Proposal and practicality of an alternative blue-light hazard risk assessment method for high-intensity white-light sources at workplaces

Abstract

Objectives. It is prescribed to determine blue-light hazard (BLH) weighted radiances, $L_{\text{B}}$, for an assessment of spotlights with an angular subtense $\alpha \ge 11\text{mrad}$. The BLH weighted irradiance, $E_{\text{B}}$, can be used alternatively for smaller sources. Appropriate instruments are not common among persons commissioned with risk assessment (RA), and especially $L_{\text{B}}$ measurements may be challenging. Therefore, a practical BLH RA approach is proposed that is based on illuminance, $E_{\text{v}}$, pre-calculated blackbody BLH efficacies of luminous radiation, $K_{\text{B,v}}^{\text{Planck}}$, and solid angle considerations. Methods. The practicality of this method was examined and compared against other RA approaches. Results. To ensure comparability of the applied instruments, measurements were performed close to a radiance standard, showing deviations within the lamp’s expanded uncertainties ($<4\mathrm{\%}$), whereas the deviations were $\pm 15\mathrm{\%}$ for longer distances. Focusing on a complex light-emitting diode (LED) spotlight, all detected values could be converted to $L_{\text{B}}$ by means of the RA methods within $\pm 20\mathrm{\%}$. Two field tests with several spotlights yielded maximum permissible exposure durations (MPED) obtained from the different RA approaches that agreed among each other within uncertainties largely below $\pm 30\mathrm{\%}$. Conclusion. The general practicality of the proposed $E_{\text{v}}$ method can be concluded for a workplace BLH RA of white-light sources.

Keywords:

• blue-light hazard
• risk assessment
• occupational safety and health
• instrument comparison
• blue-light hazard efficacy of luminous radiation
• workplace
• spotlight

1. Introduction

The human eye plays a key role in the perception of our world because about $80\mathrm{\%}$ of the surrounding’s information is gathered by the visual sense. It is therefore not surprising that artificial light is omnipresent at workplaces day and night, e.g., at the office in the form of screens (laptops, tablets, smartphones) or lighting. Such light sources do not pose any risk for photochemically induced acute retinal damage [1], typically referred to as blue-light hazard (BLH). General lighting based on light-emitting diodes (LEDs) can also be regarded as BLH safe [2–4].

In contrast to the aforementioned low-power light sources, very bright spotlights are necessary for structural and civil engineering, for large storage areas or in theaters, news studios and on stages, which might pose a BLH risk for workers [5,6]. If it is not undoubtedly clear whether the BLH exposure limits (EL) are either well exceeded or complied with at the workplace, the emitted light must be assessed metrologically. The International Commission on Non-Ionizing Radiation Protection (ICNIRP) published recommendations for EL in their ‘Guidelines on limits of exposure to incoherent visible and infrared radiation’ [7]. Therein, it is specified under which conditions depending on the angular subtense of the source, $\alpha$, a risk assessment (RA) can be based on the BLH weighted irradiance, $E_{\text{B}}$, or requires knowledge about the BLH weighted radiance, $L_{\text{B}}$.

The determination of $L_{\text{B}}$ may be challenging, and appropriate devices are expensive. Radiance meters are error-prone because of their high field-of-view (FOV) sensitivity, so experienced operators are needed to achieve reliable and reproducible results. Fiorentin and Scroccaro [8] write that ‘it requires pretty sophisticated instruments to measure the involved quantities’ (792); thus, alternative and practical methods are preferable to lower the hurdles for a BLH RA. A common way to circumvent radiance measurements is the conversion of $E_{\text{B}}$ to $L_{\text{B}}$ by means of the source’s solid angle, Ω, seen at the position of the detector. This procedure is standardized in ‘The Lamp Safety Standard’ IEC 62471:2006 [9] and will be described in more detail later.

Several research groups also attempted to find appropriate solutions. By relating $E_{\text{B}}$ to the illuminance, $E_{\text{v}}$, of about 100 lamps, Schulmeister et al. [10] derived a transformation factor in order to express the BLH EL in terms of illuminance. The permitted $E_{\text{v}}$ values were presented as a function of correlated color temperature (CCT). They concluded that at the ‘$500\text{lx}$ distance’ the BLH EL according to the ICNIRP [7], the International Electrotechnical Commission (IEC) [9] and the American Conference of Governmental Industrial Hygienists (ACGIH) [11] are not exceeded, except for very high CCTs.

