Background and scope

Why this specification exists, what it does and does not cover, and the design choices behind it. This page frames the rest of the specification; it introduces no model content. Where the scope differs from the older Task Force PDF, the difference reflects what BMOPFTools actually implements.

Motivation

Distribution networks are unbalanced: single-phase loads, untransposed lines, and single-phase laterals mean the phases cannot be collapsed to a positive-sequence equivalent without losing the physics. A faithful optimal power flow (OPF) for distribution must therefore reason at the level of individual conductors.

At the same time, the range of utility problems posed as network-constrained optimisation has grown — power flow, state estimation, volt-var control, DER scheduling, dynamic operating envelopes, optimal droop settings. Yet there is little standardisation: few openly-licensed unbalanced network models exist, so papers rely on ad-hoc modified cases and results are hard to compare. This specification, with its companion data library, provides a common, openly-licensed model so approaches can be compared directly — the distribution analogue of transmission-side benchmark libraries.

Scope: beyond classical OPF

Although "OPF" names this document, the goal is broader than minimising generation cost. The unifying requirement of the target problems is an accurate conductor-level representation of an unbalanced network subject to a selectable set of bounds — not any one objective. Bounded cost minimisation is a convenient, solver-comparable starting point. With fixed demand and one uniform non-negative price on every real-power injection, minimizing total injection is equivalent to minimizing real losses. It does not make the nonconvex feasible set or optimum unique. The same physics underpins maximum load delivery, conservation voltage reduction, dynamic operating envelopes, and state estimation.

This is why bounds are optional throughout the data model: different formulations activate different subsets of the feasible region. The specification is a reusable foundation, not infrastructure for a single problem.

Design choices

• SI units. All physical quantities in SI (see Data input formatting), so the data model commits to no per-unit base. Per-unit is a solver-internal convenience, out of scope here.
• JSON + schema. Parseable from any language; key–value structure lets extensions add nested entries without breaking readers. A JSON Schema provides basic structural checks.
• Real numbers, not complex. Every complex quantity is a pair of real fields, and the model solves in real variables — for cross-language compatibility (see Notation).
• Explicit buses. Buses are a first-class list with their own terminals and bounds — unlike OpenDSS, where buses are implicit in element connectivity.
• String identifiers. Buses, lines, loads, etc. carry unique string IDs, not forced sequential integers (unlike MATPOWER).
• Wire coordinates. Quantities are per-conductor ("wire"), not sequence/symmetrical components, keeping the format flexible and extendable to any wire count.

What is modelled

The specification covers the common branch and nodal elements, with these capabilities:

• Topology: meshed networks, electrically parallel branches, radial or looped.
• Conductors: 1- to 4-wire lines with full mutual coupling; explicit neutral and earth (no Kron reduction); perfect grounding and grounding through impedance.
• Branch elements: lines, switches, galvanically-isolated transformers (single-phase, centre-tap, wye–delta, delta–wye, n-winding), and non-isolated regulators (single-phase autotransformer, open-delta).
• Nodal elements: loads (constant-power and voltage-dependent ZIP/exponential), generators, shunts, capacitors, voltage sources, and IBRs with smart-inverter control profiles.
• DC: an MVDC/LVDC subsystem with AC/DC converters, enabling converter stations, back-to-back SOPs, and MVDC ties.
• Impedance: direct linecodes/matrices or geometry-derived via Carson/Deri.
• Time variation: time-series scaling of parameters (snapshot-based).

Present limitations

The model deliberately targets features universally required for distribution OPF, not every possible device. In this version:

• A single voltage source (one reference bus).
• Snapshot solves — time series scale parameters per instant with no inter-temporal coupling (no storage state-of-charge dynamics, no OLTC time-domain control).
• Transformer saturation and detailed frequency-dependent effects are not modelled (a no-load magnetising shunt is supported).
• The default objective is linear generation/dispatch cost; quadratic cost terms are not included.

Out of scope

The format is a practical, lightweight model for OPF research — it does not aim to replace the Common Information Model (CIM), and it does not prescribe solver software. Its OPF formulation was inspired by PowerModelsDistribution's IVRENPowerModel, but has been generalized well beyond it — native JSON data, the full set of element configurations above, and careful grounding.