equivalent to electrical output. Hydrogen reconversion includes multiple conversion stages. Assess location, civil works, water or fuel infrastructure, self-discharge, cycling frequency, degradation, safety and lead time with technology-specific evidence rather than a single universal ranking.

Begin with a chronological demand or supply gap and a common connection boundary. Simulate dispatch, reserve floors, losses, recharge limits and repeated adverse events. For each candidate, show energy delivered when needed, unserved residual, lifetime cycling assumptions and the cost items included. A daily arbitrage winner may be unsuitable for infrequent multi-day resilience.

Separate screened feasibility from investable design. Reservoir geology, connection capability, manufacturer warranties, approvals, construction risk and independently reviewed models are project evidence, not details an annual revenue estimate can replace. State which decision is justified now and which studies must be completed before committing capital.