The science
I = sum_i ( s_i * g_i ) / sum_i s_i blended intensity, g CO2/kWh
E_CO2 = I * Q 1 g/kWh x 1 TWh = 0.001 Mt
dI/ds_i = g_i - I what one more unit of source i does
dispatch: residual_h = D_h - (nuclear + hydro + wind_h + solar_h)
thermal_h = residual_h (gas : coal in the ratio of their shares)
| Source | g CO2-eq/kWh (lifecycle) |
|---|
| Coal | 820 |
| Natural gas (CCGT) | 490 |
| Solar PV (utility) | 48 |
| Hydro | 24 |
| Nuclear | 12 |
| Wind (onshore) | 11 |
The factors are lifecycle medians as quoted from memory of the IPCC AR5 Working Group III annex (2014): they include fuel supply, construction and decommissioning, not just the chimney. Real plants scatter widely around them (hydro reservoirs with decaying biomass or leaky gas supply chains can be far higher), so treat the numbers as typical, not exact.
The annual mix gives the average. The day view then asks what happens hour by hour: demand follows a two-hump daily curve, solar follows the sun, wind drifts, nuclear and hydro run flat, and gas and coal fill whatever is left. When clean energy exceeds demand, the surplus charges the battery (with losses) and the rest is curtailed. With no curtailment and no storage the day average equals the annual formula exactly (check 3).
Simplifications. One region, no transmission or imports, no ramp limits, no minimum-stable-load for thermal plants, hydro treated as flat run-of-river output, no seasonal storage, fixed typical profiles rather than measured data, energy shares not installed capacity, and a thermal fleet always large enough. Marginal emissions (what the next kWh causes) differ from the average shown here. Educational model, not a forecast.
Running...