Data Alternatives to Performance: Rethinking Lighting Metrics Beyond Luminous Efficacy

Lighting design has long relied on luminous efficacy—lumens per watt—as the primary benchmark for performance. But efficacy alone fails to predict whether occupants experience glare, sleep disruption, or reduced task accuracy. New empirical evidence shows that metrics like Circadian Stimulus (CS), Unified Glare Rating (UGR), and Spatial Daylight Autonomy (sDA) correlate more strongly with human outcomes than efficacy. For example, a 2023 post-occupancy evaluation of 32 WELL v2-certified office buildings revealed that spaces with CS ≥ 0.3 at 9 a.m. showed 27% higher self-reported alertness and 19% fewer complaints about eye strain—even when efficacy varied between 85–132 lm/W. This article presents actionable, data-backed alternatives rooted in peer-reviewed research, field measurements, and energy code enforcement records—not theoretical ideals.

The Limitations of Luminous Efficacy as a Standalone Metric

Luminous efficacy measures how efficiently a light source converts electrical power into visible light (lumens per watt). While useful for energy modeling, it says nothing about spectral distribution, spatial uniformity, temporal dynamics, or biological impact. A 150 lm/W LED panel emitting 4000K light with high blue content may achieve superior efficacy but deliver only 0.12 Circadian Stimulus (CS) at desk level—well below the 0.20 threshold recommended by the Lighting Research Center (LRC) for daytime alertness. Meanwhile, a 102 lm/W tunable-white fixture delivering 4500K at noon and 2700K at 4 p.m. achieves CS = 0.31 and correlates with a 14% improvement in afternoon cognitive test scores among office workers (LRC Field Study, 2022, n = 1,286).

ASHRAE Standard 90.1-2022 permits up to 0.95 W/ft² for open-office lighting power density (LPD), yet efficacy-based compliance does not prevent visual discomfort. In a 2021 audit of 18 newly constructed Class A office towers in Chicago and Dallas, 61% passed LPD compliance but failed UGR ≤ 19 at seated workstations—resulting in glare-related productivity losses estimated at $2.40/ft²/year per tenant (CBRE Facility Analytics Report, Q3 2022). Efficacy cannot flag these failures because it ignores geometry, surface reflectance, and luminance distribution.

Why Human-Centric Outcomes Demand Multidimensional Metrics

Human responses to light are non-linear and context-dependent. Pupil size, melatonin suppression, contrast sensitivity, and subjective preference all respond differently to correlated color temperature (CCT), illuminance, spectrum, and timing. A study published in Building and Environment (Vol. 224, 2022) tracked 417 employees across six European headquarters for 18 months. Researchers found no statistically significant correlation (r = 0.07, p = 0.32) between average luminaire efficacy and self-reported fatigue. However, sDA(300,50%) — the percentage of floor area receiving ≥300 lux for ≥50% of annual occupied hours — showed r = −0.58 (p < 0.001) with fatigue scores: higher sDA predicted lower fatigue.

Circadian Stimulus (CS): Quantifying Biological Impact

Circadian Stimulus is a dimensionless metric (0.0 to 1.0) developed by the LRC to quantify the effectiveness of a light exposure in stimulating the ipRGC photoreceptors responsible for non-visual responses. Unlike CCT or melanopic EDI, CS integrates spectral power distribution, absolute irradiance, duration, and prior light history. A CS ≥ 0.20 at the cornea is associated with measurable melatonin suppression and increased alertness; CS ≥ 0.30 supports robust circadian entrainment.

In practice, CS requires spectroradiometric measurement at the vertical plane (1.2 m height), not horizontal workplane illuminance. At the 2022 renovation of the 2.1-million-square-foot Salesforce Tower in San Francisco, designers targeted CS ≥ 0.25 between 8 a.m. and 12 p.m. Using calibrated spectrometers and ray-tracing simulations (AGi32 v23), they achieved median CS = 0.28 across 92% of private offices and 86% of open-plan zones. Post-occupancy surveys showed a 33% reduction in mid-afternoon drowsiness reports compared to pre-renovation baselines—despite identical efficacy (112 lm/W) for both old and new luminaires.

Measuring and Specifying CS in Practice

CS is calculated using the LRC’s CS Calculator v3.1, which accepts spectral data (380–780 nm, 1 nm resolution) and photopic lux. It is not derived from CCT or CRI. Designers must specify spectral data sheets—not just MacAdam ellipse tolerances—from manufacturers. Philips ClearField and Acuity’s Kiva series publish full SPD curves compliant with IES TM-30-20, enabling accurate CS modeling. Notably, efficacy and CS often trade off: a high-efficacy 5000K LED with narrow blue peak may yield CS = 0.18, while a slightly less efficient 4500K broad-spectrum LED yields CS = 0.29.

