Troubleshooting#

Common issues and their solutions.

Installation Issues#

Conda Environment Creation Fails#

Symptom: Error during conda env create

Solutions:

  1. Update conda first:

    conda update -n base conda

  2. Try mamba (faster resolver):

    conda install -n base mamba
mamba env create -f envs/pypsa-gb.yaml

  3. Clear package cache:

    conda clean --all

Gurobi License Not Found#

Symptom: GurobiError: No Gurobi license found

Solutions:

  1. Check license file location:

    • Windows: C:\Users\<username>\gurobi.lic

    • Linux/Mac: ~/gurobi.lic

  2. Regenerate license (if academic):

    grbgetkey YOUR-LICENSE-KEY

  3. Use HiGHS instead:

    solver:
  name: "highs"

Snakemake Not Found#

Symptom: snakemake: command not found

Solutions:

  1. Ensure environment is activated:

    conda activate pypsa-gb
which snakemake

  2. Reinstall snakemake:

    conda install -n pypsa-gb snakemake

Workflow Issues#

“Missing Input Files”#

Symptom: Snakemake reports missing input files

Solutions:

  1. Check the file exists:

    ls path/to/expected/file

  2. Generate the missing file:

    snakemake path/to/missing/file -j 1

  3. Check for typos in scenario name

Rule Fails with Exit Code 1#

Symptom: Rule fails, exit code 1

Solutions:

  1. Check the log file:

    cat logs/<rule_name>/<scenario>.log

  2. Re-run with verbose output:

    snakemake target.nc -j 1 --verbose

  3. Check Python errors in log

“No Rule to Produce Target”#

Symptom: No rule to produce <target>

Solutions:

  1. Check target filename matches pattern

  2. Verify scenario is defined in scenarios.yaml

  3. Check for wildcards mismatch

Network Issues#

All Generators at Same Bus#

Symptom: All generators mapped to one bus (e.g., INDQ41)

Cause: Coordinate system mismatch (WGS84 vs OSGB36)

Solution: The issue is in spatial mapping. Check:

from scripts.spatial_utils import detect_coordinate_system

# Check bus coordinates
print(detect_coordinate_system(network.buses[['x', 'y']]))
# Should output: 'OSGB36'

Network Not Connected#

Symptom: Isolated buses, disconnected regions

Solution:

import pypsa

n = pypsa.Network("resources/network/my_network.nc")

# Check connectivity
from pypsa.topology import find_disconnected_buses
disconnected = find_disconnected_buses(n)
print(f"Disconnected buses: {disconnected}")

Negative Line Impedance#

Symptom: Warnings about negative impedance

Solution: Usually data issue. Check raw network files for data entry errors.

Solver Issues#

Infeasible Optimization#

Symptom: Solver returns “infeasible” or “unbounded”

Causes & Solutions:

  1. Demand exceeds supply:

    n = pypsa.Network("resources/network/pre_solve.nc")
print(f"Generation: {n.generators.p_nom.sum()/1000:.1f} GW")
print(f"Peak demand: {n.loads_t.p_set.sum(axis=1).max()/1000:.1f} GW")

  2. Network disconnected: See above

  3. Zero capacity lines:

    print(n.lines[n.lines.s_nom == 0])

  4. Missing generators: Check if integration ran correctly

Solver Runs Forever#

Symptom: Solver doesn’t converge

Solutions:

  1. Add time limit:

    solver:
  TimeLimit: 3600  # 1 hour

  2. Use barrier method:

    solver:
  method: 2
  crossover: 0

  3. Reduce problem size:

    • Use reduced network

    • Shorten solve period

    • Cluster network

Out of Memory#

Symptom: Process killed, memory error

Solutions:

  1. Reduce parallel jobs:

    snakemake target.nc -j 1

  2. Use network clustering

  3. Reduce timesteps:

    solve_period:
  start: "2035-01-01"
  end: "2035-01-02"  # Shorter period

Market Dispatch Issues#

ELEXON Prices Requested For A Future Scenario#

Symptom: Market solve errors when bid_offer_source: "elexon" is used for a modelled year after 2024.

