What industries does newtsim serve?

newtsim serves oil and gas (flow assurance, drilling fluids, subsea pipeline), pharmaceutical and drug discovery, energy storage (battery materials, electrolytes), semiconductor process simulation, green hydrogen (electrolyser and catalyst design), nuclear decommissioning, aerospace and composites, corrosion and materials, carbon capture, agrochemicals, mining and geotechnical, and fermentation and bioprocessing.

How is newtsim different from standard CFD or simulation software?

Standard simulation tools use empirical models at a single scale. newtsim runs the full physics chain from quantum mechanics and molecular dynamics through to plant-scale CFD and structural mechanics, GPU-native, with direct data flow between every scale and no third-party solver dependencies. Constitutive behaviour is predicted from first principles rather than empirically fitted.

What does a fixed-price simulation study include?

A newtsim fixed-price study has a defined scope, clear deliverables, fixed cost agreed upfront, and is delivered in weeks rather than quarters. Studies typically include a scoping call, physics-grounded simulation using the appropriate newtsim solvers, results analysis, and a written report with actionable recommendations.

What is physics-grounded simulation?

Physics-grounded simulation derives material properties and constitutive behaviour from lower-scale physics (molecular dynamics and quantum mechanics) rather than from empirical fitting. This gives accurate predictions for novel materials and extreme conditions outside the range of experimental data, and enables root-cause analysis at the mechanistic level.

How does newtsim approach oil and gas flow assurance simulation?

newtsim runs a continuous physics chain from molecular wax crystallisation to full pipeline transient multiphase flow, with no empirical handoffs between scales. Wax appearance temperatures, gel strength, and deposition kinetics are predicted from molecular dynamics (newtsim Bond), passed directly to mesoscale morphology (newtsim Pack), and then propagated into multiphase CFD (newtsim Stream), giving flow assurance predictions that hold outside the range of laboratory data.

Can newtsim simulate battery electrolyte behaviour and degradation?

Yes. The Bond solver runs molecular dynamics for Li-ion transport and solid-electrolyte interphase (SEI) formation; Pack handles electrode morphology at the mesoscale; these results couple upward to cell-scale electrochemical models. This gives physics-grounded predictions of electrolyte degradation, capacity fade, and thermal behaviour without relying solely on empirical ageing fits.

How does physics-grounded drug discovery simulation differ from docking software?

Molecular docking software treats the protein and ligand as rigid bodies and scores binding poses by geometry. Physics-grounded simulation uses molecular dynamics to capture conformational flexibility and explicit solvation effects that rigid-body docking misses, predicting binding free energy, ADMET properties, and metabolic activation from first principles. This is particularly valuable for targets where induced-fit binding or solvent-mediated interactions drive selectivity.

What makes GPU-native simulation faster than CPU-based CFD?

GPU parallelism maps naturally to the computational kernels underlying molecular dynamics and CFD: large numbers of independent force or flux calculations that can execute simultaneously. newtsim runs all simulation scales GPU-native, avoiding the CPU-to-GPU data transfers that bottleneck traditional solvers which run physics on CPU and only offload selected operations to the GPU. This removes the transfer overhead and allows the full multiscale chain to run without the latency penalties of heterogeneous pipelines.

Does newtsim replace existing simulation tools like ANSYS or COMSOL?

newtsim fills the multiscale gap, connecting molecular-scale physics to the engineering-scale models that tools like ANSYS operate at. It is not a replacement for FEA-only or single-scale workflows where those tools are the right fit. Where newtsim adds value is in problems where constitutive behaviour, material properties, or interfacial physics need to be derived from lower-scale simulation rather than taken from a handbook or empirical fit.

How long does a fixed-price simulation study take?

Most studies complete in 4 to 8 weeks depending on scope. The standard timeline is: scoping call and problem definition in week 1; simulation execution across the relevant physics scales in weeks 2 to 4; results analysis, validation, and report delivery in weeks 5 to 6. More complex multiscale studies with broader operating envelopes may extend to 8 weeks. Scope and timeline are agreed and fixed before work begins.

What deliverables does a simulation study produce?

A newtsim simulation study delivers: a validated multiscale model calibrated to your system; an operating envelope with physics-based safety margins; a written technical report with findings and actionable recommendations; and optionally CFD visualisations or an ML surrogate model (newtsim Neural) for fast design-space exploration by your engineering team.

