Milestone · verified in public
1,000+

validations against LBNL THERM — and still counting.

Always striving. Always transparent.

Every window and façade section AFRC Lab models in CFL — our own 2-D ISO 15099 solver — is also run in LBNL THERM 7.8, the reference the industry already trusts. Not once, as a launch stunt. Every section, every time, written into a public register anyone can read. We've now passed 1,000 head-to-head checks, and the two solvers agree to 99.0%.

Live count, read straight from the record: — THERM runs · — compared values · updated —

—
THERM runs compared
—
Compared values
—%
Agreement with THERM
—%
Median |difference|
Read the live register → Try BTW yourself →
Why it matters

It's the science, not the software.

Two-dimensional heat transfer is governed by physics — the equations set out in ISO 15099 — not by whose logo is on the software. Any engine that solves those equations, under the same geometry and the same boundary conditions, lands in the same place. That's not an opinion. It's something you can either assert, or show.

We chose to show it. So we built CFL — the Computational Facade Lab solver — to model the actual installed window straight from CAD, with a detailed radiation model and clean meshing of the thermal breaks that trip up hand-traced models. And then we did the unglamorous part: we ran the same sections through LBNL THERM as well, and compared them, and kept the receipts.

A thousand times over. Because "trust me" isn't evidence — and the one answer that survives scrutiny is the one you can watch being checked.

A rating asks you to trust a number. We'd rather show you the working — a thousand times, in public.
What the record shows

1,000+ checks. 99.0% agreement. Nothing hidden.

Independent

CFL is AFRC Lab's own solver, checked against LBNL's THERM — the tool the industry already accepts. Two engines, the same geometry, the same conditions, compared on the same tagged zones.

Rigorous

Frame, edge-of-glass, centre-of-glass and total heat flow, across aluminium, uPVC, timber, thermally broken and steel systems, curtain wall and skylights at the NFRC 20° slope. 99.0% agreement, median difference about 0.6%.

Transparent

Every section is logged, anonymised, in a register that updates itself from live data. We even show where THERM itself won't converge, and leave those out rather than lean on a shaky reference. The honest bits are the point.

This is why the milestone matters beyond a round number. When a reviewer asks whether an AFRC Lab result holds up — including the fair question of whether a non-NFRC-listed tool belongs in a submission — the answer isn't a sentence of reassurance. It's a register they can open and read, and a method that stands on the physics rather than on a brand.

Questions, answered

The things people actually ask us.

What is CFL, AFRC Lab's thermal solver? +
CFL (Computational Facade Lab) is AFRC Lab's own two-dimensional finite-element heat-transfer solver, run to ISO 15099 and NFRC 100. It models the real installed window, door or junction — frame, glass and the interface around it — rather than a generic catalogue rating.
How is CFL validated? +
Every section AFRC Lab models in CFL is also built and run in LBNL THERM 7.8 under the same geometry and boundary conditions. The two are compared and logged in a public register — more than 1,000 THERM runs to date, with 99.0% agreement and a median difference of about 0.6%.
Is the software NFRC-listed or AFRC-accredited? +
No. CFL is not on the nominated-software list for a formal AFRC (Australian Fenestration Rating Council) rating label, and AFRC Lab does not claim NFRC accreditation. AFRC Lab is an independent consultancy — it is not the AFRC. Its results are reliable evidence of 2-D heat transfer, suited to an NCC Performance Solution; final NCC compliance is determined by your assessor or certifier. The validation record exists precisely to show it's the science, not the software, that governs the result.
Why build your own solver instead of only using THERM? +
CFL imports the section directly from CAD (removing manual-tracing error), uses a detailed view-factor radiation model, and meshes complex thermal breaks cleanly. Because 2-D heat transfer is governed by the physics in ISO 15099, any engine solving the same equations under the same conditions returns the same answer — which is exactly what the register demonstrates, section by section.
Where can I see the data? +
The full, anonymised record is at afrclab.com.au/verification. It reads from a live file and updates automatically as new sections are verified — the numbers are never typed by hand.
See it for yourself

Send us one window. We'll model it free.

The best way to judge a method is to watch it work on something you know. Send one window or one junction you've already rated, and we'll model the real installed system in CFL and hand back the result, the model screenshots and the QA checklist — so you can see exactly what independent, validated proof looks like.

Try BTW yourself → Send us a window → Read the register →

CFL is AFRC Lab's own 2-D finite-element solver (Computational Facade Lab), run to ISO 15099 / NFRC 100; THERM 7.8 is LBNL's reference tool. This is an independent method check — AFRC Lab does not claim NFRC accreditation or certification, and "AFRC Lab" is not "the AFRC" (the Australian Fenestration Rating Council). Modelled results are indicative to the systems and sizes provided; final NCC compliance is determined by your assessor / certifier. Figures shown update live from the verification record and reflect the count at the time of viewing.