How to Optimize Smart Glass Facades for High Performance with LBNL Software

The Importance of Smart Glass Simulation

Smart glass is the only building material that can dynamically reject solar thermal energy, enhance privacy and reduce visual glare, on-demand.

You can read about the basics of smart glass in our tutorial article.

Smart glass reduces HVAC & lighting costs, improves user comfort and can be driven by weather sensors, live energy feeds, and building occupancy data.

Application areas include: 

• Commercial & residential building facades
• Vehicles & transportation hubs
• Data centres and industrial plants, and
• Smart city infrastructure, such as air-conditioned tram stops

The Role of Software Simulation

But how can we estimate the energy and cost savings on a project, prior to build? 

How will the smart glass and low-e layers affect the solar heat gain (SHGC)?

How will the choice of insulating gas, or the vacuum gap pillar dimensions, affect the U-value?

We need ballpark answers to these questions before we close out the glazing design, since the facade will affect the operational efficiency of the building over its lifetime.

The aim of this article is thus to guide architects, facade engineers and window manufacturers in how to optimize glazing for high performance, before embarking on fabrication or procurement.

Overview of LBNL Simulation Tools

In this article, we use the Lawrence Berkeley National Laboratory (LBNL) software tools, namely OPTICS v6 and WINDOW v8.1, to model the thermal and optical characteristics of a smart glass insulating glazing unit (IGU).

We will use SPD smart glass as the underlying technology in this case, but we could re-run this example with electrochromic glass, electrophoretic glass or liquid crystal glass.

Reach out if you are interested in exploring this idea.

Here is a summary of the most relevant LBNL software tools. Please note that we will only be using OPTICS and WINDOW in this article.

Designing an Energy-Efficient Smart Glass IGU

Our smart glass IGU will look like below (dimensions not to scale, obviously).

The normal convention is to number the glass surfaces, starting from #1 on the left-hand (outer) glass surface, and increasing as we move rightward (and inward) into the stack.

IGU Stack Design

This glass stack can also be written as follows, which is useful shorthand when communicating with your IGU supplier:

Glass + PVB + SPDCore + PVB + Glass + Vacuum + LowE + Glass

What is SPDCore?

The ‘SPDCore’ is our PET-encapsulated SPD smart film, structured internally like this:

PET + ITO + SPD + ITO + PET

The ‘SPD’ (Suspended Particle Devices) layer is a smart switchable emulsion, sandwiched between Indium Tin Oxide (ITO) transparent conductors, and Polyethylene Terephthalate (PET) plastic substrates on both sides.

• PET: functions as a laminating adhesive and bonds the smart film to the glass;
• ITO: transparent conductive layer that connects the smart glass electrically and is deposited onto the PET film directly;
• SPD: microscopic, rod-like particles, suspended in a liquid or gel matrix, which absorb or transmit sunlight.

How SPDCore Works

When we switch off the voltage on the smart glass, the microscopic particles inside the SPD emulsion orient themselves randomly and absorb sunlight, giving the glass a dark tint.

The SPD film will re-radiate some of this absorbed energy as longer-wavelength infrared. This is what we call ‘secondary heatflow’. We will reflect this heat away from the building interior using a low-emittance (low-e) layer in the IGU (see later).

When we apply an electric voltage to the ITO conductors, the SPD particles align with the electric field, allowing light to pass through and turning the glass clear.

Splitting the IGU Design

• We will first specify the ‘SPDCore’ as an ‘Interlayer’ in LBNL OPTICS, then design the enclosing components into a ‘Laminate’.
• We then import the Laminate into LBNL WINDOW where we add the vacuum gap, low-e layer, and the Glass #C layer, to give the complete IGU.

IGU Tech Spec

Here is the full technical specification of our IGU that we will model in the LBNL toolchain:

PVB or EVA or SGP: Which is better?

PVB, EVA and SGP are clear plastic films we use to laminate smart glass for safety and durability.

• PVB (Polyvinyl Butyral): standard for outdoor buildings because it blocks UV and sound very efficiently. 
• EVA (Ethylene Vinyl Acetate): better for wet areas because it resists water. However, it suffers from a yellow discoloration under sunlight, so it is often used in interiors.
• SGP (SentryGlas® Ionoplast): superior to PVB and EVA in durability, rigidity and safety, and ideal for extreme weather applications.