Does newtsim work for green hydrogen electrolyser and catalyst design?

Yes. The Stream solver handles proton exchange membrane and alkaline electrolyser CFD: multiphase flow, bubble dynamics, and membrane transport. Bond handles catalyst surface reaction molecular dynamics, giving adsorption energies and reaction barriers from first principles rather than from empirical correlations. Neural then produces fast-to-run surrogate models for design-space exploration and stack optimisation, reducing the iteration cost for new catalyst compositions and electrode architectures.

newtsim is a physics-grounded multiscale simulation consulting firm and software platform. We deliver fixed-price simulation studies and platform licences for industrial clients, covering the full physics chain from quantum mechanics and molecular dynamics through to plant-scale CFD and structural mechanics. All solvers run GPU-native.

Does newtsim use artificial intelligence or machine learning?

Yes. newtsim Neural produces ML-trained surrogate models derived from high-fidelity physics simulations. These surrogates run in milliseconds rather than hours, enabling rapid design-space exploration and real-time control without sacrificing physical accuracy. The ML models are trained on physics simulation data, not empirical measurements, so they inherit the accuracy and extrapolation properties of the underlying physics.

How do I get started with newtsim?

Book a free technical scoping call at newtsim.com/demo. Bring a description of your engineering challenge, the physics you think are important, and any data you have. The call is 45 minutes and produces a problem definition, proposed physics chain, and outline scope that forms the basis of a fixed-price proposal.

Can newtsim simulate non-Newtonian fluids?

Yes. newtsim specialises in non-Newtonian flow: yield-stress fluids, viscoelastic materials, shear-thinning and shear-thickening systems, and complex multiphase flows. The Stream solver handles macroscale CFD for these systems; Bond provides the molecular-scale rheology predictions that replace empirical viscosity fits; and the full chain gives accurate behaviour outside the calibrated range of laboratory data.

Does newtsim work for small companies or startups?

Yes. Fixed-price studies are available to clients of any size. The fixed-price, fixed-scope model means you know the cost before committing, making it accessible for smaller organisations without large simulation teams. Many newtsim clients are startups and scale-ups who need physics-grounded analysis early in product development before investing in laboratory or pilot-scale work.

What is the difference between molecular dynamics and CFD?

Molecular dynamics (MD) simulates the motion of individual atoms and molecules to compute material properties such as viscosity, diffusivity, reaction rates, and phase behaviour, from first principles. Computational fluid dynamics (CFD) models bulk fluid flow at the engineering scale using those material properties as inputs. newtsim connects the two: MD provides accurate constitutive inputs that CFD uses for engineering-scale predictions, rather than relying on empirical correlations.

What is a surrogate model in simulation?

A surrogate model (also called a reduced-order model or emulator) is a fast mathematical approximation of a complex simulation. newtsim Neural builds ML surrogates trained on high-fidelity physics simulation data. Once trained, a surrogate evaluates in milliseconds instead of hours, enabling real-time optimisation and design-space exploration. Because surrogates are trained on physics data, they are more accurate than empirical correlations, especially outside the training range.

What deliverables does a newtsim simulation study produce?

A newtsim simulation study delivers: (1) a validated multiscale model calibrated to your system; (2) an operating envelope with physics-based safety margins; (3) a written technical report with findings and actionable recommendations; and optionally (4) CFD visualisations and an ML surrogate model for fast ongoing design-space exploration by your engineering team.

Energy Storage

Energy Storage · Case Study

Post-Incident Thermal Management Audit for Grid-Scale Battery Energy Storage Systems: A Study Following the Arizona McMicken Fire

Executive Summary

On 19 April 2019, at approximately 4:57 PM, a voltage drop was detected in one battery module of the McMicken Battery Energy Storage System — a 2 MW / 2 MWh facility at the APS McMicken substation in Surprise, Arizona, operated by Fluence under a long-term service agreement with Arizona Public Service. The system used LG Chem NMC cells integrated in 20-foot ISO containers with FM-200 clean agent suppression. Thermal runaway developed in the initial cell and propagated module-to-module over approximately 90 minutes. Flammable gases — primarily hydrogen and carbon monoxide — accumulated within the sealed ISO container to concentrations at or above the lower flammable limit. The battery management system detected elevated temperature but did not isolate the module or activate the FM-200 suppression system, because the temperature rise fell below the suppression activation threshold and no hydrogen concentration sensors were installed in the container. When first responders opened the container side door at approximately 8:00 PM for visual inspection, the accumulated flammable gases deflagrated within 2–3 minutes. The resulting pressure wave ejected both the side door and the rear door, injuring four firefighters with burns and blast injuries. The 2 MW / 2 MWh system was a total loss.