Solution: ELEXON bid/offer data is historical. Use derived prices for future scenarios:

market:
  balancing:
    bid_offer_source: "derived"

Or use bid_offer_source: "auto" to prefer ELEXON only when valid historical files are available.

Balancing Solve Is Infeasible#

Symptom: Wholesale solves, but solve_balancing_mechanism fails.

Common causes:

  • BM participation filters exclude assets needed for feasibility

  • non-participants are set to behavior: "fixed"

  • interconnector or outage settings over-constrain the physical network

Start from permissive participation settings, then tighten filters:

market:
  balancing:
    participation:
      generators:
        mode: "all"
        behavior: "priced_out"
      storage_units:
        mode: "all"
        behavior: "priced_out"

Wholesale Price Spread Looks Too High#

Symptom: Averaging network.buses_t.marginal_price directly gives an unrealistic price spread in the wholesale network.

Solution: Use resources/market/{scenario}_wholesale_price.csv. The market reporting code filters to demand buses and avoids internal DC-link buses that can retain directional shadow prices.

Data Issues#

FES Data Not Found#

Symptom: Error loading FES data

Solutions:

  1. Check FES year is available:

    ls resources/FES/

  2. Regenerate FES data:

    snakemake resources/FES/FES_2024_data.csv -j 1 -R

DUKES Coordinates Missing#

Symptom: Generators without coordinates

Solutions:

  1. Check for manual coordinate file:

    ls data/generators/manual_coordinates.csv

  2. Run coordinate geocoding:

    python scripts/geocode_missing_coordinates.py

Weather Cutout Missing#

Symptom: “Cutout not found” for renewables year

Solutions:

  1. For years 2010-2024 (automatic Zenodo download):

    # Configure year in config/cutouts_config.yaml
# Then run cutout workflow
snakemake -s Snakefile_cutouts --cores 1

    Cutouts will be automatically downloaded from Zenodo (~5-10 minutes per year). No CDS API credentials required for these years!

  2. For other years (requires CDS API credentials):

    • Register at: https://cds.climate.copernicus.eu/user/register

    • Set up ~/.cdsapirc with your API key

    • Run: snakemake -s Snakefile_cutouts --cores 1

  3. Quick fix - Use an available year:

    renewables_year: 2019  # Year with available cutout

  4. Check what’s available:

    ls resources/atlite/cutouts/
# Should show: uk-2019.nc, uk-2020.nc, etc.

Note: Zenodo cutouts are automatically verified with MD5 checksums.

Logging Issues#

Empty Log Files#

Symptom: Log files exist but are empty

Cause: Scripts not using snakemake.log[0]

Solution: This was fixed in v0.1. If still occurring:

# In script's main block
log_path = snakemake.log[0] if snakemake.log else "fallback"
logger = setup_logging(log_path)

“Logger Not Defined”#

Symptom: NameError: name 'logger' is not defined

Solution:

from scripts.logging_config import setup_logging
logger = setup_logging("my_script")

Getting Help#

Before Asking#

  1. Check this troubleshooting guide

  2. Search existing GitHub Issues

  3. Check the relevant log file

Information to Include#

When reporting an issue:

  1. Environment:

    conda list | grep -E "pypsa|snakemake|pandas"
python --version

  2. Command run:

    snakemake target.nc -j 4

  3. Full error message (copy from terminal)

  4. Log file contents (from logs/)

  5. Configuration snippet (relevant scenario)

Quick Diagnostic#

Run this to collect system info:

import sys
import pypsa
import pandas
import snakemake

print(f"Python: {sys.version}")
print(f"PyPSA: {pypsa.__version__}")
print(f"Pandas: {pandas.__version__}")
print(f"Snakemake: {snakemake.__version__}")