In this case, we have chosen PVB, because EVA and SGP were not available in the database.

Modeling Workflow: Our 5-Step Guide

Our IGU modeling process follows the following steps:

1. Prepare smart glass spectral data
2. Import spectral data into LBNL OPTICS as an Interlayer
3. Design the Laminate in LBNL OPTICS
4. Import Laminate into LBNL WINDOW
5. Model IGU thermal and optical properties in LBNL WINDOW

Let’s dive in:

1. Prepare Smart Glass Spectral Data

Please refer to the instructions for the allowed data format in the IGDB Data File Format document.

Spectral Graph

First, we need to capture the spectral data as a Tab-Separated-Values (TSV) file, one for each ‘state’ of the glass. In our case, we limit our states to ‘ON’ (transparent) and ‘OFF’ (opaque) to keep the model simple.

The spectral data shows how Visible Light Transmittance (VLT) on the Y-axis varies with wavelength on the X-axis for two cases: 

• Blue line: voltage OFF (SPD tinted)
• Red line: voltage ON (SPD clear)

In the graph, the red line shows the glass when powered, indicating an average VLT of about 65%. The flat characteristic across the visible waveband (380-780 nm) shows that the SPD film does not add any extra coloration to light transmitted through it.

In the unpowered state (the blue line), the VLT averages around 3-5% across most of the visible waveband, but shows a blueish tint. We can see this from the blue spectral peak at 420 nm wavelength. 

Spectral Data File

We used GeminiPro to create a TSV data file from this graph, interpolated to 5 nm intervals, and extrapolated to cover the full wavelength range required by OPTICS, namely 300 to 2,500 nm.

Finally, we prepended the data file with the required headers, identifying the units, thickness, conductivity, infrared transmittance, product range and so on. 

The 4 column headers are not marked in the file but correspond to the following (an excerpt is shown here for brevity):

1. Wavelength (microns)
2. Transmittance (from 0 to 1)
3. Front Reflectance (from 0 to 1)
4. Back Reflectance (from 0 to 1)

2. Import Spectral Data into LBNL OPTICS as Interlayer

Next, we import the spectral data into OPTICS as an ‘Interlayer’.

SPD Interlayer (OFF)

This OPTICS screenshot shows the glass in its OFF state (tinted). You can see that the visible waveband (from 380 nm to 780 nm) exhibits low transmittance (the dip in the yellow line), indicating a tinted state when OFF.

SPD smart glass interlayer in opaque state (OFF)

The ‘Spectral Properties’ section shows the following key data.

Solar Transmittance

This represents how much raw energy from the sun is transmitted by the glass and includes UV, visible light and near-infrared (NIR).

The software multiplies the raw radiometric data with a standard Solar Spectral Irradiance Distribution (such as the NFRC/ISO global solar spectrum curve).

It then divides it by the total energy of the reference sun, resulting in a flat, radiometric percentage of solar heat transmission.

Under ‘Spectral Properties’, the field marked “solar, T” indicates 0.222 which means that only 22.2% of total direct solar radiation passes through the smart glass.

The rest (77.8%) is blocked by the glass, reducing HVAC costs.

Photopic Transmittance

Whereas the ‘Solar’ section is used for thermal calculations, the ‘photopic’ (visible) section highlights daylight performance and human comfort, and measures what the human eye can see.

OPTICS does this by applying a mathematical filter (called the ‘photopic response curve’) to the raw spectral data to mimic how the human eye sees sunlight. 

For example, the human eye is more sensitive to green light than to red light.

This allows OPTICS to calculate how bright the room will look to an occupant.

From the “photopic, T” number, shown as 0.033, we see that the smart glass will allow only 3.3% of human-visible sunlight to pass through in its dark state. 

This reduces visual glare, improving occupant productivity and comfort.

SPD Interlayer (ON)

In the following image, the SPD smart glass is in its transparent state and averages about 60% VLT across the visible band (the highest yellow peak).