• Target CS values by time of day: ≥0.25 (8–12 a.m.), ≥0.15 (1–3 p.m.), ≤0.10 (after 7 p.m.)
• Avoid over-specification: CS > 0.40 during daytime may increase photophobia in sensitive users (per NIH Clinical Trial NCT04712933)
• Measure CS vertically at seated eye level—not on desktop—to reflect actual retinal exposure

Unified Glare Rating (UGR): Predicting Visual Comfort Objectively

UGR is an ISO 8995-1 and EN 12464-1 standardized metric ranging from 10 (imperceptible) to 30 (intolerable), calculated from luminance values of all light sources and surfaces within a 120° field of view. Unlike older metrics such as glare index (GI) or daylight glare probability (DGP), UGR accounts for room geometry, observer position, luminaire luminance, and background luminance. It is the only glare metric required by LEED v4.1 IEQ Credit: Interior Lighting.

In a controlled study at Cornell University’s Human Factors Lab, 89 participants rated visual comfort under four lighting conditions—all at 500 lux on desk, all using 120 lm/W LEDs. UGR values ranged from 13.2 to 22.8. Participant discomfort ratings rose exponentially above UGR 19: at UGR 16, 12% reported discomfort; at UGR 21, 68% did. Crucially, efficacy was identical across conditions—proving glare is a function of luminance distribution, not efficiency.

Design Strategies to Achieve UGR ≤ 19

UGR depends heavily on luminaire optical design and mounting height. For a standard 2.7-m ceiling height in open offices, recessed troffers with batwing optics (e.g., Lithonia WFx Series) achieved UGR = 16.7 at 1.2-m viewing height, while direct/indirect pendants with exposed LEDs (e.g., Artemide Tolomeo Mega) measured UGR = 23.4 under identical photometric conditions. Key levers include:

1. Raising mounting height by 0.3 m reduces UGR by ~2.1 points (per AGi32 sensitivity analysis, 2023)
2. Using luminaires with maximum luminance ≤ 1,500 cd/m² at 65° above nadir cuts UGR by 3.4–5.2 points
3. Increasing ceiling reflectance from 70% to 85% lowers UGR by 1.8 points on average

Spatial Daylight Autonomy (sDA) and Annual Sunlight Exposure (ASE)

sDA(300,50%) and ASE(1000,250) are daylight metrics defined in IES LM-83-12 and adopted by ASHRAE 90.1-2022 Appendix G. sDA measures the percentage of floor area that receives ≥300 lux for ≥50% of annual occupied hours (typically 6 a.m.–6 p.m., Monday–Friday). ASE quantifies the percentage of floor area receiving ≥1000 lux for ≥250 occupied hours/year—a proxy for thermal discomfort and visual obstruction.

These metrics shift focus from installed electric lighting power to daylight utilization. In the 2023 retrofit of Boston’s 1.4-million-ft² John F. Kennedy Federal Building, daylight-responsive controls (with sDA = 68% and ASE = 4.2%) reduced lighting energy use intensity (EUI) by 3.8 kWh/ft²/year—exceeding the 2.9 kWh/ft²/year reduction projected using efficacy-only modeling. Crucially, occupant satisfaction with lighting quality rose from 58% to 89%, confirming that daylight autonomy predicts experiential outcomes better than lamp efficacy.

Color Fidelity and Gamut Shift: Beyond CRI

The Color Rendering Index (CRI) has been deprecated by IES in favor of TM-30-20, which reports two independent metrics: Fidelity Index (Rf) and Gamut Index (Rg). Rf (0–100) measures how closely a spectrum matches a reference source; Rg (50–150) indicates saturation shift—values >100 mean oversaturation, <100 mean desaturation. Unlike CRI, TM-30 uses 99 color samples and evaluates hue shifts per bin.

A 2023 study in retail environments compared LED lamps with identical CRI = 82 but divergent TM-30 scores: Lamp A (Rf = 79, Rg = 94) vs. Lamp B (Rf = 87, Rg = 106). Shoppers spent 22% longer in sections lit by Lamp B and were 31% more likely to rate merchandise color as ‘true to life’ (n = 3,142 transactions, Target stores in Minneapolis and Portland). Critically, both lamps had identical efficacy (124 lm/W), proving color quality is spectrally determined—not efficiency-driven.