Shibuya et al. [12,13] applied a hyperspectral camera to examine the effect of non-uniform luminance, $L_{\text{v}}$, distributions on BLH risk levels. They assessed the light emissions of LED luminaires, incandescent and fluorescent lamps at the standardized ‘$500\text{lx}$ distance’ [9], demonstrating the usefulness of two-dimensional light profiles. Liu et al. [14] pursued a similar method combining the measured relative spectral power distribution with the maximum $L_{\text{v}}$ value determined from a camera for LEDs and compact fluorescent lamps. However, such imaging luminance meters are even more expensive than radiance instruments, their handling is also demanding and they cannot be used at workplace-typical large distances.

Other RA approaches applied the BLH efficacy of luminous radiation, $K_{\text{B,v}}$, i.e., the ratio of $E_{\text{B}}$ to $E_{\text{v}}$ or $L_{\text{B}}$ to $L_{\text{v}}$. Fantozzi et al. [15] describe a simplified BLH RA method according to IEC TR 62778:2014 [16] based on CCT-dependent photometric threshold curves between risk group (RG) 1 and RG 2. By means of $K_{\text{B,v}}$, the $L_{\text{v}}$ and $E_{\text{v}}$ values measured at a distance of $20\text{cm}$ can be applied for a quick RG check: if both values are smaller than the threshold curves, the lamp is classified as RG 1 at most, thus posing no photobiological risk. Otherwise, two different quantities must be determined. After the present study had been completed, the new revision of IEC 62471-7:2023 was published [17] and is currently discussed to include the threshold curves from IEC TR 62778:2014 as an informative appendix.

A mere analytical approach parametrized the light emission spectra of phosphor-converting white LEDs (pc-wLEDs) via a linear combination of two Gaussian functions [18]. A comparison of calculated with experimentally determined $K_{\text{B,v}}$ values revealed percentage deviations better than $10\mathrm{\%}$. Although this finding demonstrates its general applicability, it remains unclear how to transfer results derived from normalized single pc-wLEDs to an actual workplace situation. The so-called ‘Planck Approximation’ [19] is another RA method using $K_{\text{B,v}}$. It is based on a CCT-dependent pre-calculated $K_{\text{B,v}}$ value for a blackbody radiator, $K_{\text{B,v}}^{\text{Planck}}$, in conjunction with a single $E_{\text{v}}$ measurement at the exposure site. Considering $\alpha$ (assuming a circular extent of the source), the intended FOV and the BLH EL, three simple equations for the maximum permissible exposure duration (MPED; the exposure duration resulting in an exposure level equal to the EL) were derived.

The aim of this work is to evaluate the uncertainty and practicality of the Planck Approximation with regard to other RA approaches within the Partnership for European Research in Occupational Safety and Health (PEROSH) project ‘High-Power Spotlights Risk Assessment’ (HiPoSisAs). For a thorough analysis, the inter-comparability of the HiPoSisAs partners’ instruments was tested by means of a calibrated radiance standard. The applied RA approaches were evaluated by more demanding measurement conditions for a multi-chip RGBW spotlight. Field tests completed this evaluation.

2. Materials and methods

2.1. Basic equations for BLH risk assessment

Most photobiological effects are wavelength dependent and can be described by an action spectrum that is used for spectral weighting. Regarding photochemically induced retinal damage, $E_{\text{B}}$ is determined via the convolution of the source’s spectral irradiance, $E(\lambda )$, with the BLH action spectrum, $B(\lambda )$, in the wavelength range $300-700\text{nm}$ [7]: (1) $E_{\text{B}}={\int }_{300\text{nm}}^{700\text{nm}}E(\lambda )B(\lambda )\text{d}\lambda [E_{\text{B}}]=\text{W}{\text{m}}^{-2}.$(1) The convolution of $E(\lambda )$ with the photopic luminous efficiency function, $V(\lambda )$, that reflects the brightness perception of the human eye [20], results in $E_{\text{v}}$: (2) $E_{\text{v}}=K_{\text{m}}{\int }_{360\text{nm}}^{830\text{nm}}E(\lambda )V(\lambda )\text{d}\lambda [E_{\text{v}}]=\text{lm}{\text{m}}^{-2}=\text{lx}.$(2) The constant $K_{\text{m}}=683\mathrm{lm}{\text{W}}^{-1}$ represents the maximum spectral luminous efficacy. Replacing $E(\lambda )$ by $L(\lambda )$ in Equations (1) and (2) yields $L_{\text{B}}$ and $L_{\text{v}}$, given as $\text{W}{{\text{m}}^{-}}^2\text{s}{{\text{r}}^{-}}^1$ (watt per square meter per steradian) and as $\text{lm s}{\text{r}}^{-1}{\text{m}}^{-2}=\text{cd}{\text{m}}^{-2}$ (lumen per steradian per square meter = candela per square meter), respectively.

