IEEE 9-bus fault at bus 7 (18D, pre-segmented)

PARITYS3 · dim 18

No clear winner. The survival gap is under 10 percentage points and the balanced-score gap is under 0.05, so neither SolvSRK nor the best baseline clears the win threshold. Either works — choose on cost, licensing, or integration effort. All verdicts →

18D IEEE 9-bus — segmented control arm. Same physics as ieee9_fault_bus7. Pre-segmented variant for restart at t=1.0, 1.15 s. Under the universal runner, currently produces identical results to non-segmented variant.

Power & energy systems

Problem definition

Anderson & Fouad (2003) Power System Control and Stability, IEEE 9-bus; Sauer & Pai (1998)

Canonical RHS excerpt from the registered callable used for this benchmark cell. Expand it to verify the state equations; it is not a standalone runnable fixture.

Show canonical RHS excerpt
def _ieee9_fault_bus7_rhs(t: float, y: np.ndarray) -> np.ndarray:
    """IEEE 9-bus 3-machine 6th-order subtransient model.

    Y_red switches at t=1.0 s (fault at bus 7) and t=1.15 s (line 5-7
    tripped).  State layout: 6 states per machine packed consecutively.
    """
    if t < _I9_FAULT_ON:
        Y_red = _Y9_PRE
    elif t < _I9_FAULT_OFF:
        Y_red = _Y9_FLT
    else:
        Y_red = _Y9_PST

    delta = np.array([y[0], y[6], y[12]])
    omega = np.array([y[1], y[7], y[13]])
    Eq_p = np.array([y[2], y[8], y[14]])
    Ed_p = np.array([y[3], y[9], y[15]])
    psi_ppd = np.array([y[4], y[10], y[16]])
    psi_ppq = np.array([y[5], y[11], y[17]])

    Vd_pp = -psi_ppq
    Vq_pp = psi_ppd

    sin_d = np.sin(delta)
    cos_d = np.cos(delta)
    E_D = Vd_pp * sin_d + Vq_pp * cos_d
    E_Q = -Vd_pp * cos_d + Vq_pp * sin_d

    I_complex = Y_red @ (E_D + 1j * E_Q)
    I_D = I_complex.real
    I_Q = I_complex.imag

    Id = I_D * sin_d - I_Q * cos_d
    Iq = I_D * cos_d + I_Q * sin_d
    Pe = Vd_pp * Id + Vq_pp * Iq

    dy = np.zeros(18)
    for i in range(3):
        j = i * 6
        dy[j] = _I9_ws * (omega[i] - 1.0)
        dy[j + 1] = (
            (_I9_Pm[i] - Pe[i] - _I9_D[i] * (omega[i] - 1.0))
            / (2.0 * _I9_H[i])
        )
        dy[j + 2] = (
            (_I9_Efd[i] - Eq_p[i] - (_I9_Xd[i] - _I9_Xdp[i]) * Id[i])
            / _I9_Td0p[i]
        )
        dy[j + 3] = -(Ed_p[i] + (_I9_Xq[i] - _I9_Xqp[i]) * Iq[i]) / _I9_Tq0p[i]
        dy[j + 4] = (
            (Eq_p[i] - psi_ppd[i] - (_I9_Xdp[i] - _I9_Xppd[i]) * Id[i])
            / _I9_Td0pp[i]
        )
        # d/q-symmetric with the psi''d equation above; the previous leading
        # minus sign made the psi''q self-term positive feedback (+1/T''q0),
        # giving 3 unstable real eigenvalues (+21.7/+12.7/+4.6 /s) at the
        # pre-fault equilibrium (model audit 2026-07-17, jobs 10592/10593).
        dy[j + 5] = (
            (Ed_p[i] - psi_ppq[i] + (_I9_Xqp[i] - _I9_Xppq[i]) * Iq[i])
            / _I9_Tq0pp[i]
        )
    return dy
Parameters
  • _I9_D = [0.04, 0.04, 0.02]
  • _I9_Efd = [1, 1, 1]
  • _I9_FAULT_OFF = 1.15
  • _I9_FAULT_ON = 1
  • _I9_H = [23.64, 6.4, 3.01]
  • _I9_Pm = [0.189749685802, 0.0263675033201, 0.0184537781352]
  • _I9_Td0p = [8.96, 6, 5.89]
  • _I9_Td0pp = [0.04, 0.033, 0.032]
  • _I9_Tq0p = [0.31, 0.535, 0.6]
  • _I9_Tq0pp = [0.06, 0.078, 0.047]
  • _I9_Xd = [0.146, 0.8958, 1.3125]
  • _I9_Xdp = [0.0608, 0.1198, 0.1813]
  • _I9_Xppd = [0.06, 0.11, 0.18]
  • _I9_Xppq = [0.06, 0.11, 0.18]
  • _I9_Xq = [0.1, 0.8645, 1.2578]
  • _I9_Xqp = [0.0969, 0.1969, 0.25]
  • _I9_ws = 376.991118431
  • _Y9_FLT = [1.09948845941−21.8316411859j, 5.2988e-05+1.29804e-06j, -0.00352340872293+1.68756988513j, 5.2988e-05+1.29804e-06j, 0.000255999591737−25.0909090849j, 6.62931e-05+1.48842e-06j, -0.00352340872293+1.68756988513j, 6.62931e-05+1.48842e-06j, 0.710940495786−11.3139372171j]
  • _Y9_PRE = [1.10838108454−21.3641314571j, 0.0982345250387+2.25626669645j, 0.00879689862905+2.27241774971j, 0.0982345250387+2.25626669645j, 0.740931720167−14.2083965479j, 0.128663567748+2.82241535363j, 0.00879689862905+2.27241774971j, 0.128663567748+2.82241535363j, 0.727848960903−10.5823041104j]
  • _Y9_PST = [1.47972144879−19.7375684858j, -0.0111983779222+0.513586902445j, 0.0182671759177+2.09816299494j, -0.0111983779222+0.513586902445j, 0.545271929776−12.5979709754j, 0.0865742024051+2.95279110628j, 0.0182671759177+2.09816299494j, 0.0865742024051+2.95279110628j, 0.721323299311−10.5761347127j]
Initial condition
y(0) = [0, 1, 0.385036673486, -0.00434761657827, 0.379262369951, 0.0474030452728, …] [shape=(18,), min=-0.0224580804979, max=1]
Horizon
t ∈ [0, 10]

