foxes_opt.constraints¶

Constraints for wind farm optimization problems.

Guarded Minimum-Distance Pairs¶

MinDistConstraint.active_components selects pairs that can approach the nominal spacing limit anywhere within enforced X/Y coordinate bounds. For min_dist_unit="D", the required distance uses the larger diameter in each pair. Fixed turbines keep their coordinates. Unbounded coordinates and variable diameters are conservative; state-dependent coordinates and unsupported variable parameterizations retain all pairs.

The caller of active_components must enforce the supplied coordinate bounds over the entire solve before using its selection. Without useful movement bounds, a safe selection may retain all pairs. Current iwopy SLSQP evaluates all registered spacing components and does not invoke this selection hook.

iwopy Problem.get_jacobian returns a SciPy CSR matrix whose rows are selected function components and whose columns are selected floating variables, both in requested order. Analytical spacing and geometry derivatives feed this sparse assembly directly. SLSQP converts the requested rows to dense native arrays; SciPy trust-constr and PyGMO IPOPT retain sparse constraint derivatives at their native boundary.

Nearest connected groups¶

NearestGroupConstraint combines an upper nearest-turbine distance with a minimum connected-group size. For each selected turbine, it constrains the smallest distance threshold at which the turbine’s graph component contains max(2, min_nearest_group_size) turbines. This permits chains of nearby turbines while rejecting isolated turbines and undersized components.

The connection radius is piecewise equal to one active edge length, so the constraint provides analytical derivatives for direct turbine X/Y variables. At equal edge lengths, the deterministic active branch supplies the derivative. Other layout parameterizations remain marked as dependent and can be handled by LocalFD.

Configure it through LayoutOptimizerStage.constraints. A custom constraint list replaces the default boundary entry, so include that entry explicitly:

constraints = [
    {"constraint_type": "FarmBoundaryConstraint"},
    {
        "constraint_type": "NearestGroupConstraint",
        "max_nearest_dist": 1200.0,
        "min_nearest_group_size": 10,
    },
]

Constraint families own their tolerances independently. For example, combine a boundary tolerance with separate automatic minimum-distance parameters:

stage = LayoutOptimizerStage(
    optimizer_family="iwopy",
    optimizer_pars={"optimizer": "SLSQP"},
    constraints=[
        {"constraint_type": "FarmBoundaryConstraint", "tol": 0.5},
    ],
    min_dist_constraint_pars={"tol": 0.01},
)

foxes_opt.constraints