SPD smart glass interlayer in clear state (ON)

The Spectral Properties are now quite different in its powered state:

• “solar, T” is now about 58%, allowing more heat into the building
• “photopic, T” is about 63%, allowing more daylight into the building

This indicates that the smart glass will transmit heat and light whenever we switch it on, which can be useful for cooler climates or darker days. 

This reduces artificial lighting costs.

3. Design the Laminate in LBNL OPTICS

Next we encapsulate the SPDCore ‘Interlayer’ with PVB and glass on both sides to give us a ‘Laminate’ in OPTICS.

The following image shows the Laminate in its opaque state, with the layers numbered as follows:

• #1: Glass
• #2: PVB
• #3: SPDCore
• #4: PVB (not visible in the image)
• #5: Glass (not visible in the image)

SPD Laminate in opaque state (OFF)

• “solar, T” is 12.5%
• “photopic, T” is 2.8%

The tab called ‘Schematic’ shows the structure of the Laminate more clearly (the numbers have been rounded up by OPTICS):

• 4.85 mm Glass (on the left)
• 0.76 mm PVB
• 0.40 mm SPDCore interlayer (shown asa dark layer)
• 0.76 mm PVB
• 4.85 mm Glass on the right
• Total thickness = 11.62 mm

The following image shows the same Laminate in its clear state, with the SPDCore layer now shown in a grey color, and with the following spectral characteristics:

• “solar, T” is 40%
• “photopic, T” is 55%

SPD Laminate in clear state (ON)

4. Import Laminate into LBNL WINDOW

We now import the Laminate from OPTICS into the WINDOW Glass library, which shows key metrics in the OFF state:

• Solar Transmittance (front) = 0.125, i.e. the SPD Laminate transmits 12.5% of solar radiation into the building
• Visible Transmittance (Front) = 0.028 i.e. the SPD Laminate transmits 2.8% of VLT into the building

WINDOW: SPD Laminate in opaque state (OFF)

And the corresponding data for the SPD Laminate in its clear state:

• Solar Transmittance (front) = 0.401, i.e. the SPD Laminate transmits about 40% of solar radiation into the building
• Visible Transmittance (Front) = 0.549 i.e. the SPD Laminate transmits about 55% of visible light into the building

WINDOW: SPD Laminate in clear state (ON)

Here, we are looking at the ‘Laminate’ in the WINDOW Glass Library. Next, we need to create the IGU itself in the WINDOW ‘Glazing System’ library.

5. Model IGU Thermal and Optical Properties in LBNL WINDOW

Now finally, we can create the IGU itself in WINDOW by joining the Laminate with a Vacuum Gap layer, a Low-E layer and the inner Glass layer.

The result looks like this for the IGU in its tinted state. The low-e layer appears as a dotted line on the inner surface of the inner glass panel.

WINDOW: SPD IGU in opaque state (OFF)

WINDOW calculates the following parameters for the whole IGU:

• Centre-of-Glass (COG) U-Factor = 1.146 W/m2.K
• Solar Heat Gain Coefficient (SHGC) = 0.174
• Visible Light Transmittance (Tvis) = 0.018

This shows that our triple-glazed SPD smart glass with vacuum gap and low-e layer gives us an energy-blocking glass wall that reduces glare in its tinted state.

Now for the clear state:

WINDOW: SPD IGU in clear state (ON)

In its activated (clear) state, the IGU has the following parameters:

• Centre-of-Glass (COG) U-Factor = 1.146 W/m2.K (same as before)
• Solar Heat Gain Coefficient (SHGC) = 0.380
• Visible Light Transmittance (Tvis) = 0.343, i.e. 34% VLT when clear

This confirms that we have successfully designed a smart glass IGU that would meet stringent energy codes, minimising heat gain and maximising daylighting.

This reduces energy consumption and artificial lighting costs, while enabling daylighting whenever needed.

Summary and Industry Applications

Smart glass is not a monolithic product but a complex, multilayered system. 

As such, it requires a rigorous, physics-based analysis of the entire stack in order to simulate its thermal and optical behavior.

By leveraging simulation tools like LBNL OPTICS and WINDOW, AEC professionals can optimize IGU designs for specific project requirements.

The workflow outlined in this article moves beyond simple “ballpark” estimates, providing the confidence needed to finalize glazing specifications before fabrication or procurement.