For a BLH RA of light sources with an angular subtense $\alpha \ge 11\text{mrad}$, it is necessary to determine $L_{\text{B}}$ based on the solid angle. The Ω value is calculated by the ratio of a spherical calotte with area $A_{\text{sc}}$ and the sphere’s radius d. A common approximation is the use of a plane circular area with diameter D instead of $A_{\text{sc}}$. The Ω value can also be described via the resulting cone’s opening angle $\gamma$: (3) $\begin{array}{rl}\mathrm{\Omega }& =2\pi \left[1-\cos \left(\frac{\gamma }{2}\right)\right]=\frac{A_{\text{sc}}}{d^2}\approx \frac{\pi }{4}{\left(\frac{D}{d}\right)}^2\\ [\mathrm{\Omega }]& ={\text{m}}^2{\text{m}}^{-2}=\text{sr}.\end{array}$(3) Although the solid angle is a dimensionless quantity, it is usually expressed in units of steradian. The relation between radiance and irradiance follows as: (4) $L(\lambda )={\mathrm{\Omega }}^{-1}E(\lambda )[L]=\text{W }{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}.$(4) The ratio of $L_{\text{B}}$ and $L_{\text{v}}$, or equally of $E_{\text{B}}$ and $E_{\text{v}}$, is a common quantity to characterize white-light sources [16], and the resulting $K_{\text{B,v}}$ is usually presented as a function of CCT, $T_{\text{cp}}$: (5) $K_{\text{B,v}}(T_{\text{cp}})=\frac{L_{\text{B}}}{L_{\text{v}}}=\frac{E_{\text{B}}}{E_{\text{v}}}.$(5) The Planckian radiator’s $K_{\text{B,v}}^{\text{Planck}}$ value is calculated the same way and can be described empirically via an S-shaped four-parameter logistic function [19]: (6) $K_{\text{B,v}}^{\text{Planck}}=K_{\mathrm{\infty }}{\left[1+{\left(\frac{T_0}{T}\right)}^h\right]}^{-s},$(6) with saturation value $K_{\mathrm{\infty }}=2.289\pm 0.001\text{W}\text{kl}{\text{m}}^{-1}$ (watt per kilo lumen), with temperature $T_0=1573\pm 32\text{K}$ for that the curvature changes direction or sign, with hill slope $h=1.307\pm 0.003$ (steepness of the curve) and asymmetry parameter $s=5.3\pm 0.1$ (with respect to $T_0$, $s=1$ corresponds to symmetry). Note that $K_{\text{B,v}}^{\text{Planck}}$ is constant for any fixed CCT; thus, it can be pre-calculated (see Appendix 1) and directly used on site at workplaces.

2.2. Exposure limits and averaging angles

The BLH EL according to the ICNIRP [7] are expressed in terms of weighted quantities and depend on the exposure duration, t, as well as on the angular subtense of the source, $\alpha$: (7a) $\begin{array}{rl}{L}_{\text{B}}^{\text{EL}}& =\{\begin{array}{ll}\text{1 MJ }{\text{m}}^{-2}\text{s}{\text{r}}^{-1}{t}^{-1}& 0.25\text{s}\le t\le 10,000\text{s}\\ \text{100 W}{\text{m}}^{-2}\text{s}{\text{r}}^{-1}& 10,000\text{s}<t<30,000\text{s}\end{array}\\ & \alpha \ge 11\text{mrad,}\end{array}$(7a) (7b) $\begin{array}{rl}{E}_{\text{B}}^{\text{EL}}& =\{\begin{array}{ll}\text{100 J}{\text{m}}^{-2}{t}^{-1}& 0.25\text{s}\le t<100\text{s}\\ \text{1 W}{\text{m}}^{-2}& 100\text{s}\le t<30,000\text{s}\end{array}\alpha <11\text{mrad}.\end{array}$(7b) For sources with $\alpha \ge 11\text{mrad}$, the EL are given as BLH weighted radiances whereas the alternative irradiance limit may be determined for smaller sources. $L_{\text{B}}^{\text{EL}}$ and $E_{\text{B}}^{\text{EL}}$ are time-dependent for exposure durations shorter than 10,000 and $100\text{s}$, respectively; otherwise, they are constant values.

