Pypsa Asset Economics

PyPSA Asset Economics

• PyPSA default answers "what minimizes SYSTEM cost", not "what does THIS asset earn".
• ! correct (or disclose) three biases in every revenue number.

Bias 1 - perfect foresight

• optimizer sees all prices/weather in advance.
• perfect-foresight storage arbitrage overstates revenue ~10-30% (daily cycling) | more for multi-day strategies.
• Fixes, increasing effort:

1. report perfect-foresight numbers WITH label + haircut range.
2. rolling horizon dispatch (24-48h windows, limited lookahead) -> native optimize_with_rolling_horizon + its 3 traps: pypsa-solve-and-debug/references/performance.md item 4.
3. dispatch vs FORECAST price series -> settle vs outturn. Multi-market sequence (DA -> ID -> imbalance) -> references/multi-market-dispatch.md.

Bias 2 - system vs merchant optimization

• system-cost run -> asset dispatched to help SYSTEM; merchant asset maximizes own profit vs prices.
• price-TAKING asset setup:

1. SOLVE: system model WITHOUT asset (or asset marginal) -> price series.
2. BUILD: single-asset network = one bus + ONE bidirectional market-interface Generator (p_min_pu=-1, marginal_cost = +price(t)) + asset -> solve. Cost linearity: buy (p>0) pays price | sell (p iterate | accept system-run dispatch as equilibrium approximation. STATE: which.

Bias 3 - wholesale price != asset price

• merchant assets pay grid fees | levies | taxes ON TOP of wholesale — jurisdiction-specific. ! storage double-charging (fees on charge AND discharge) + exemptions w/ SUNSET dates can flip a BESS business case alone. STATE: which non-market cost components included; never quote model arbitrage as investable w/o them.

Revenue accounting (post-solve, any run)

• RUN: scripts/revenue_report.py solved_network.nc.
• per-asset decomposition: energy revenue = sumt p lambdabus(t) w(t) | energy cost (links' bus0 side) | VOM | annualized capex | net margin.
• USES: n.busest.marginalprice -> run must produce meaningful prices (which runs do: pypsa-market-design).
• figures from this output -> pypsa-reporting chart-catalog #7 (diverging net-margin bar).

Metrics

• LCOE/LCOH = (annualized capex + fixed O&M + sum variable costs) / annual output.
• LCOH: electricity at MARGINAL price vs PPA price shifts answer EUR/kg levels. STATE: power sourcing assumption. 1 kg H2 = 33.33 kWh LHV.
• capture price = market value = revenue-weighted mean price; capture RATE = capture price / time-weighted mean.
• VRE capture rates fall w/ penetration (cannibalization). ! never extrapolate today's capture rate -> high-RES scenario.
• IRR/NPV on model cashflows: STATE discount rate + lifetime consistent w/ annuity in capital_cost | else double-discount.

Technology references - READ matching one

• references/storage-revenue.md -> BESS + TES arbitrage | stacking | degradation cost.
• references/multi-market-dispatch.md -> DA + intraday + imbalance sequencing | 15-min granularity | co-optimization traps.
• references/hydrogen-economics.md -> electrolyzer utilization frontier | LCOH.
• references/heat-economics.md -> heat pump vs boiler dispatch economics.
• references/vre-ppa.md -> capture prices | cannibalization | PPA structures.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: nimabahrami
• Source: nimabahrami/pypsa-skills-kit
• License: MIT

Install and usage instructions live in the source repository linked above.