Battery Thermal Runaway 4stage

ADVANTAGES2 · dim 8

SolvSRK wins. At the comparison noise level, SolvSRK beats the best baseline by at least 10 percentage points of survival, or by at least 0.05 balanced score when survival is tied. Use SolvSRK for this class of problem. All verdicts →

Single Li-ion cell with 4-stage Arrhenius thermal runaway (Hatchard-Dahn model): SEI, anode-electrolyte, cathode, and electrolyte decomposition with gas pressure and resistance tracking.

Batteries & energy storage

Problem definition

Canonical benchmark implementation

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 _arrhenius_rate(A, E, T):
    """Safe Arrhenius rate: A * exp(-E/(R*T)) with clamped exponent."""
    arg = np.clip(E / (_R_GAS * T), 0.0, _EXP_CLAMP)
    return A * np.exp(-arg)

def _four_stage_decomposition(alpha_sei, alpha_ae, alpha_ca, alpha_el, T):
    """Compute the four decomposition rates and total volumetric heat.

    Returns (d_sei, d_ae, d_ca, d_el, q_dot) where q_dot is W/kg (mass-specific).
    """
    k_sei = _arrhenius_rate(_A_SEI, _E_SEI, T)
    k_ae = _arrhenius_rate(_A_AE, _E_AE, T)
    k_ca = _arrhenius_rate(_A_CA, _E_CA, T)
    k_el = _arrhenius_rate(_A_EL, _E_EL, T)

    # SEI: consumed (α decreases)
    d_sei = -k_sei * alpha_sei

    # Anode-electrolyte: consumed fraction increases
    d_ae = k_ae * alpha_ae * (1.0 - alpha_ae)
    # At α_ae=0 this would give zero rate, but the seed is the
    # SEI decomposition products — use a small baseline nucleation
    # once SEI has started decomposing.
    if alpha_ae < 1e-12 and alpha_sei < 0.15 - 1e-6:
        d_ae = k_ae * 1e-6

    # Cathode: decomposed fraction increases
    d_ca = k_ca * (1.0 - alpha_ca)

    # Electrolyte: decomposed fraction increases
    d_el = k_el * (1.0 - alpha_el)

    q_dot = (_Q_SEI * _W_SEI * abs(d_sei)
             + _Q_AE * _W_AE * d_ae
             + _Q_CA * _W_CA * d_ca
             + _Q_EL * _W_EL * d_el)

    return d_sei, d_ae, d_ca, d_el, q_dot

def _btr_4stage_rhs(t, y):
    alpha_sei = np.clip(y[0], 0.0, 1.0)
    alpha_ae  = np.clip(y[1], 0.0, 1.0)
    alpha_ca  = np.clip(y[2], 0.0, 1.0)
    alpha_el  = np.clip(y[3], 0.0, 1.0)
    T         = np.clip(y[4], 250.0, 2000.0)
    Q_total   = y[5]
    P_gas     = y[6]
    R_int     = y[7]

    d_sei, d_ae, d_ca, d_el, q_dot = _four_stage_decomposition(
        alpha_sei, alpha_ae, alpha_ca, alpha_el, T,
    )

    # Temperature
    q_gen = q_dot * _M_CELL
    q_cool = _H_CONV * _A_SURF * (T - _T_AMB)
    dT = (q_gen - q_cool) / (_M_CELL * _CP)

    # Cumulative heat
    dQ = q_gen

    # Gas pressure: gas moles proportional to electrolyte decomposition
    # n_gas ≈ α_el * n_gas_max;  dn/dt = n_gas_max * d_el
    n_gas_max = 0.01  # mol of gas at full electrolyte decomposition
    n_gas = alpha_el * n_gas_max
    dn_gas = n_gas_max * d_el
    dP = (dn_gas * _R_GAS * T + n_gas * _R_GAS * dT) / _V_HEAD

    # Internal resistance growth
    dR = _R0 * (0.5 * abs(d_sei) + 2.0 * d_ae)

    dy = np.empty(8)
    dy[0] = d_sei
    dy[1] = d_ae
    dy[2] = d_ca
    dy[3] = d_el
    dy[4] = dT
    dy[5] = dQ
    dy[6] = dP
    dy[7] = dR
    return dy
Parameters
  • _A_AE = 2.5e+13
  • _A_CA = 6.667e+13
  • _A_EL = 5.14e+25
  • _A_SEI = 1.667e+15
  • _A_SURF = 0.000818
  • _CP = 830
  • _EXP_CLAMP = 80
  • _E_AE = 135080
  • _E_CA = 139600
  • _E_EL = 274000
  • _E_SEI = 135080
  • _H_CONV = 10
  • _M_CELL = 0.044
  • _Q_AE = 1.714e+06
  • _Q_CA = 314000
  • _Q_EL = 155000
  • _Q_SEI = 257000
  • _R0 = 0.02
  • _R_GAS = 8.314
  • _T_AMB = 298
  • _V_HEAD = 1e-06
  • _W_AE = 0.5
  • _W_CA = 0.25
  • _W_EL = 0.217
  • _W_SEI = 0.033
Initial condition
y(0) = [0.15, 0, 0, 0, 420, 0, 101325, 0.02]
Horizon
t ∈ [0, 600]

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: high

Recommendation snapshot

Clean best: SolvSRK

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: Battery Thermal Runaway 4stage (battery-thermal-runaway-4stage)

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
1SolvSRK
100%
14.8245,1856.44 s0.972
2SciPy RadauSciPy
100%
12.368632 ms0.911
3SciPy DOP853SciPy
100%
11.12187 ms0.883
4SciPy LSODASciPy
100%
10.43189 ms0.867
5SciPy BDFSciPy
100%
10.040623 ms0.857
6Tsit5external
100%
9.9264756 ms0.854
7SciPy RK45SciPy
100%
9.82729 ms0.853
8SciPy RK23SciPy
100%
9.879427 ms0.851
9CVODE BDFexternal
100%
9.717112 ms0.850
10CVODE Adamsexternal
100%
9.212311 ms0.838

At Clean, best balanced arm is SolvSRK.

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_battery_thermal_runaway_4stage_2026,
  title        = {Resonix Evidence Portal: Battery Thermal Runaway 4stage},
  author       = {{Resonix Labs (Canada) Inc.}},
  year         = {2026},
  howpublished = {\url{https://resonix.tech/evidence/problems/battery-thermal-runaway-4stage}},
  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