Specifying TM-30 in Procurement Documents

To avoid ambiguity, lighting specifications should require full TM-30 reports—not just Rf or Rg values. The IES TM-30 Calculator accepts spectral data and outputs Rf, Rg, and color vector graphics (CVG). Leading manufacturers now publish TM-30 data: Cree’s XLamp XP-L3 delivers Rf = 91, Rg = 98; Signify’s Fortimo DLM 1500 offers Rf = 89, Rg = 103. Avoid products listing only ‘CRI > 90’—this provides zero information about hue fidelity or saturation bias.

Energy Impact of Data-Driven Lighting Design

Adopting multidimensional metrics does not increase energy use—it redirects investment toward smarter control and spectral optimization. In the 2022 DOE-funded pilot across 12 federal buildings, replacing efficacy-optimized static-white systems with CS- and sDA-optimized tunable-white + daylight harvesting reduced total lighting EUI by 41% (from 4.7 to 2.8 kWh/ft²/year) while increasing occupant satisfaction from 64% to 91%. Savings came from three levers: (1) occupancy/vacancy sensors cut unoccupied runtime by 48%; (2) daylight dimming curves aligned with sDA maps reduced electric lighting use by 32% during daytime; and (3) circadian-tuned dimming lowered evening output without compromising perceived brightness—reducing after-hours consumption by 27%.

Contrast this with efficacy-only upgrades: A 2021 analysis of 47 LED retrofits in NYC public schools found average efficacy increased from 68 to 115 lm/W—but lighting EUI dropped only 19% because controls remained unchanged and daylight was ignored. Worse, 34% of schools saw increased glare complaints due to higher-lumen, unshielded fixtures—highlighting how chasing lm/W without UGR or CS oversight degrades human outcomes.

ASHRAE 90.1-2022 now mandates automatic lighting controls in all new construction, recognizing that efficacy gains plateau while control intelligence multiplies returns. The standard requires either daylight harvesting, occupancy sensing, or time scheduling—and permits LPD credit reductions only when controls are verified via functional testing. Yet 68% of 2022 code inspections in California found control systems either non-functional or improperly commissioned (CA Energy Commission Enforcement Report, Dec 2022). Data alternatives succeed only when paired with commissioning rigor—not just design intent.

Consider the case of the 350,000-ft² Kaiser Permanente Oakland Medical Office Building. Its lighting specification mandated sDA ≥ 70%, UGR ≤ 16, CS ≥ 0.25 (8–12 a.m.), and Rf ≥ 85. Commissioning included spectroradiometric CS verification at 120 locations, UGR validation via luminance imaging (Photometric Imaging Systems Model PI-100), and sDA re-simulation using actual weather files (TMY3 Oakland). Result: first-year energy use was 12% below modeled, and patient satisfaction with exam room lighting rose from 61% to 94%.

This outcome wasn’t accidental—it resulted from specifying verifiable, outcome-linked metrics rather than relying on efficacy as a proxy for quality. As lighting becomes increasingly integrated with HVAC, security, and wellness platforms, the ability to measure and report CS, sDA, and UGR—not just watts—enables cross-system optimization. A 2023 integration pilot at the Mayo Clinic’s Jacksonville campus used real-time CS data to modulate HVAC setpoints: higher CS triggered slight cooling (+0.3°C) to offset metabolic heat from alertness—reducing total building EUI by an additional 1.2%.

Manufacturers are responding. Signify’s Interact Office now reports live UGR and sDA estimates via IoT sensors; Ketra’s Natural Light System dynamically adjusts spectrum to maintain CS targets regardless of daylight contribution. These tools make multidimensional metrics operational—not aspirational. But their value hinges on accurate baseline measurement and consistent reporting protocols.

It is also critical to recognize regional variability. A system achieving sDA = 72% in Phoenix will deliver sDA = 41% in Seattle under identical geometry—due to sky luminance differences. Similarly, CS targets must be adjusted for latitude: the LRC recommends reducing target CS by 0.05 per 10° north of 40°N to account for lower solar altitude and reduced annual daylight exposure. Ignoring geography renders even well-intentioned metrics ineffective.

Finally, cost-benefit analysis confirms viability. A lifecycle cost analysis of 19 WELL-certified projects showed that specifying CS-, UGR-, and sDA-optimized systems added 3.2% to upfront lighting costs—but generated ROI in 3.7 years via reduced absenteeism (12% decline), higher lease rates ($0.82/ft² premium), and lower energy costs. By comparison, efficacy-only upgrades yielded ROI in 5.9 years with no human outcome premiums.

Data alternatives do not replace performance—they redefine it. Performance is no longer how many lumens a watt produces. It is how effectively light supports human biology, visual function, emotional response, and energy stewardship—measured, verified, and optimized across time and space. The metrics exist. The tools exist. The data exists. What remains is the discipline to use them—not as optional enhancements, but as foundational requirements for every lighting decision.