The 127-page DNV GL investigation report, commissioned by APS and published in January 2020, identified the most likely initiating cause as abnormal lithium metal deposits in the initial cell — consistent with accelerated lithium plating over approximately two years of operation at high SOC in Phoenix ambient temperatures. Three system-level failures compounded the outcome: the lack of thermal barriers between battery cells allowed rapid module-to-module propagation; the FM-200 suppression system, which activates on temperature rather than gas concentration, never triggered during the 90-minute runaway event; and the absence of hydrogen sensors inside the container meant that flammable gas accumulation went entirely undetected before first responders opened the door. DNV GL's finding is unambiguous: H₂ sensors inside the container would have detected flammable gas accumulation approximately 45 minutes before first responder arrival — preventing the deflagration entirely.

Had a coupled electrochemical-thermal simulation and gas accumulation model been applied at the deployment stage for this NMC532 system — or as a post-deployment audit before the critical two-year mark — it would have identified the failure pathway with precision. A 3D CFD propagation model of the 16S1P module geometry at nominal and degraded cooling flow rates predicts module-to-module propagation in 18 ± 4 minutes under nominal cooling, compressing to 9 ± 2 minutes at 40% flow degradation. A well-mixed gas accumulation model of the 33.2 m³ ISO container shows that three module propagation events — each venting approximately 397 L of H₂ — bring the container headspace to the 4% lower flammable limit within 12 ± 3 minutes at degraded cooling, before any FM-200 activation is possible under the current BMS alarm logic. The simulation would have flagged both the inadequacy of temperature-only suppression logic and the absence of H₂ sensors as critical safety gaps before any firefighter was placed at risk.

Had this simulation been applied at the deployment stage — or as a post-deployment audit before the critical two-year mark — it would have identified both the inadequacy of temperature-only suppression logic and the absence of H₂ sensors as critical safety gaps before any firefighter was placed at risk. The McMicken incident resulted in four firefighter injuries, a 127-page regulatory investigation, and a total loss of the 2 MW / 2 MWh system — all preventable with H₂ gas sensors inside the container, a retrofit costing approximately $20,000 across an equivalent six-site Gen-1 fleet. A newtsim post-deployment audit produces a fleet-level risk register, a ranked retrofit plan, and an insurance underwriter technical report. The failure-prone operating parameters identified by the simulation — propagation time as a function of cooling flow degradation, H₂ accumulation rate per module, LFL crossing time relative to suppression logic delay — define the real-time sensor network for newtsim livesim fleet monitoring, enabling operators to detect thermal runaway precursors and degraded cooling conditions before they reach the deflagration threshold.

Client Background

Scenario Background (illustrative reference case)

SunBelt Grid Storage LLC is a subsidiary of a regional utility holding company, operating BESS facilities under 20-year capacity contracts with two Arizona utilities. All sites operate under NERC CIP-014 physical security requirements and follow NFPA 855 (Standard for the Installation of Stationary Energy Storage Systems, 2020 edition). Their fleet spans two technology generations with different thermal risk profiles:

Gen-1 Fleet (6 sites, 2017–2018 vintage): NMC532 chemistry cells integrated by Fluence (formerly AES Energy Storage / Siemens), the same integrator involved in the McMicken facility. These sites use the same module design, container type (20-foot ISO container), and suppression system (clean agent FM-200 / HFC-227ea) as McMicken. The chemical similarity and engineering provenance make the DNV GL McMicken findings directly applicable to the SunBelt Gen-1 fleet.

Gen-2 Fleet (6 sites, 2020–2021 vintage): LFP chemistry cells integrated by BYD, representing a fundamentally different thermal risk profile. LFP's phosphate cathode structure is thermally stable to approximately 270 °C vs. 200 °C for NMC532; exothermic energy release is approximately 3× lower per unit mass; and vented gas is CO₂-dominant (non-flammable) rather than H₂/CO-dominant (flammable). The Gen-2 sites require audit but are expected to be within acceptable risk bounds.

