More elaborate CO2 emission calculations

[ArnoClaude]

Description

Improve CO₂ emission modeling by first researching the most relevant emission sources and emission factors (ranked by impact), and only then implementing them. The objective is to move from a highly simplified CO₂ model to one that reflects reality closely enough to influence optimization decisions correctly.

Current Behavior

• Only grid electricity imports are considered as a CO₂ emission source.
• All grid imports use a single, global CO₂ intensity constant.
• No differentiation by region, grid market, or time.
• No CO₂ emissions from:
  • On-site fuel usage
  • Building own assets (e.g. PV, ESS)
  • Operating own assets
• This simplification can strongly distort results.

Proposed Behavior

1. Research first (ordered by importance)

Research and quantify, in order of expected impact:

1. Grid electricity CO₂ intensity
   • Regional differences between grid markets.
   • Time-varying intensity (hourly vs. annual averages).
   • Average vs. marginal CO₂ intensity.
2. On-site operational emissions (scope 1)
   • Fuel combustion for heat, CHP, generators, etc.
   • Emission factors per fuel type.
3. Embodied emissions of assets
   • CO₂ emitted by building own PV systems.
   • CO₂ emitted by building energy storage systems (ESS).
   • Typical lifecycle values (kgCO₂ per MW or per MWh capacity).
4. Operational emissions of own assets
   • PV: typically zero during operation but may include lifecycle amortization.
   • ESS: emissions caused by round-trip losses leading to additional grid imports.
5. CO₂ pricing / penalties
   • Differences by region and market.
   • Time-dependent CO₂ price trajectories.

2. Implementation after research

• Incrementally extend the CO₂ model following research results.
• Start with highest-impact sources and highest data confidence.

Implementation Notes

• Introduce emission factors:
  • grid_co2_intensity[region, time] (kgCO₂/MWh)
  • fuel_co2_factor[fuel] (kgCO₂/unit)
  • asset_embodied_co2[technology] (kgCO₂ per MW or MWh capacity)
• Model emissions as linear terms:
  • Operational: emissions = activity * emission_factor
  • Embodied: emissions = new_capacity * embodied_co2_factor
• Attribute ESS operational emissions via increased grid imports due to losses.
• Aggregate all emission sources into total_emissions.
• Apply CO₂ penalty in objective:
  • co2_cost = Σ emissions * co2_penalty[region, time]
• Keep a simplified fallback with a single global CO₂ factor.

Open Questions

• How far to go with embodied emissions vs. operational focus?
• Over which lifetime should embodied CO₂ be amortized?
• Required time resolution for grid CO₂ intensity (hourly vs. yearly)?
• Use average or marginal grid emissions for optimization decisions?
• Should CO₂ limits be soft (penalized) or hard (budget constraint)?