Glazing Performance Calculator (Open-Data Beta)

How the open-data model works

The open-data model replaces the current calculator's parametric glass and coating curves with physics computed from published optical constants. Everything else is shared: both models feed the same multiple-reflection optics, the same ISO 15099 center-of-glass thermal model and the same NFRC 100 and 200 conditions, so the two columns of the results differ only in how each pane's spectra are made.

Glass. Each pane uses the refractive index \(n(\lambda)\) and extinction coefficient \(k(\lambda)\) that Rubin measured for soda-lime float glass: clear, low-iron and three tints (green, gray and bronze). The internal transmittance of a pane of thickness \(d\) is

and the two air–glass surfaces add Fresnel reflection, so any thickness can be modeled. Against the measured IGDB clear-glass records shipped with this page, the Rubin glass is within about 0.1 point in visible transmittance and 0.5 to 0.8 point in solar transmittance.

Coatings. A low-e coating is a stack of thin films a few nanometers to a few hundred nanometers thick: silver for infrared reflection, dielectric layers (titanium dioxide, silicon nitride) to suppress visible reflection, and a thin nickel blocker. Its optics follow from the characteristic matrix of each layer at normal incidence,

with \( R = |(n_0 B - C)/(n_0 B + C)|^2 \) and \( T = 4 n_0 n_s/|n_0 B + C|^2 \), where \(N_j = n_j - i k_j\) is the complex index of layer \(j\). The optical constants are public-domain (CC0) data from the refractiveindex.info database; the pyrolytic hard coat uses published Drude–Lorentz models of tin oxide and fluorine-doped tin oxide.

The coated pane. The film and the thick glass combine incoherently:

where \(T_c\) is the film's transmittance, \(R_0\) the bare glass surface's reflectance, and \( R_g \) the film's reflectance seen from inside the glass. The classic model reuses the air-side value for \(R_g\). Within the visible band the two differ by up to about 14 points for the double-silver stack and 5 points for the triple-silver stack, mostly near the ends of the band, and by less than 1 point for the single-silver stack.

Calibration. Four generic stacks (hard coat, single, double and triple silver) had their layer thicknesses fitted offline only to published datasheet values: visible transmittance, reflectance, SHGC, UV transmittance (computed as LBNL WINDOW computes it for datasheets) and, for the hard coat, solar transmittance. Color windows taken from the manufacturers' appearance statements keep the reflected color neutral. The emissivity of the coated face is then inferred from the published air and argon U-factors. A fitted stack is an effective model of a coating family, not any manufacturer's actual design. The panel “How the open coatings are built” shows every layer and every residual, and the comparison panel shows the datasheet rows the fit never saw.

Measured products. The nine named products are unmodified International Glazing Database (IGDB) records that LBNL distributes in its open-source pyWinCalc and Windows-CalcEngine projects. Your browser reads each file when you first select the product and resamples its spectra to the calculator's 5 nm grid for that calculation only; a measured pane is identical in both models.

Solar spectrum. The solar quantities (solar transmittance, SHGC, UV) depend on the solar spectrum used to weight them. This page defaults to the ASTM E891 direct-normal spectrum, which LBNL's NFRC calculations in Windows-CalcEngine use, so its numbers are comparable with manufacturers' datasheets computed in LBNL WINDOW; the datasheet comparisons always use it. ASTM G173 (direct plus circumsolar), the newer reference spectrum, has more ultraviolet and visible light; with it, SHGC comes out about 0.01–0.03 higher for silver low-e coatings.

Assumptions and limits

• Center-of-glass values at normal incidence only. Frames, edge spacers, angular optics, laminates, films and shades are not modeled.
• Nickel stands in for the nickel-chromium blockers used in production, for which no open optical data exist.
• The hard coat's emissivity comes from its published U-factors, not from the film's infrared optics.
• Some manufacturers' clear glass is clearer than Rubin's: Cardinal's 6 mm clear double glazing reads about 1.8 points of visible transmittance higher than the open model.
• The color windows used in the calibration are published descriptors (“neutral”), not measured color coordinates.
• The thermal model is the same as the current calculator's.

References

1. M. Rubin, “Optical properties of soda lime silica glasses,” Solar Energy Materials 12, 275–288 (1985).
2. M. N. Polyanskiy, “Refractiveindex.info database of optical constants,” Scientific Data 11, 94 (2024), doi:10.1038/s41597-023-02898-2.
3. H. U. Yang et al., “Optical dielectric function of silver,” Phys. Rev. B 91, 235137 (2015); T. Siefke et al., Adv. Opt. Mater. 4, 1780–1786 (2016); K. Luke et al., Opt. Lett. 40, 4823–4826 (2015); I. H. Malitson, J. Opt. Soc. Am. 55, 1205–1208 (1965); A. D. Rakić et al., Appl. Opt. 37, 5271–5283 (1998).
4. K. von Rottkay and M. Rubin, “Optical indices of pyrolytic tin-oxide glass,” Mater. Res. Soc. Symp. Proc. 426, 449 (1996), LBNL-38586.
5. H. A. Macleod, Thin-Film Optical Filters (CRC Press), for the characteristic-matrix method.
6. ISO 15099:2003, Thermal performance of windows, doors and shading devices — Detailed calculations.
7. NFRC 100 (U-factor), NFRC 200 (SHGC and VT) and NFRC 300 (optical properties) procedures.
8. ISO 9050:2003, Glass in building — Determination of light transmittance, solar direct transmittance, total solar energy transmittance, ultraviolet transmittance and related glazing factors.
9. ASTM E891-87, direct normal solar spectral irradiance (as tabulated in LBNL Windows-CalcEngine), and ASTM G173-03 reference solar spectra.
10. Lawrence Berkeley National Laboratory, WINDOW and the International Glazing Database.