For determination of the exposure level, $L_{\text{B}}$ must be spatially averaged over a certain FOV. The ICNIRP recommends a time-dependent $\gamma$ value between a lower ($11\text{mrad}$) and an upper ($110\text{mrad}$) limit with the important note that ‘If the visual task and the behavior can be characterized, a safety analysis can account for more realistic eye movements and a larger averaging angle can be used’ (80) [7]. IEC 62471:2006 [9] defines more or less the same $\gamma$ values but with an upper limit of $100\text{mrad}$. In praxis, it is common to average $L_{\text{B}}$ over 11 or $100\text{mrad}$, both constant, and most instruments are designed for these two FOVs.

2.3. Instruments

Each partner of the HiPoSisAs project brought its own device(s) that is (are) usually applied for BLH RA at workplaces. Table 1 presents these 21 instruments, first with regard to their measurands and then in alphabetical order. For the sake of simplicity, a shortcut is assigned. If different sensors or entrance optics (EOP) can be mounted, the applied one is indicated.

Table 1. Devices used for the instrument comparison and the RAs.

Measurand	Shortcut	Device	Sensor/EOP	Spectral range (nm)	Calibration	Manufacturer
$E_{\text{v}}$^a	HCT	HCT-99D^b	CT-4501-4	380–780	November 2010	Gigahertz-Optik (Germany)
	VL-3701	X11	VL-3701	380–780	August 2019
	L100	Luxometer L-100	G.L-100	400–700	March 2022	Sonopan (Poland)
$E_{\text{B}}$^a	ILT	ILT1400-BL/U	SEL033/TBLU SCS395/TD	305–700	August 2020	Inter. Light Technol. (USA)
	XD45	X13	XD-45-HD-4^c	300–700	May 2012	Gigahertz-Optik (Germany)
$L_{\text{v}}$	LDM-VL	LDM-9805^d	VL	400–730	February 2022	Gigahertz-Optik (Germany)
	PD-VL	LDM-9811^d	PD-16VL	410–710	April 2017
$L_{\text{B}}$	LDM-BLH	LDM-9805^d	CD-2001, BLH	390–590	February 2022
	PD-BLH-1	LDM-9811^d	PD-16BLH	400–520	August 2019
	PD-BLH-2				April 2017
$E(\lambda )$^a	AvaSpec-1	AvaSpec-ULS2048XLS		200–1050	March 2021	Avantes (France)
	AvaSpec-2	AvaSpec-ULS2048CL-EVO		200–1100	February 2023
	CAS-152	CAS140CT-152	EOP-146	250–800	May 2022	Instrument Systems (Germany)
	Specbos	Specbos 1211UV-2		250–1000	May 2022	JETI (Germany)
	Flame	Flame-S-XR1-ES	ISP-50-8-I^e	250–1000	April 2022	Ocean Insight (Germany)
	Spectis	Spectis 5.2 Touch	Opti Probe 5.1.51	200–1050	November 2021	GL Optic (Poland)
	VL-TEC	BTS2048-VL-TEC		350–1050	February 2017	Gigahertz-Optik (Germany)
$L(\lambda )$	Specbos-ACC	Specbos 1211UV-2	ACC 024^f	250–1000	May 2022	JETI (Germany)
	Spectis-Tel	Spectis 5.0 Touch	Radiance Telescope 5.0.93^g	300–1050	November 2021	GL Optic (Poland)
$L_{\text{v}}(x,y)$	Canon	Canon EOS 6D MARK II^h		380–780	November 2021	Canon (France)
	Opticam	Opticam 3.0 4k TEC		380–780	September 2021	GL Optic (Poland)

^a Open FOV, cosine-corrected EOP.

^bAlso provides the correlated color temperature (CCT).

^cMeasures $E_{\text{B}}$ and $E_{\text{v}}$.

^dPossible $\gamma$ = 1.7, 11 and 100 mrad.

^eAlso applied in conjunction with a plano-convex lens of focal length $300\text{mm}$ and a circular aperture of $7\text{mm}$ diameter to determine $L_{\text{B}}$ according to the IEC 62471:2006 standard method [9].