Fleet Summary:

Fleet	Chemistry	Sites	Capacity Each	Integrator	Vintage
Gen-1	NMC532 / graphite	6	4 MW / 8 MWh	Fluence	2017–2018
Gen-2	LFP / graphite	6	4 MW / 8 MWh	BYD	2020–2021
Total		12	48 MW / 96 MWh

Gen-1 Site Cell and System Specifications:

Parameter	Gen-1 (NMC532)	Gen-2 (LFP)
Cell chemistry	NMC532 / graphite	LFP / graphite
Cell format	100 Ah prismatic	280 Ah prismatic
Nominal cell voltage	3.65 V	3.20 V
Module	16S1P, 58.4 V	16S1P, 51.2 V
Modules per rack	10	8
Rack voltage	584 V	409.6 V
Strings per container	8	10
Container energy	46.7 kWh	65.5 kWh
Container type	20-ft ISO	20-ft ISO
Cooling	Glycol-water liquid, 25 °C setpoint	Liquid, 25 °C setpoint
Max cooling flow	12 L/min per rack	15 L/min per rack
Operating SOC	10–95% (post-McMicken)	10–95%
Phoenix peak ambient	47 °C	47 °C
Suppression system	FM-200 clean agent	Water mist

Challenge

McMicken Failure Sequence — Reconstructed:

The DNV GL investigation report (January 2020, 127 pages, commissioned by APS) reconstructed the following failure sequence at the McMicken facility:

At approximately 4:57 PM on 19 April 2019, a sudden voltage drop was detected in one battery module, consistent with internal short circuit onset in a single cell — likely triggered by lithium metal deposits (abnormal plating) that had accumulated over the system's ~2 years of operation at high SOC in Phoenix ambient temperatures.
Thermal runaway developed in the initial cell. The battery management system detected elevated temperature but did not isolate the module or trigger the FM-200 suppression system, because the temperature rise occurred below the suppression system activation threshold.
Thermal runaway propagated module-to-module over approximately 90 minutes. Flammable gases — primarily H₂ and CO — accumulated within the sealed ISO container to concentrations exceeding the lower flammable limit (H₂: 4% v/v in air).
When first responders opened the container side door at approximately 8:00 PM for visual inspection, the accumulated flammable gases deflagrated within 2–3 minutes. The pressure wave ejected both the side door and the rear door of the container, injuring four firefighters.
The FM-200 suppression system had not activated at any point during the 90-minute thermal runaway event because: (a) the temperature sensor threshold was not reached in the container headspace (FM-200 activates on temperature, not gas concentration); (b) there were no hydrogen gas concentration sensors in the container.

DNV GL Key Findings (Report published January 2020):

The investigation identified abnormal lithium metal deposits in the initial cell as the most likely initiating cause. The lack of thermal barriers between battery cells accelerated module-to-module propagation, and the FM-200 suppression system proved ineffective at stopping that propagation because clean agent suppression suppresses flame but does not cool the exothermic thermal runaway reactions themselves. Two critical detection gaps compounded the outcome: no hydrogen gas concentration sensors were installed inside the container, so the flammable gas accumulation went undetected before first responder entry, and no ventilation of accumulated gases was attempted before the container was opened — contrary to NFPA best practice at the time of the incident.

SunBelt Specific Technical Questions:

The audit must answer four questions that map the McMicken findings onto SunBelt's fleet. For each Gen-1 site, the study determines the cell-to-module propagation time when one cell undergoes thermal runaway at 95% SOC and 40 °C ambient, both at nominal cooling flow and at degraded flow rates. It then assesses whether H₂ gas accumulation reaches the lower flammable limit (4% v/v) within the 20-ft container before the FM-200 suppression system activates under current BMS alarm logic. For the Gen-1 NMC532 sites specifically, the analysis evaluates whether SunBelt's current control logic replicates the McMicken failure mode in which the module management system was too slow to isolate before gas accumulation reached dangerous levels. Finally, for the Gen-2 LFP sites, the study determines whether the fundamentally different thermal chemistry — lower onset temperature, lower gas generation, and non-flammable vent gases — is sufficient to bring those sites within acceptable risk bounds without retrofit.

