Skip to content
GridArb

How we calculate

What this report is: a screening check before you spend on a site. What it is not: a bank-ready revenue study. Every rule with the date it was last checked: Assumptions and dates. Every source with its terms: Data sources and terms.

Data sources and dates

PJM day-ahead and real-time hourly zonal LMPs
2023-12-31 to 2026-10-12; from PJM Data Miner 2 (dataminer2.pjm.com, feeds da_hrl_lmps and rt_hrl_lmps) or gridstatus.io's hosted copy (gridstatus.io). Real-time zonal values in the gridstatus feed are estimates.
PJM Regulation market results
2024-01-01 to 2026-10-13 (24,383 hours); Data Miner 2 feed reg_market_results (RMCCP, RMPCP, mileage, performance score).
PJM capacity prices
Base Residual Auction reports for 2026/27 (July 2025), 2027/28 (December 17, 2025) and 2028/29 (July 14, 2026): pjm.com RPM.
PJM ELCC class ratings
Published with each BRA, 2025-2026; the 2-hour rating is GridArb's assumption (energy / 4 at the 4-hour class) until confirmed in PJM's ELCC calculator.
NJ GSESP
NJ BPU order of June 18, 2025 (program) and March 4, 2026 (Phase 1 Tranche 1 awards, $60,000 to $81,840 per MW-year for 15 years): nj.gov/bpu.
Federal ITC
Section 48E clean electricity investment credit, 30% base for standalone storage (2026 summary); FEOC content rules and prevailing-wage requirements apply.
Battery cost benchmark
NREL ATB 2025 (Utility-Scale Battery Storage, Moderate, year-2024 values, 2023 USD), OEDI submission 8759, workbook dated 2026-08-27, downloaded 2026-10-12: atb.nrel.gov.
Interconnection deposits
PJM Manual 14H, Generation and Transmission Interconnection Process (2026): study deposit $75,000 plus $4,000/MW readiness deposit for 0-20 MW.
Queue, substations, geocoding
interconnection.fyi (PJM queue map), OpenStreetMap Overpass (substations), US Census Geocoder (addresses). Utility hosting-capacity maps are read by hand.

Energy arbitrage: the optimizer

For each day we solve a small linear program that knows that day's hourly prices in advance and picks the hours to charge and sell, within the battery's power, energy, efficiency, state of charge window, cycle limit and a degradation cost per MWh. The state of charge carries from one day to the next. That is a ceiling: real operators bid on forecasts. The achievable line is 80% of the ceiling (range 70% to 90%). We test two calendar years (2024 and 2025) of real PJM prices; 2025 is the headline year.

The forecaster (owner schedules)

For tomorrow's plan we train a gradient-boosted model per zone on past day-ahead and real-time prices, weather, load forecasts and outages, and publish a real-time price range (p10 to p90) for each hour. Its record against actual prices is public at /track-record. It plays no part in the site report numbers.

Capacity

Revenue = effective MW x ELCC x clearing price x 365. Clearing prices used: 2026/27 $329.17/MW-day, 2027/28 $333.44/MW-day, 2028/29 $325.00/MW-day (all at the FERC-approved cap). ELCC storage ratings: 2026/27: 4 h 0.50, 6 h 0.58, 2027/28: 4 h 0.58, 6 h 0.67, 2028/29: 4 h 0.59, 6 h 0.68. A battery under 4 hours is rated as energy / 4 at the 4-hour class, an assumption to be confirmed in PJM's ELCC calculator. The price collar ends after 2029/30; later prices are unknown, so the fifteen-year view uses labelled low / base / high paths ($175 / $250 / $500 per MW-day) with ELCC sliding to 0.40 by 2035.

Regulation

History: the credit per MW-hour of regulation = RMCCP + RMPCP x the performance multiplier PJM paid (mileage ratio x performance score, from the hour's credit totals in PJM's zone billing feed; the market-results feed supplies mileage and performance where the totals are missing). Today's history: pre-redesign average $36.22, post-redesign average $116.16 (median $36.76), last 12 months $115.82 per MW-hour. The history is shown as context only ("what the last 12 months paid if you had cleared every hour").

First operating year. Every number that feeds payback and NPV starts in the project's first operating year (COD), not today: by default 2029 for a transmission-connected project (PJM's cycle-based interconnection process) and 2028 for a distribution-connected one (the utility's WMPA route); cod_year=... on any report changes it. The report also shows "if this battery were running today", the same stack at today's prices and competition, as context only.

Forward price path, $ per MW-hour: base $150 in 2026 falling to $35 (Modo Energy's long-run level) by 2029, then flat; low the pre-redesign average falling to the battery's cost floor (margin adder plus energy losses, about $4.30) by 2028; high the last-12-month average falling to $35 by 2033 and never below the base price in any year (Modo Energy's PJM regulation outlook).

Cleared-hours share from competition: a battery clears min(80%, 750 MW requirement / expected battery fleet) of the hours it offers. Fleet path (PJM's Energy Storage Resource paper and Modo Energy): 2026 600 MW, 2027 2,500 MW, 2028 4,000 MW, 2029 5,000 MW, 2030 onward 7,000 MW. The low case uses 1.5x that fleet, the high case 0.5x. Regulation runs in the idle hours of a once-a-day cycle (24 minus two times the duration) and, where it beats the hour's arbitrage margin, in the cleared share of the active hours, with that share of the arbitrage forgone. From December 2026 the price is shown split into RegUp and RegDn at 50/50, a placeholder until PJM publishes the two. Regulation prices are expected to fall as more batteries join PJM: the market is a few hundred MW, so new storage moves it quickly. Treat this line as upside, never as the base for a loan.