^f$\gamma$ = 1.7 mrad at $200\text{mm}$, $11$ and $100\text{mrad}$ at any larger distance.

^g$\gamma =100\text{mrad}$.

^hCamera lens: Canon EF $100\text{mm}$ f/2.8L Macro IS USM.

Note: To the right of the measurands, a shortcut is assigned to every device or, if applicable, to the combination with its sensor/EOP. Spectral measurement range, date of the last (re)calibration and the manufacturer (country) are also given. $E(\lambda )$ = spectral irradiance; $E_{\text{B}}$ = blue-light hazard (BLH) weighted irradiance; EOP = entrance optic; $E_{\text{v}}$ = illuminance; f = focal point; $\gamma$ = field of view (FOV); $L(\lambda )$ = spectral radiance; $L_{\text{B}}$ = BLH weighted radiance; $L_{\text{v}}$ = luminance; $L_{\text{v}}(x,y)$ = two-dimensional luminance distribution; RA = risk assessment.

Participants were encouraged not to perform a recalibration before the two measurement series, one in May 2022 and the other in March 2023. This is important to allow for a best possible field instrument comparison because recalibrations are usually done no more than once a year. However, as can been seen in Table 1, the variety of (re)calibration dates is quite large: seven devices had not been calibrated for at least 3 years whereas seven instruments were calibrated prior to the start of the measurement campaign in 2022. In addition, the institute operating the Flame is accredited according to ISO/IEC 17025:2017 [21] to test to IEC 62471:2006, even on site; thus, the device is recalibrated very frequently. Note that not all HiPoSisAs partners were able to take part in each measurement series.

2.4. Radiation sources and measurement locations

A Bentham Halogen Spectral Radiance Standard (SRS 8) with a ground glass window of $D=50.4\text{mm}$ diameter was measured by most devices presented in Table 1 to test their inter-comparability. The SRS 8 is equipped with a $50\text{W}$ quartz-halogen lamp placed in an integrating sphere of $200\text{mm}$ diameter coated with BaSO₄. Its spectral radiance was calibrated last time in August 2019. Based on the expanded uncertainty with coverage factor $k=2$, the SRS 8 BLH weighted radiance, $L_B=3.3\pm 0.1\text{W}{\text{m}}^{-2}\text{s}{\text{r}}^{-1}$, and luminance, $L_{\text{v}}=12.0\pm 0.4\text{kcd}{\text{m}}^{-2}$, were calculated by means of Equations (1) and (2), respectively.

A Robe Spiider washbeam luminaire was used to compare RA approaches under increased measurement demands. The spotlight is equipped with 18 circularly arranged 40 W RGBW LED multichips and one centered 60 W RGBW LED multichip, all situated within a measured inner front diameter of $D=25.8\text{cm}$; however, the actual emitting surface is much smaller due to empty spaces between the LEDs. Measurements were performed at $2.58\text{m}$ (equal to $\alpha =100\text{mrad}$) and at $11.4\text{m}$ (in-laboratory exposure scenario), approximately at eye level ($1.6\text{m}$). The Robe Spiider was operated at $100\mathrm{\%}$ intensity and 150/255 zoom with a DMX-controlled CCT of T_cp = 5600 K.

The field tests were performed in two different places. The spotlights of setting I (see Table 2) were compiled and arranged by the authors in a large room. It was necessary to perform these measurements in a controlled surrounding to be able to try and discuss the applied RA approaches. Setting II was a stage used commercially for varying events. The three spotlights presented in the lower part of Table 2 were chosen from eight different options. At both settings I and II, the horizontal distances were large enough so that no spotlights overlapped with each other within a FOV of $\text{11}\text{mrad}$.

Table 2. Lamp characteristics and measurement distances for Bentham’s Spectral Radiance Standard SRS 8 (inter-comparability test), the Robe Spiider (demanding measurement conditions), for the in-laboratory exposure scenario (setting I) and at the assessed workplace (setting II).

