Thermal Modelling · Simulation · Optimization

Thermal answers,
not thermal guesswork.

NovaTherm couples a cited materials-property database with a deterministic SINDA/G solver and Latin-hypercube scenario sweeps — so you evaluate active and passive thermal management across every operating mode and environment before you cut metal.

Start a project Explore the solver →
114
Cited materials indexed
Flexible importCAD geometry (STL / OBJ / STEP), SINDA/G decks & material datasets fold into one model.
Modular librariesReusable cited-material and native-model libraries (Bridgman cuvette, enclosure parametric).
Powerful outputsPercentile envelopes, node-temperature fields & model-vs-test overlays.
CASE: heatsink_v7 · steady-state
T_max 91.2°C
°C 25 95
SOLVING · 84%
IMPORT · ANY
CAD, measured curves, spec sheets & datasets fold into one model.
SWEEP · MONTE CARLO
Latin-hypercube scenario sweeps with percentile envelopes across operating modes.
OPTIMIZE · A+P
Active and passive management, tuned to your objective function.
MATERIALS · 114
Temperature-dependent properties, cited and version-controlled.
Try it · In-browser

Interactive thermal calculators

Estimate a junction temperature rise before you open the solver. Free, in-browser, no sign-in — the first step from guesswork to a defensible number.

Open the toolbox →
θ_JA ESTIMATOR
Power dissipated8.0 W
Airflow200 LFM
Est. junction rise +38.6 K
Order-of-magnitude estimate (θ_JA airflow model, illustrative). Final coefficients pending NovaTherm validation. Run the solver for a defensible result.
Applications · What you can model

One solver. The whole thermal-management design space.

From a Bridgman crystal-growth cuvette to a satellite avionics box to a building envelope, NovaTherm composes the same pipeline — cited materials, a deterministic SINDA/G solver, and Latin-hypercube scenario sweeps — across domains that usually demand separate tools.

NovaThermal THERMAL CONSOLE Missions Models Jobs MODEL INPUT Native parameters Built-in cuvette model Direct ingest SINDA/G deck to solver CAD geometry STL / OBJ / STEP to FEM Hybrid compose Sub-models + port wiring Enclosure (parametric) Electronics box, CFD-lite Propose spec via AI scout · qwen3.6 · Gemini Flash RUN CONFIGURATION Run type Steady state Transient Sweep parameter T_hot — hot boundary (deg C) v Sampling Latin hypercube (LHS) Range (deg C) low 28 to high 32 Cases 16 Seed 12345 Percentiles 5 , 50 , 95 SWEEP RESULTS (schematic) 72 64 56 48 40 70C LHS cases P95 P50 P5 limit P50 peak 56 C P95 margin +6 C Limit check node S4 max 70 C ok Artifacts summary_csv stats_json npz provenance margin_report MISSION TREE · snapshot to fork to sweep to stitch to versioned audit trail | CASE SETS · named, reusable run configs
Modelling Workbench

Model Inputs and Sweep Ranges Under Engineer Control

Five model paths, Latin-hypercube boundary sweeps, and per-node limit checks — composed in one engineered submission form
From a single submission form, engineers configure analyses that once demanded bespoke scripting. Choose your model source — native SINDA/G node parameters, a direct deck ingest, or CAD geometry (STL, OBJ, or STEP) meshed to FEM shells or voxels. Set run type to steady-state or transient, pick a thermal boundary to sweep, and let Latin-hypercube sampling span the ranges you define. P5/P50/P95 envelopes emerge automatically; per-node temperature thresholds flag margin exceedances against the sweep results before they reach hardware. Run outputs — summary CSV, statistics JSON, and provenance records — are emitted as selectable artifacts.
Live capability
MATERIALS LIBRARY 114 entries · ASM Handbook + manufacturer-datasheet citations · full property sets representative entries shown 114 CITED PROPERTY SET: ρ cp k CTE ε αs Tmin / Tmax ✓ source citation Al 6061-T6 Aluminum alloy k 167 W/(m·K) ε 0.05 αs 0.09 CTE 23.6 ppm/K T -73 / 149 °C ASM Vol. 2 Ti-6Al-4V Titanium alloy k 6.7 W/(m·K) ε 0.10 αs 0.40 CTE 8.6 ppm/K T -196 / 315 °C ASM Vol. 2 Kapton HN Polyimide film k 0.12 W/(m·K) ε 0.86 αs 0.37 CTE 20.0 ppm/K T -269 / 400 °C DuPont Datasheet · · · 111 more entries: optical coatings · beryllium · graphite-epoxy · invar · TIMs · paints · adhesives · · · ROADMAP — NOT YET AVAILABLE NIST temperature / pressure-dependent property profiles · k(T), cp(T) curves per material Planned: NIST SRD + JANAF sourcing · T-dependent curves · direct integration with solver parameter sweeps 114 sourced materials 7 properties per entry 100% cited (ASM / datasheet) Spacecraft metals Optical coatings + composites, TIMs, more
Materials Library