Benchmark check, apples to apples: the first operating year's total per MW-year is compared with Modo Energy's post-redesign PJM fleet average (about $62/kW-month, $744,000 per MW-year), and the 15-year average total per MW-year with Modo's central case (about $84,000 per MW-year for a 4-hour battery, 2027-2049). The report shows an "above industry benchmark" warning, and GridArb's own test fails, only when the 15-year average is above 1.5x the central case.

Grid connection check

Hosting capacity comes from the utilities' own maps, read through the public ArcGIS feature layers behind them: PSE&G's service points (circuit availability in kW and voltage), JCP&L's solar accommodation limit (remaining kW AC, floor "less than 100 kW"), the Pepco Holdings feeder layer for Atlantic City Electric (allowable kW and kV; last edited November 2022, so dated) and Orange & Rockland's storage map for Rockland Electric (feeder discharge and charge capacity in MW). GridArb sends one small radius query per site, caches it 30 days, refreshes each layer's data date monthly and stores the small ACE layer whole; the report shows the nearest feeder's figure with the map link and the data date. If a layer cannot be read, the report shows the map link and "check the map manually". Utility terms: the maps are screening tools; the interconnection study decides.

PJM queue: the public cycle service request status list (about 1,800 cycle-process projects, AE1 onward) is imported weekly. It carries no coordinates, so active projects with the zone's transmission owner are matched by name to the OpenStreetMap substations within 10 miles of the site and shown at the nearest substation, within 5 miles and within 10 miles; unmatched projects are counted, not placed. Network-upgrade costs: PJM's cycle System Impact Study executive summaries (TC1 Phase 2 and 3, TC2 Phase 1 and 2) are parsed weekly into upgrades and per-project cost allocations; the report shows the allocated cost per MW for recent projects in the zone as a warning signal (yellow above $150,000 per MW, red above $500,000; GridArb thresholds).

Jurisdiction: a planned connection below 69 kV follows the distribution route (N.J.A.C. 14:8-5 with the utility plus a PJM WMPA for wholesale sales); 69 kV and above follows PJM's cycle process (Manual 14H deposits: $75,000 plus $4,000 per MW for 0-20 MW). The report shows the deposits and typical timelines for the route chosen. A pre-application report request (default fee $300; NJ's PAVE process may not be live at every utility yet) and an authorization letter are drafted from the report's inputs. PJM Queue Scope is linked with a short how-to, and a site-report customer can attach the Excel export to the private report page.

Connection score: green, yellow or red with the reasons listed, from the feeder's available kW against the battery's kW (1.0 and 0.5), the nearest substation's distance (3 and 10 km), queued projects at the same substation (1 and 3) and within 5 miles (3 and 8), and the upgrade-cost signal. Screening thresholds, GridArb assumptions. A box lists what only a study can tell you: exact feeder capacity, upgrade costs, capacity rights, protection requirements.

GSESP and the ITC

The NJ Garden State Energy Storage Program pays a fixed incentive for 15 years to projects of at least 5 MW that win a competitive award (Tranche 1 range $60,000 to $81,840 per MW-year). It is never counted in the default total. For sites of 5 MW and more the report adds a second line, "with GSESP award" ($60,000 / $75,000 / $81,840 per MW-year for 15 years), labelled "only if awarded; Tranche 2 decision expected December 2026". The federal ITC is 30% of the battery cost in year 0 (Section 48E); your interconnection estimate is left out of the credit, which is the conservative reading.

Cost, payback and the fifteen-year view

Costs default to the NREL ATB benchmark (NREL ATB 2025 (Utility-Scale Battery Storage, Moderate, year-2024 values, 2023 USD), OEDI submission 8759, workbook dated 2026-08-27, downloaded 2026-10-12) plus a small-project premium; you can enter your own quote and your own interconnection estimate, and every report shows payback and NPV at cost -20% and +20%. Fixed O&M is 4% of capex a year; usable energy fades 2.5% a year with an augmentation when it falls below 80%. NPV uses an 8% discount rate over 15 years. The fifteen-year view is three labelled scenarios, not a forecast.

Grid connection

We show the nearest mapped substations and distances, PJM's study and readiness deposits for 0-20 MW, typical timelines, a link to count queue projects within 10 miles on interconnection.fyi, and a field for your own interconnection cost estimate. What GridArb cannot know: network-upgrade costs (only PJM's studies and the utility's facilities study decide them); distribution hosting capacity on your circuit (read it from the utility's map); permits, zoning and site control; the queue position and withdrawal of other projects ahead of you; real vendor quotes for the battery, the transformer and the switchgear.

Limits

  • Backtests on past prices are not guarantees; every revenue line carries a low / base / high.
  • Only the day-ahead market is used for arbitrage; real-time settlement, intra-hour volatility and co-optimisation are ignored.
  • Degradation is a flat cost per MWh, not a cycle-life model.
  • Capacity revenue needs Capacity Interconnection Rights (and a WMPA for distribution-connected projects); queue timing is not modelled.
  • Nothing here is a bid, an offer or financial advice. GridArb recommends; you decide.

Questions about a number? Contact.