	Spotlight	Type	D (cm)	d0 (m)	α (mrad)	$T_{\text{cp}}$ (K)
	Bentham SRS 8	Halogen	0.0504^a	0.5, 1, 2	100, 50, 25	3200^a
	Robe Spiider	(18 + 1) × RGBW	25.8	2.58	100	5600^b
I	ARRI L7-C	Fresnel RGBW	17.5^a	11.1	16	6550^b
	Robe MegaPointe	Short arc	6 (15)^a	11.1	5.4 (13)	6300^a
	Robe Spiider	(18 + 1) × RGBW	25.8	11.4	23	5600^b
	Ultralite PAR 56 (A)	PAR 56	13.5	11.2	12	2750^a
	Ultralite PAR 56 (B)			11.5
II	Martin MAC Quantum Profile	pc-wLED	10 (11.1)^a	14.5	6.9 (7.7)	6000^a
	Robe Robin 600+	37 × RGBW	24.5	14.0	18	8000^b
	Showtec Stage Blinder 2	2 × PAR 36	Each 8.5, 12^c	11.5	each 7.4, 10	$3200$^a

^aGiven in the manual.

^bDMX controlled.

^cCalculated via the summed up areas of both output lenses regarded as one circular area yielding a diameter of $8.5\text{cm}\cdot \sqrt{2}$.

Note: All lamps have a circular light output profile. The front side inner diameter, D, and direct line-of-sight distance between the spotlight and EOP, $d_0$, are listed together with the calculated angular subtense of the source, $\alpha \approx D{d}_0^{-1}$, and the CCT. For the MegaPointe and the Quantum Profile, both the diameter of the beam as well as of the output lens (in parentheses) are given. $\alpha$ = angular subtense of the source; EOP = entrance optic; PAR = parabolic aluminized reflector; pc-wLED = phosphor-converting white LED; SRS 8 = Bentham Halogen Spectral Radiance Standard; $T_{\text{cp}}$ = correlated color temperature (CCT).

2.5. BLH risk assessment methods

The basic idea of any BLH RA approach is to identify each spotlight that a worker might look at during an 8h working day, and to find out (by asking about the visual task and the behavior) for what length of time eye exposures accumulate. Then, a comparison of the determined $L_{\text{B}}$ or $E_{\text{B}}$ values with the related EL in Equations (7a) and (7b) allows the calculation of MPED and thus a recommendation concerning BLH workplace safety.

BLH RA methods that are common among the HiPoSisAs partners will be presented in the following. Note that glare usually prevents long-time viewing into the source, so that exposure durations less than $100\text{s}$ will be assumed hereafter for an 8h working day. Consequently, only the upper parts of Equations (7a) and (7b), the time-dependent parts, must be regarded.

2.5.1. Lamp safety standard approach

The $L_{\text{B}}$ value can always be detected and directly compared to the EL of Equation (7a). However, this can lead to problems for sources with $\alpha <11\text{mrad}$, as discussed later with respect to the Showtec Stage Blinder 2. In addition, radiance instruments are expensive, and an experienced operator is needed, e.g., because of potential misalignments. IEC 62471:2006 [9] defines $\gamma =100\text{mrad}$ for exposure durations of t ≥ 10,000 s with the related EL equal to $\text{100}\text{W}{\text{m}}^{-2}\text{s}{\text{r}}^{-1}$, in accordance with the ICNIRP. If the measured $L_{\text{B}}$ value is smaller than this EL, no MPED needs to be considered.

The determination of $E_{\text{B}}$ is usually based on an open FOV and not limited by an angle of acceptance. The respective irradiance EL of Equation (7b) can be applied for sources smaller than $\text{11}\text{mrad}$. Using this method also in the case of $\alpha \ge 11\text{mrad}$, the exposure distance, $d_0$, and the source diameter, D, must be known to calculate $L(\lambda )$ (or equally $L_{\text{B}}$) via Equation (4) in conjunction with Equation (3). In a simple approach, the open FOV $E_{\text{B}}$ value can be directly compared to the irradiance EL of $\text{1}\text{W}{\text{m}}^{-2}$ stemming from the multiplication of $\text{100}\text{W}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$ and Ω = 9.5 msr for ICNIRP’s $\gamma =110\text{mrad}$.

2.5.2. Luminance distribution approach

A BLH RA should consider the spectral as well as the spatial distribution of the light source, because the retina’s sensitivity is wavelength dependent and the highest risks are found where the source’s image is formed on the retina. Consequently, a spectral irradiance measurement together with the luminance distribution, $L_{\text{v}}(x,y)$, aim to collect the required information. For this combined method, first, the measured $E(\lambda )$ value is used to determine $K_{\text{B,v}}$ via Equations (1), (2) and (5). Then, the spatial luminance distribution recorded with a specifically calibrated camera is converted to $L_{\text{B}}(x,y)$ by means of this experimental $K_{\text{B,v}}$ value. Based on the new image, considering the exposure duration and distance, it is possible to average $L_{\text{B}}$ in the area that is limited by the averaging angle $\gamma$.