114 Sourced Materials, Every One Cited

A thermal engineer's library built on ASM Handbooks and manufacturer datasheets — not guesswork.
NovaThermal ships with 114 materials, each carrying seven cited thermal properties: density, specific heat, conductivity, CTE, emissivity, solar absorptivity, and operating-temperature bounds — single reference-condition values, not full temperature curves. Every entry traces to a published source — an ASM Handbook volume or a manufacturer datasheet — so every material assumption you carry into peer review or a customer audit cites where it came from. Select a material, verify the inline citation, run the model. Documented input properties, every time.
114 cited entries
RoadmapNIST temperature/pressure-dependent property curves and custom-material upload are roadmap — today’s library is a fixed, cited, single-value set.
THERMAL FIELD Bridgman Cross-Section Front Hot Mid Cold T distribution · SINDA/G FEM SCENARIO SWEEP · PERCENTILE ENVELOPES 390 370 350 330 310 K 1 4 8 12 16 Scenario Index (n=16, LHS sampling) LIMIT P95 P50 P5 PROVENANCE RECEIPT PASS deck-sha a4f7c2e1 solver SINDA/G sweep LHS n=16 pct=5/50/95 limits PASS (user bounds + SSP-51721: 45C / 49C) materials 114 cited (ASM / manufacturer datasheets) geometry STL/OBJ/STEP to FEM (fem-shell | fem-voxel) NovaThermal · thermalresults.com · deterministic solver: same deck, same result, every time
Post-Processing

Deterministic, Traceable, Review-Ready

Percentile envelopes, limit checks, and provenance — NovaThermal delivers traceable thermal artifacts, not just temperatures
A number without a lineage is an engineering liability. NovaThermal's deterministic SINDA/G solver maps every input deck to exactly one output — so any run can be fingerprinted and reproduced — and material properties drawn from the library trace to their 114 ASM/manufacturer-cited entries. Attach a provenance record and statistics JSON as run artifacts, and Latin-Hypercube scenario sweeps return 5/50/95 percentile envelopes over your defined parameter ranges — the sampled parametric spread across those sweep scenarios, not just a nominal answer. Limit checks fire automatically, including ISS crew-contact thresholds per SSP-51721. What you hand to a review board is a traceable thermal result with its own sampled sweep envelope.
Traceable output
CAD INGEST · FEM MESH · THERMAL SOLUTION INPUT GEOMETRY .STL .OBJ .STEP h Bracket geometry (schematic) AUTO MESH FEM MESH fem-shell fem-voxel Nodes + elements auto-generated SINDA/G SOLVE THERMAL FIELD steady transient Tmax Tmin T50 Nodal temperature distribution GEOMETRY FORMATS STL · OBJ · STEP MESH PATHS fem-shell · fem-voxel SOLVER SINDA/G · steady + transient MATERIAL LIBRARY 114 ASM/datasheet-cited THERMALRESULTS.COM · NOVATHERMAL · ILLUSTRATIVE SCHEMATIC
CAD → Solver

Your Geometry, Straight Into the Solver

Drop in an STL, OBJ, or STEP file, mesh it to fem-shell or fem-voxel, and hand it to the SINDA/G solver — in one session.
Geometry rebuilding is where thermal analysis projects stall. NovaThermal collapses it: import STL, OBJ, or STEP directly, mesh it to fem-shell for thin-walled structures or fem-voxel for solid bodies, apply boundary conditions, and draw material data from 114 ASM/manufacturer-cited entries — in the same submission form. What comes out is a discretized SINDA/G thermal model tied to your actual geometry: a mesh, not a hand-rebuilt proxy. Steady-state or transient, the path from CAD file to nodal temperature field stays in one session instead of a separate geometry-prep handoff.
In use on real NASA work
SCENARIO SWEEP · LHS SAMPLING → PERCENTILE ENVELOPES 50 100 150 200 10 30 50 70 90 110 Heat Load (W) Peak Temp (°C) 95th 50th 5th LHS sample scenarios Median (50th pct) 5th pct (lower bound) 95th pct (upper bound) Sweep: boundary-T · heat-load · flow-rate SINDA/G solver · LHS sampling · pct 5 / 50 / 95
Scenario Sweeps

Sweep the Design Space. See the Percentile Envelope.

