Skip to content

Hosting capacity

How much DER a bus can take before a planning criterion binds — or, the mirror case, how much new load (EV charging, electrification). The Hosting tab runs it; the Hosting overlay colours every bus by the answer; the result exports as GeoJSON or CSV for a GIS or a public hosting-capacity map, and the study report prints it.

What is evaluated

A trial unit — a generator at unity power factor for DER, a constant-power load at 0.95 for the load case — is placed at each candidate bus and grown until one of the criteria below fails. Candidates are every three-phase bus above 1 kV by default; both choices are options.

Criterion Test Default limit
Voltage DER: any bus above the upper limit. Load: any bus below the lower limit The project's abnormal-condition limits (1.05 / 0.95 pu), or a tighter value typed in the tab
Thermal Any line, transformer or device above the active load level's rating (normal or emergency) 100 %
Voltage change on trip The unit is switched off and the circuit re-solved; any bus moves more than the limit (IEEE 1547-2018 §4.7 rapid voltage change) 3 %
Reverse power at a regulator Real power flows backwards through a regulator more than 10 kW
Reverse power at the substation The feeder exports at its head more than 0 kW
Fault-current increase The unit's fault contribution against the base fault current at its bus (the FERC SGIP screen) 10 %
Interrupting duty Base fault current at a protective device plus the unit's contribution, against the device's interrupting rating (the SGIP screen) 87.5 %
Islanding ratio DER against the rated load of its protection section (the branches below the nearest fuse, recloser or relay) at the active level's multiplier 2/3

Each is a tick box. The fault criteria are analytic: a base fault study (mode=faultstudy) gives the three-phase fault current at every bus once, and the unit contributes 1.2 × its rated current (an inverter) or 5 × (a synchronous machine, xd″ ≈ 0.2). That is how utilities compute the screens, and it spares a fault solve per trial. Protection reach (the device's view of a downstream fault shrinking as DER feeds it) is not evaluated yet; it needs the sliding fault planned for the protection milestone.

Pre-existing violations do not count. A bus already over the voltage limit, a line already over its rating, a regulator already carrying reverse flow, a device already past its interrupting duty in the base case is excluded from that criterion and listed under the result, so a congested feeder still gets an answer about the rest.

Methods

Iterative. At each bus the unit is grown by doubling until a criterion fails, then bisected between the last feasible and the first infeasible size to within the tolerance (10 kW). Every evaluation is one solve, two when the trip criterion is on. Exact for the criteria it checks; the reference the other method is held to.

Sensitivity. One probe solve per bus (the unit at 100 kW) gives dV/dP at every bus and the change in loading of every branch; the capacity each criterion allows follows by linear extrapolation — for a branch whose loading first falls as the unit offsets the flow, the root on the far side where the reversed flow reaches the rating — and the smallest of them is the estimate. It is then confirmed by a solve at that size, and tightened by a few bisection steps if the linear guess was optimistic (regulators and voltage-dependent loads bend the line). Three or four solves per bus instead of ten to twenty: the shape of EPRI's DRIVE method, and what makes a feeder-wide or hourly study affordable. The per-criterion estimates are shown in the table, which answers "what binds second".

Time-granular. With loadshapes assigned, a list of hours (2, 12, 18) runs the study at each hour of the daily shapes; the capacity of a bus is its worst hour, and the per-hour values are kept, which is the form a published hosting-capacity map takes.

The run is at the active load level (minimum daytime load is where DER hosting capacity usually binds) and on the open scenario, and says so.

Outputs

  • A table, bus by bus: capacity, the binding criterion, the per-criterion estimates (sensitivity method), the per-hour values, and the load of the bus's protection section.
  • The Hosting overlay: every bus coloured red (nothing) through amber to green (the largest capacity in the run), with a legend; on the map too.
  • GeoJSON (points at each bus's coordinates, with capacity, binding criterion and per-criterion values as properties, in the project's coordinate system) and CSV.
  • The report's Hosting capacity section.

Interconnection screening

Analysis → Screen a DER interconnection… (or Screen a DER here… on an element's right-click menu) runs one application at a point of common coupling through the tariff screens in one go and prints the standard table, with the outcome pass, supplemental review or fail:

Screen Test Default
Aggregate generation on the line section DER already on the section plus this unit, against a share of the section's peak load (the fast-track screen) 15 %
Minimum daytime load (supplemental) When the aggregate screen fails and the unit exports: aggregate against the section's minimum load, taken from the project's level named Minimum… (else 30 % of peak) 100 %
Fault-current contribution The unit's contribution against the maximum fault current at the PCC 10 %
Interrupting duty Every protective device with the unit's contribution added, against its interrupting rating 87.5 %
Rapid voltage change The unit tripped, at every bus 3 %
Steady-state voltage With the unit running, inside the project's limits 0.95–1.05 pu
Loading With the unit running, nothing newly over its rating 100 %
Effective grounding at the PCC X0/X1 ≤ 3 and R0/X1 ≤ 1 (IEEE 142), from the Thevenin impedances
Smart inverter settings The IEEE 1547-2018 Category A or B characteristics the unit must run; reported, not judged Category B

The screens run at the active load level and on the open scenario, as every study does. The report's Interconnection screening section prints the table. POST /api/screen is the route; POST /api/faultpoint gives one fault of any kind (three-phase, line-to-ground, line-to-line, double-line-to-ground, through resistance) at one bus with the current in every series element, and POST /api/jobs/fault-sweep the maximum and minimum fault currents at every bus — the pieces the protection work of M16 builds on.

What has been checked

tests/test_hosting_capacity.py holds the method to a feeder built for it: the iterative result is confirmed by placing a generator of that size and re-solving (every criterion inside its limit; a third more and one is outside), the sensitivity method agrees with the iterative one within 8 % with a fifth of the solves, reverse power at the head binds at the feeder's load, the islanding ratio at two thirds of the section's load, load hosting capacity on undervoltage or thermal, the worst hour of a time-granular run is taken, and pre-existing violations are excluded.

tests/test_screening.py holds the screens: a 100 kW unit passes, 500 kW on a 2 MW section needs supplemental review (15 % fails, 100 % of minimum load passes), 3 MW at the end of a weak feeder fails on voltage and rapid voltage change, existing DER counts toward the aggregate, a non-exporting unit gets no minimum-load screen, and the fault-element solves agree with the fault study within 5 % for the bolted cases.

Not yet done: a comparison with a published reference (EPRI's DRIVE results on the IEEE 123-bus feeder, NREL's hosting-capacity papers on the same feeders). The numbers in those reports depend on criteria choices this page documents, so the comparison is a matter of setting the same ones; it will be added to this page when run.