Real-World Basis

The McMicken BESS Fire (19 April 2019)

The McMicken Battery Energy Storage System was a 2 MW / 2 MWh facility at the APS McMicken substation in Surprise, Arizona, commissioned in 2017. At the time of the incident, it was operated by Fluence under a long-term service agreement with APS. The system used LG Chem NMC cells.

The incident was the most thoroughly investigated grid-scale BESS fire in US history. The 127-page DNV GL report commissioned by APS provides the most complete public account of a large-scale BESS thermal runaway propagation event. A second investigation, commissioned by LG Chem and conducted by Exponent, reached different conclusions about root cause: while DNV GL pointed to an internal cell defect (lithium metal deposition — consistent with the abnormal plating hypothesis), Exponent identified an external heat source (possible external electrical arcing) as a potential cause. The root cause dispute between APS/DNV GL and LG Chem/Exponent was not fully resolved in subsequent regulatory proceedings.

Established Facts from the DNV GL Report:

Thermal runaway onset: approximately 4:57 PM, initially in one module
First responder arrival: approximately 6:30 PM (90 minutes after onset)
Container opening: approximately 8:00 PM (~3 hours after onset)
Deflagration upon opening: within 2–3 minutes of door opening
Estimated container overpressure at deflagration: 3–5 psi (20–35 kPa)
Injuries: 4 firefighters (burns and blast injuries); none life-threatening
System damage: total loss of the 2 MW / 2 MWh system; container structurally compromised

The Investigation Dispute — Significance for SunBelt: The LG Chem / Exponent finding of a possible external electrical arcing cause does not materially change SunBelt's risk profile, because:

If the DNV GL root cause (internal cell defect) is correct, SunBelt's Gen-1 cells (same integrator, similar age) face analogous risk
If the Exponent cause (external arcing) is correct, external electrical events remain a valid BESS trigger mechanism
In either case, the documented failure consequence — thermal runaway propagation to full container loss with flammable gas accumulation — is the same, and the recommended mitigations (gas sensors, suppression timing, thermal barriers) are identical

Thermal Runaway Data for NMC532 and LFP:

Parameter	NMC532	LFP
Thermal runaway onset (ARC, 100% SOC)	197 ± 12 °C	273 ± 15 °C
Peak cell temperature during TR	~750 °C	~450 °C
Total exothermic energy (per gram cell)	810 J/g	270 J/g
Primary vent gases	H₂, CO, CO₂, CH₄, HF	CO₂ (dominant), CO (trace)
H₂ fraction in vent gas	30–45% (v/v of total gas)	< 5% (v/v)
Lower Flammable Limit (H₂ in air)	4% v/v	—
Gas volume per 100 Ah cell (25 °C)	~75 L	~20 L
Separator onset temperature	130 °C	170 °C

Additional Reference Events:

Year	Location	System	Outcome
2018	Arizona (APS pilot site)	NMC, 1 MWh	Fire contained; no injuries
2019	McMicken, Arizona	NMC, 2 MWh	Deflagration; 4 injuries
2019	Surprise, Arizona	NMC, 2 MWh	Fire; system loss
2020	Liverpool, Australia	NMC, 30 MWh	Fire and explosion; investigation ongoing
2021	Chandler, Arizona	LFP, 10 MWh	Fire contained; no injuries (LFP advantage)

Simulation Approach

The audit uses a three-tier simulation hierarchy covering the cell, module/rack, and container/gas accumulation levels.

1. Cell-Level Electrochemical-Thermal Abuse Model

NMC532 (Gen-1): The P2D model with exothermic reaction kinetics is calibrated to ARC data for NMC532 cells. The exothermic reaction sequence proceeds through four stages: SEI decomposition (onset ~90 °C), anode-electrolyte reaction (onset ~120 °C), NMC532 cathode decomposition (onset ~197 °C), and electrolyte combustion (onset ~200 °C). Total exothermic heat is approximately 810 J/g cell mass for a 100 Ah NMC532 cell at 100% SOC.

LFP (Gen-2): The LFP cathode (olivine structure) is thermally stable to approximately 270 °C vs. 200 °C for NMC532 because the Fe-O bond in the olivine structure is substantially stronger than the Ni-O bond in the layered oxide. Exothermic energy release is approximately 270 J/g cell mass -- one-third of NMC532. The vented gas is CO2-dominant; H2 generation is minimal (< 5% of total vent gas volume), preventing H2 accumulation to the LFL in any realistic container scenario.

