Facility siting

The canonical FRED workflow. Give FRED a target region and a facility size, get back a specific recommended location with supporting data.

#What it does

A siting query asks FRED to choose a specific location for a new facility within a target region. FRED considers cluster supply density, harvest cost, transport efficiency, and (optionally) fire risk in the surrounding terrain, then returns a single recommended coordinate.

Behind the scenes, FRED evaluates candidate centroids across the target region, computes the procurement economics for each, and picks the centroid that minimizes the weighted objective.

#How to ask

The reliable phrasings include:

  • "Where should I site a 20MW facility near Redding?"
  • "Find the best location for a 15MW biomass plant in Plumas County."
  • "Recommend a facility location with 30,000 BDT/yr capacity around Lake Almanor."
  • "What's the optimal site for a 25MW plant near Burney, prioritizing fire risk?"

FRED accepts capacity in MW or biomass demand in BDT/year. The target region can be a town, county, watershed, or coordinate point with an implicit radius.

#What comes back

A typical siting answer includes:

  • The recommended site: latitude/longitude, the nearest named place, and the county.
  • Annual supply: BDT/year available to that site under the operational radius.
  • LCOE: per-BDT levelized cost at the recommended site, broken down into harvest cost and transport.
  • Operational radius: the actual procurement radius needed to meet demand. This is often smaller than people expect for cluster-rich regions and larger for sparse ones.
  • A map: the recommended site marked, contributing clusters color-coded by contribution, with the procurement boundary overlaid.
  • A justification: why this site won out over nearby alternatives. Usually mentions terrain, transport efficiency, or supply density.

#How FRED decides

FRED's siting algorithm is straightforward in concept: evaluate a grid of candidate centroids across the target region, run the procurement model for each, pick the best. The implementation details that matter for understanding the output:

  • The candidate grid resolution adapts to the target region size. For a county-scale search, FRED evaluates roughly 100-200 candidates. For a town-scale search, it evaluates a finer grid.
  • Each candidate's "score" combines weighted LCOE (after the alpha weighting if fire risk is enabled) with a sustainability check: can this site sustain operations for at least 10 years without depleting available clusters?
  • The expanding-ring radius search is rerun for each candidate. FRED does not assume a fixed radius across candidates; the radius itself is part of the optimization.
NOTE
FRED's recommended coordinate is a centroid suggestion, not a precise parcel address. Real facility placement involves access roads, water rights, zoning, and grid interconnect that FRED does not model. Treat the recommendation as the center of a 1-2km search area for site-specific evaluation.

#Common follow-ups

Useful follow-up questions after a siting query:

  • "What were the top three alternative sites and why did they lose?"
  • "Show me the cost surface across the search region."
  • "What does this look like if I move the facility 5km north?"
  • "How does the recommendation change at 25MW instead of 20MW?"
  • "Project this over 10 years."