Working contributions to the climb

megawatt.fun

Small experiments and tools that support the larger energy roadmap. Each should make one real decision easier to inspect.

1 MW ≈ 200 Texas homes at peak · 1 MW for 1 hour = 1 MWh

Dispatch evaluation · synthetic experiment

What survives forecast error?

Compare idle, clock-based, forecast-committed and hindsight battery schedules under price error, losses and a household reserve. The Go planner selects profitable partial cycles, and a separate replay checks physical constraints and realized settlement.

A measured failure is useful evidence. With 88% efficiency and a 5 kWh reserve, the strongest synthetic error setting gives the forecast strategy 38 losing days out of 120. Its mean is $0.145/day versus $0.097 for the clock baseline. These are synthetic stress outcomes, not validated customer economics or a production controller.

go test ./...
go run ./cmd/dispatch-study

Tool 001 · fleet dispatch

One day's arbitrage is a scenario, not a valuation.

A day-ahead dispatch planner for a fleet of home batteries, written in Go and run on real ERCOT day-ahead prices (hub average, captured 2026-08-13). It charges the midday solar trough, discharges the evening ramp, and refuses to cycle when the spread does not clear round-trip losses.

ERCOT DAM hub average · $/MWh · hour ending charge discharge idle
$11.02
cheapest hour bought (10:00)
$38.46
priciest hour sold (22:00)
$0.43
net per battery, this day (25 kWh / 5 kW / 88% RT)
$7,312
per day across 17,000 battery homes (Series D coverage, Aug 2026)

This is not a revenue floor. One cycle, energy-only, supplied prices, no forecast error or backup reserve. Omitted revenues and optimistic operating assumptions pull in opposite directions; one day cannot establish fleet economics. The local Go study compares forecast schedules with clock and idle baselines under synthetic price error, losses, and reserves.

Tool 003 · live prices

ERCOT, right now

Real-time 15-minute and day-ahead hourly settlement prices for every ERCOT hub and load zone, from the same public feed the dispatch planner uses, refreshed every minute at the edge.

Tool 002 · the climb

Kardashev: where humanity actually is

The earlier scale analysis is preserved as an archive. It explores the formula, the assumptions, the climb since 1900, and how many years of growth Type I actually takes at any growth rate you pick.

Study · the titans

The greatest energy companies of all time

Standard Oil to Base Power: the five mechanisms that repeat across 150 years of energy empires, with sourced numbers, one table of seven empires, six lessons for a builder, and the contested claims listed instead of dropped.