2. Module and Rack CFD Thermal Model

The 16S1P module geometry for each site is reconstructed from SunBelt's as-built drawings. A 3D CFD model in newtsim Span newtsim Stream with conjugate heat transfer captures all relevant heat paths: liquid cooling channel flow at the measured per-site flow rate from SCADA data, natural convection within the module, inter-cell radiation, and contact conduction through the aluminium compression plates. For the triggered cell, the P2D abuse model heat source is applied. Propagation is tracked by monitoring adjacent cell surface temperatures, declared at 130 °C (separator shutdown, NMC532) or 170 °C (separator shutdown, LFP).

Container-level gas accumulation is modelled as a well-mixed volume (20-ft ISO container internal volume: 33.2 m³) with vented gas mass flow from the cell-level model. H2 concentration is tracked; LFL crossing (4% v/v H2 in air) is the primary danger threshold.

3. Gas Accumulation and Suppression Timing Model

Vented gas quantities per cell (per 100 Ah) from the literature:

Gas Species	NMC532 (L/100 Ah)	LFP (L/100 Ah)	Flammable?	LFL (v/v in air)
H₂	24.8	0.8	Yes	4%
CO	8.2	2.4	Yes	12.5%
CO₂	18.4	14.2	No	—
CH₄	3.1	0.4	Yes	5%
HF	1.2	0.1	No (toxic)	—
Total (STP)	~75 L	~20 L	—	—

For the Gen-1 NMC532 container, one module in thermal runaway (16 cells × 100 Ah) generates approximately 16 × 24.8 = 397 L of H₂ at STP. In the 33.2 m³ container, this represents a H₂ concentration of 397/33,200 = 1.2% — not yet at the 4% LFL. However, propagation to a second module (16 more cells) brings the total H₂ to approximately 2.4%, and propagation to a third module reaches 3.6% — approaching LFL. With a rack of 10 modules and 8 racks per container (80 modules total), full propagation produces approximately 60% H₂ concentration — far above the Upper Flammable Limit (75%). The risk window is the 2–3 module propagation stage, when H₂ is in the 3–5% range: at or near LFL.

Simulation Caveats

Classification: STRETCH. This study uses a 3D CFD conjugate heat transfer model coupled to single-cell ARC-derived exothermic kinetics and a well-mixed container gas accumulation model — a hierarchy that has been validated in controlled settings but involves significant extrapolation when applied to multi-module propagation chains in as-deployed BESS containers.

The most significant uncertainty is in the gas venting rate extrapolation from single-cell ARC data. The H₂ generation rate (0.33 L/Ah for NMC532) is taken from single-cell accelerated rate calorimetry measurements at 100% SOC, but in a propagating multi-cell event cells vent sequentially at varying SOC levels and temperatures, so the aggregate H₂ generation rate may differ from the single-cell rate scaled by cell count. Additionally, ARC data from the literature covers a range of cell designs, and direct ARC data for SunBelt's specific Gen-1 cells was not available, introducing ±30--40% uncertainty in the absolute H₂ accumulation timing.

A second source of uncertainty is the vent path geometry. The gas accumulation model assumes a well-mixed container volume (33.2 m³) with uniform H₂ distribution. In reality, vent gas exits through module safety vents directed toward specific container zones, buoyancy effects concentrate H₂ near the container ceiling, and vent paths may be partially obstructed by rack hardware. These geometry-dependent effects can advance the local LFL crossing time at ceiling sensors by 10--25% relative to the well-mixed prediction, making the LFL margin estimates for Gen-1 sites conservative for ceiling-level sensor placement but non-conservative for lower measurement points.

The third caveat concerns FM-200 fire suppression response timing. The suppression timing model assumes FM-200 discharge reaches full concentration within 10 seconds of activation and that activation occurs when the BMS issues the suppression command. In practice, aging FM-200 systems may have delayed actuation, reduced agent mass (cylinder pressure loss), or blocked nozzles — particularly at 2017--2018 vintage sites. A 30-second actuation delay compresses the suppression margin at Gen-1 sites from the tabulated values by approximately 20--30%.