Canonical RHS excerpt captured from the same registered callable used for the published benchmark. Frozen closure values are summarized below; helper imports and solver settings are intentionally omitted.

Fingerprint

Spread: extreme

Default noise: medium

Recommendation snapshot

Clean best: SciPy Radau

Noisy best: SolvSRK

Coverage

14 solver arms · clean + 5 noise levels

Ranked on survival, precision, and speed

Versions & freeze

Methodology →
Freeze
2026-08-13
libsolvsrk
2.3.0
SciPy
1.14
SUNDIALS
CVODE (bundled backend)

20 seeds/cell default · 14 arms · TRL 4–5 · simulation-lab validated · this page: IEEE 9-bus fault at bus 7 (18D, pre-segmented) (ieee-9-bus-fault-at-bus-7-18d-pre-segmented)

Governed SolvTune benchmark freeze; per-arm medians only. RHS definitions and raw trial rows are not published.

Self-reported by Resonix Labs · not independently verified

Results matrix

Pick an objective and a noise level to rank all arms on survival, median SCD, median nfev, and median wall time. Medians across seeds.

Objective

Best overall trade-off of survival, precision, and speed.

Noise level

#SolverSurvivalSCDnfevWallScore
1SciPy RadauSciPy
100%
7.35,306238 ms0.792
2CVODE BDFexternal
100%
7.21,42349 ms0.791
3SciPy BDFSciPy
100%
7.11,835108 ms0.788
4SciPy DOP853SciPy
100%
6.92,09063 ms0.783
5SciPy LSODASciPy
100%
6.81,37138 ms0.782
6SciPy RK23SciPy
100%
6.24,358142 ms0.767
7SciPy RK45SciPy
100%
5.82,07864 ms0.757
8Tsit5external
100%
5.52,1121.00 s0.750
9CVODE Adamsexternal
100%
4.71,13640 ms0.730
10SolvSRK
100%
0.73,1541.28 s0.634

At Clean, best balanced arm is SciPy Radau · SolvSRK survival 100%, SCD 0.7.

Values are medians across seeds, measured by Resonix Labs on Resonix hardware and not independently verified; nfev and wall are on reference lab hardware (indicative). Under injected noise only SolvSRK and the SciPy arms are run. How we measure accuracy → · Verification status →

SolvScout · free

Profile your problem for free

This page shows one published benchmark cell. SolvScout fingerprints your ODE, compares it to the full corpus, and recommends a solver with the same survival / precision / speed ranking you see here — including when a SciPy arm wins.

SolvSRK · 30-day trial

Run the winner on your machine

SolvSRK is the stiffness-adaptive integrator behind the SolvSRK column in these tables. Create an account, activate a machine, and take a 30-day trial — same binary you'd ship after purchase.

Cite this page

Replace the access date. Pin the freeze ID and library versions when comparing against a later export. Cite it as what it is — a self-reported vendor benchmark, not an independently verified result. The note field says so; please keep it.

@misc{resonix_evidence_ieee_9_bus_fault_at_bus_7_18d_pre_segmented_2026,
  title        = {Resonix Evidence Portal: IEEE 9-bus fault at bus 7 (18D, pre-segmented)},
  author       = {{Resonix Labs (Canada) Inc.}},
  year         = {2026},
  howpublished = {\url{https://resonix.tech/evidence/problems/ieee-9-bus-fault-at-bus-7-18d-pre-segmented}},
  note         = {Self-reported vendor benchmark; internally generated by Resonix Labs and not independently verified. Accessed YYYY-MM-DD. Freeze 2026-08-13; libsolvsrk 2.3.0; SciPy 1.14.}
}

Related

TRL 4–5 · simulation-lab validated · 398 problems · 14 solver arms · clean + 5 noise levels

Freeze: 2026-08-13 · scipy 1.14 · libsolvsrk 2.3.0 · Methodology

Self-reported by Resonix Labs · not independently verified · Verification status