With advanced simulation, we ensure that projects deliver thermal efficiency targets, sustainable daylighting and occupant comfort, resulting in responsive, high-performance building skins.

Technical FAQ

Why does LBNL WINDOW model Front and Back Transmittance?

We often think of light transmittance as passing through glass from one side to the other. However, it is not that simple. 

Smart glass often has coatings placed on only one side. Additionally, light can reflect internally within an IGU, due to multiple layers and diverse coatings, as we have seen in this article.

Transmittance must therefore be tracked from both directions.

Design Asymmetry

Low-emissivity (low-e) coatings and electrochromic layers are ultra-thin chemical depositions located on only one side of a glass substrate.

• When incident light hits the coated side first, it reflects some light, then absorbs or transmits the rest.
• When light enters from the uncoated side, it travels through the absorbing glass first, before hitting the coating on the other side.

This alters the measured transmittance, such that Front Transmittance can differ substantially from Back Transmittance.

Matrix Calculations

Both OPTICS and WINDOW are based on ISO 15099. This determines how they calculate a glazing system, and depends on light that can bounce repeatedly between multiple layers inside the smart glass.

To model these internal reflections accurately, OPTICS and WINDOW need to model each pathway:

• Front Transmittance
• Back Transmittance
• Front Reflectance
• Back Reflectance

What exactly is Infrared Radiation?

The entire solar spectrum looks like this, divided as follows:

• Ultraviolet (UV): 100 nm to 380 nm and plit into UVC, UVB and UVA
• Visible light: 380 nm to 780 nm
• Near Infrared (NIR): 780 nm to 2500 nm
• Far Infrared: 2500 nm to 1,000,000 nm (1 mm)

Infrared represents a large span of solar radiation. We split it into two main regions:

Near-Infrared (NIR): 

This spans from 780 nm to 2,500 nm and corresponds to the heat coming directly from the sun. This is captured in the ‘Solar’ section of the WINDOW software.

Far-Infrared / Thermal Infrared: 

This region spans from 2,500 nm to 1,000,000 nm (although WINDOW focusses on the 5,000 nm to 50,000 nm band only). This is radiant heat emitted by room-temperature objects, and is captured in the ‘IR (Emissivity)’ section of OPTICS and WINDOW.

How does LBNL WINDOW use Far-Infrared Data?

WINDOW simulates heat flow under steady-state conditions and calculates thermal resistance using three mechanisms:

Surface Radiation Exchange

In a multi-pane glass unit, the panes do not just conduct heat through the gaps. They radiate energy across the gap to each other. WINDOW uses the Far-Infrared Emissivity of each surface to calculate the radiation heat transfer between layers. Lower emissivity means less heat is radiated, which lowers the U-factor.

Iterative Heat Balance

Heat which passes through the window creates distinct temperatures on each surface. Because radiation depends on surface temperature, WINDOW establishes a heat balance equation for each layer and runs an iterative mathematical solver. It recalculates the thermal radiation exchange and surface temperatures until the energy balances.

Boundary Air Films

The software relies on Far-Infrared data to figure out how much heat the outer glass layer radiates to the cold sky, and how much heat the inner glass layer absorbs from warm walls and warm interior furnishings.

Why did we model the ‘SPDCore’ as an Interlayer?

Our original attempt was to import our data as a 3mm thick glass sample, with the SPDCore film inside it. 

This distorted our calculations by overestimating energy absorption, lowering VLT and skewing the weight and thickness of the assembly.

When we laminated this layer inside WINDOW, the software assumed a 3mm glass substrate as a physical part of the layer. If we then sandwich that layer between two 5mm low-iron sheets with PVB, WINDOW calculates a single, massive 13mm thick composite pane.

In other words, the 3mm glass of our sample, with its high iron content and solar absorption, was penalizing our optical calculations.

The better approach, and the reason for re-running the data file import as an ‘Interlayer’, was to isolate the ‘SPDCore’ as a standalone layer within OPTICS.

To achieve this, we built a data file with spectral measurements of only the PET+ITO+SPD+ITO+PET stack, measured as a free-standing film. 

In our OPTICS script, we modified the headers to reflect that this is an ‘Interlayer’, not a monolithic glass pane.