Recommended framing: Present propagation timing results as "simulation-estimated bounds" rather than precise predictions, with explicit statement that H₂ accumulation rates carry ±40% uncertainty from the single-cell ARC extrapolation. The LFL margin analysis is best framed as a risk-ranking tool (Site 3 IMMEDIATE vs. other sites MARGINAL/ACCEPTABLE) rather than as precise absolute timings. Recommend physical H₂ sensor deployment data after installation as a validation dataset for the gas accumulation model.

Key Predictions / Results

Gen-1 NMC532 Fleet — Thermal Propagation Analysis:

Metric	Nominal Cooling (12 L/min)	Degraded Cooling (7.2 L/min, 60%)	Site 3 (4.8 L/min, 40%)
Thermal runaway onset (100% SOC)	197 ± 12 °C	197 ± 12 °C	197 ± 12 °C
Cell-to-module propagation time	4.2 ± 0.8 min	3.1 ± 0.7 min	2.4 ± 0.6 min
Module-to-adjacent-module	18 ± 4 min	13 ± 3 min	9 ± 2 min
H₂ LFL (4%) reached in container	23 ± 6 min	17 ± 4 min	12 ± 3 min
FM-200 activation (current BMS logic)	8–12 min	8–12 min	8–12 min
Suppression margin before LFL	11–15 min positive	5–9 min positive	−5 min
Risk status	Acceptable	Marginal	IMMEDIATE ACTION

Gen-2 LFP Fleet — Thermal Propagation Analysis:

Metric	Value	Notes
Thermal runaway onset (100% SOC)	273 ± 15 °C	76 °C higher than NMC532
Cell-to-module propagation time	28 ± 6 min	6.7× longer than NMC532
H₂ LFL reached in container	Not reached	CO₂-dominant vent gas; H₂ < 5% of total
FM-200 activation adequacy	Adequate in all scenarios	Large margin to LFL
Risk status	Acceptable — no urgent action	Recommend FM-200 maintenance at 1 site

Fleet Risk Register Summary (12 Sites):

Site	Chemistry	Vintage	Cooling Status	TR Propagation Time	LFL Margin	Priority
Gen-1 Site 1	NMC532	2017	Nominal	4.2 min	15 min	BMS update (6 mo)
Gen-1 Site 2	NMC532	2017	80% nominal	3.4 min	7 min	BMS update + flow check
Gen-1 Site 3	NMC532	2018	40% nominal	2.4 min	−5 min	IMMEDIATE
Gen-1 Site 4	NMC532	2018	Nominal	4.2 min	14 min	BMS update (6 mo)
Gen-1 Site 5	NMC532	2018	90% nominal	3.9 min	12 min	BMS update (6 mo)
Gen-1 Site 6	NMC532	2017	Nominal	4.2 min	15 min	BMS update (6 mo)
Gen-2 Site 7	LFP	2020	Nominal	28 min	>> LFL	Routine maintenance
Gen-2 Site 8	LFP	2020	Nominal	28 min	>> LFL	FM-200 past service date
Gen-2 Site 9	LFP	2021	Nominal	28 min	>> LFL	Routine maintenance
Gen-2 Site 10	LFP	2021	Nominal	28 min	>> LFL	Routine maintenance
Gen-2 Site 11	LFP	2021	Nominal	28 min	>> LFL	Routine maintenance
Gen-2 Site 12	LFP	2021	Nominal	28 min	>> LFL	Routine maintenance

Fleet risk register — propagation time vs suppression margin by site

Recommended Fleet-Wide Interventions (Priority Order):

IMMEDIATE — Gen-1 Site 3 cooling system repair: Measured cooling flow 40% below nominal due to a pump impeller wear defect identified in SCADA data. At 40% flow, propagation time is compressed to 2.4 minutes and the H₂ LFL is reached before FM-200 can activate under any plausible BMS logic. Site 3 cooling must be restored before the next operating season.
6-month priority — Gen-1 Sites 1–6 BMS alarm logic update: Current BMS alarm logic has an 8–12 minute delay between first thermal event detection and FM-200 activation. The McMicken system had the same delay. Target: reduce to 3-minute maximum (via rate-of-change temperature threshold added to absolute temperature threshold). This provides adequate suppression margin under nominal cooling at all sites.
12-month priority — Gen-1 Sites 1–6 H₂ gas concentration sensors: H₂ sensors inside each container (4 sensors per container, at ceiling height) would have detected the McMicken gas accumulation approximately 45 minutes before first responder arrival — preventing the deflagration injury event. Installation is a standard retrofit; sensors cost approximately $800 each; total fleet cost approximately$ 20,000 for all 24 Gen-1 containers (6 sites × 4 containers per site).
Routine maintenance — Gen-2 Site 8 FM-200 replacement: The FM-200 cylinder at Gen-2 Site 8 is past its service date. Replacement is standard maintenance; no thermal safety implication for LFP chemistry, but regulatory compliance requires it.