Latin Hypercube sampling drives the SINDA/G solver across your defined parameter ranges — 5th, 50th, and 95th percentile temperature envelopes in a single automated pass
Every thermal design hides unknowns: boundary temperatures that drift, heat loads that spike, cooling flows that vary. NovaThermal's scenario-sweep engine samples across your defined parameter ranges via Latin Hypercube sampling — then drives the deterministic SINDA/G solver across each sampled scenario, delivering 5th, 50th, and 95th percentile temperature envelopes in a single automated pass. The result is a structured parametric spread: not just the nominal answer, but the sampled response of your design across the ranges you set. Quantify your margin across those scenarios, and see where node limits are exceeded — before hardware.
LHS + percentiles
RoadmapNo published product-wide accuracy figure — uncertainty is characterized per model, and million-sample campaigns are roadmap.
Diurnal / Annual Thermal Cycle — Building Enclosure ENCLOSURE THERMAL NETWORK EXTERIOR AMBIENT solar T_ext RadiativeExchange σ·ε·F·A → T_env CLADDING INSULATION T_wall ConductionInterface h_contact·A GYP INTERIOR ZONE T_int PLENUM mdot·cp advection fan_operating_point recirc 114 ASM-cited materials · FDM/FEM transient solver T_env BOUNDARY PROFILES 24-HOUR DIURNAL 35°C 20°C 5°C 0h 6h 12h 18h 24h lag T_ext (T_env boundary) T_int (zone response) ANNUAL SEASONAL 35°C 10°C -15°C Jan Mar Jun Sep Dec T_set heating cooling Swap T_env array: orbit eclipse profile or building climate file FDM + FEM transient Plenum + Fluid port RadiativeExchange
Cross-Domain

From Orbit to Architecture: the Same Transient Physics, Illustrated

The same transient conduction, radiation, and fluid-advection solver that tracks a satellite through eclipse composes into an illustrative building-envelope model — a physics demonstration of the 24-hour and seasonal swing, not a certified HVAC tool.
NovaThermal's Plenum component models well-mixed HVAC zone air via mdot*cp advective conductors — the same fluid-port abstraction used for spacecraft coolant loops. Couple it to ConductionInterface wall-layer conductances (h_contact*A or n_bolts*g_per_bolt), a RadiativeExchange sink whose env_node boundary the caller updates each time step to sweep T_ext across a diurnal or seasonal profile, and a fan_operating_point for duct resistance, and you have an illustrative coupled thermal network that demonstrates the same transient conduction, radiation, and fluid-advection physics used for spacecraft. In that illustrative model, the transient solver resolves peak-load timing, wall time-lag, and sizing-sensitivity trends — as a physics demonstration, not a certified HVAC calculation. The current fluid model uses constant cp from the fluids registry; NIST temperature/pressure-dependent profiles for humid air are a roadmap item, and no validated comparison against EnergyPlus or TRNSYS reference simulations exists.
Illustrative application
RoadmapNot a certified building-energy tool — no EnergyPlus / TRNSYS validation, and humid-air NIST profiles are roadmap. The physics is real; the building-energy certification is not claimed.

Proof, plotted

All worked examples →
EV BATTERY PACK
Peak cell temperature, mapped
Immersion vs. cold-plate cooling across a drive-cycle envelope.
Illustrative — real run pending
SATELLITE AVIONICS
Full eclipse-cycle survival
Transient response through orbital thermal cycling.
Illustrative — real run pending
BRIDGMAN CUVETTE
Node-temperature field vs. test
Native model with real breadboard-trace overlay — live in the library.
Real — view in library

Bring us your hardest thermal problem.

Scoping call with a thermal engineer, not a sales rep.

Talk to an engineer