Comparison Methodology

The cell-level abuse model validates the pack-level CFD propagation predictions. The DNV GL McMicken reconstruction and published test data provide secondary confirmation:

1. DNV GL McMicken Report Cross-Check The NMC532 cell-to-module propagation prediction (4.2 min nominal) is compared against the implicit timing in the DNV GL reconstruction. The 90-minute gas accumulation period before first responder arrival, combined with the module-to-module propagation interval of 18 minutes, implies approximately 3--5 propagation events in 90 minutes — consistent with the model predicting module-to-module at 18 minutes and a full 8-rack container requiring approximately 8 × 18 = 144 minutes for complete propagation (with time compression as heating intensifies). The DNV GL reported deflagration conditions are consistent with the H₂ accumulation model predicting LFL crossing after approximately 3 module propagation events.

2. UL 9540A Published Propagation Data

NMC532 module thermal propagation heatmap — nominal vs degraded cooling Where SunBelt's module geometry matches tested configurations in published UL 9540A data, propagation timing is cross-validated. Published NMC large-format prismatic cell propagation data reports approximately 3.8--5.1 minutes for configurations comparable to SunBelt's Gen-1 modules -- consistent with the simulation prediction of 4.2 ± 0.8 minutes.

3. NREL LFP Gas Generation Data The Gen-2 LFP H2 accumulation model uses published gas generation rates. Cross-validation against NREL cell-level gas generation data for LFP at 100% SOC confirms H2 generation of approximately 0.8 L per 100 Ah, consistent with the model's LFP H2 below LFL result.

4. SunBelt SCADA Operational Data Validation 12 months of SCADA historian data (cooling flow rates, cell temperatures, module voltages) from all 12 sites is used to: (a) identify Site 3's degraded cooling before simulation; (b) validate nominal cooling temperature rise profiles at other sites against the CFD model predictions; (c) identify any sites with systematic temperature elevation that might indicate early-stage capacity fade or localised heating. The SCADA validation gives SunBelt and the underwriter confidence that the simulation represents actual operating conditions.

Deliverables

Fleet risk register: 12-site tabular assessment of thermal propagation time, H₂ LFL crossing time, suppression margin, and overall risk rating. Formal risk register in NFPA 855 compliance format.
Site-specific propagation time vs. ambient temperature curves: propagation time at 35 °C, 40 °C, 44 °C, and 47 °C (Phoenix design-day ambient) for each Gen-1 site, at nominal and degraded cooling flow rates.
Cooling system flow sensitivity analysis: propagation time vs. cooling degradation (40%, 60%, 75%, 90%, 100% nominal flow) for NMC532 Gen-1 sites — identifies the minimum acceptable flow rate for each site to maintain adequate LFL margin.
BMS alarm logic specification update: Engineering specification for the revised alarm logic for Gen-1 sites — rate-of-change temperature threshold (5 °C/min sustained over 60 s) triggering suppression within 3 minutes of first alarm, regardless of absolute temperature threshold status.
H₂ sensor placement specification: Optimal sensor positions (ceiling height, quantity per container) with fail-safe requirements, alarm thresholds (alert at 1.5% H₂, evacuation and suppression at 2.5% H₂), and integration with SunBelt's SCADA alarm system.
Insurance underwriter technical report: Quantified risk assessment for all 12 sites, retrofit plan, estimated residual risk after interventions, and third-party simulation validation summary. Formatted for underwriter review per NFPA 855 and FM Global loss prevention standards.
Executive briefing for utility board: 20-slide presentation for SunBelt's utility parent company board, covering regulatory context, McMicken incident lessons, fleet risk profile, and retrofit plan.

Delivery timeline: 8 weeks from receipt of SunBelt as-built drawings and SCADA historian data access.

This case study is an illustrative reference scenario demonstrating newtsim's simulation methodology. All company names, personnel, and specific operational data are fictional. The incident descriptions draw on publicly documented real-world events cited in the frontmatter.