2.5.3. Alternative illuminance approach

An alternative BLH RA method is based on the CCT-dependent pre-calculated $K_{\text{B,v}}^{\text{Planck}}$ (see Appendix 1), in conjunction with a single $E_{\text{v}}$ measurement at the distance of exposure, $E_{\text{v}}(d_0)$ [19]. The spotlight’s CCT can be found either in the manual (for fixed $T_{\text{cp}}$) or is set by the operator via DMX. The underlying assumption that $K_{\text{B,v}}^{\text{Planck}}$ is close to the ‘actual’ $K_{\text{B,v}}$ value is fulfilled quite well not only for halogen lamps, but also for many other white-light sources like LED spotlights [16,19]. It is important to emphasize that a mismatch error of $K_{\text{B,v}}^{\text{Planck}}$ and $K_{\text{B,v}}$ is smaller than any uncertainty resulting from, e.g., a poor radiance detector alignment or imprecisely determined exposure conditions such as the distance, duration or viewing angle. In particular, the latter is difficult to find out precisely and results in the largest error.

Combining the described approach with the first lines of Equations (7a) and (7b) and the L-to-E conversion via Ω in Equations (3) and (4) yields: (8) $\begin{array}{rl}t& =(K_{\text{B,v}}^{\text{Planck}}{E}_{\text{v}}{)}^{-1}\\ & \times \{\begin{array}{ll}\text{100}\text{J}{\text{m}}^{-2}& \alpha <11\text{mrad}\\ 7.85\text{k}\text{J}{\text{m}}^{-2}& 11\text{mrad}\le \alpha <\gamma =100\text{mrad}\\ 785\text{k}\text{J}{\text{m}}^{-2}{\left(\frac{D}{d_0}\right)}^2& \alpha \ge \gamma \end{array}\end{array}$(8) as explained in the following. Provided that the distance $d_0$ between the spotlight and the exposed person is sufficiently long, the angular subtense of the source is smaller than $\text{11}\text{mrad}$, and the EL is expressed as irradiance. A direct conversion of the measured $E_{\text{v}}$ to $E_{\text{B}}$ can be done via Equation (5), yielding the first line of Equation (8). For sources with $\alpha \ge 11\text{mrad}$, the radiance EL must be used. According to the left side of Equation (3), Ω = 7.85 msr for $\gamma =100\text{mrad}$ leading to the second line of Equation (8).

A different situation exists if $\alpha$ is larger than the applied acceptance angle $\gamma$. Then, the spatially averaged ‘biologically relevant’ radiance is equal to the distance independent ‘physical’ one, at least in theory neglecting inhomogeneities of the emitting surface, and the right side of Equation (3) must be considered with a variable Ω. The resulting third line of Equation (8) can also be applied for a worst-case simplification: the over-restrictive assumption that light from a source with angular subtense $\alpha \ge 11\text{mrad}$ is emitted from a constant $\alpha =11\text{mrad}$ ($\mathrm{\Omega }=9.5\times {10}^{-5}\text{sr}\approx {10}^{-4}\text{sr}$) combined with the time-dependent EL of $\text{1}\text{MJ}{\text{m}}^{-2}\text{s}{\text{r}}^{-1}{t}^{-1}$ leads to the first line of Equation (8) that can be regarded as a worst-case approach for all types of $\alpha$–$\gamma$ combinations.

3. Results

3.1. Instrument inter-comparability test

The optical radiation emitted by the SRS 8 ground glass window was detected at a distance of $50.4\text{cm}$. Taking into account its diameter $D=50.4\text{mm}$, the angular subtense of the source is equal to $\alpha =100\text{mrad}$. Some of the devices presented in Table 1 also performed measurements at two additional distances, $100.7\text{cm}$ ($\alpha =50\text{mrad}$) and $201.6\text{cm}$ ($\alpha =25\text{mrad}$). Because of spatial limitations of the experimental set-up, $4.58\text{m}$ ($\alpha =11\text{mrad}$) could not be realized.

The measured $L(\lambda )$ and $E(\lambda )$ values were weighted with the BLH action spectrum and the photopic luminous efficiency function according to Equations (1) and (2). $E_{\text{v}}$ and $E_{\text{B}}$ values were converted to $L_{\text{v}}$ and $L_{\text{B}}$, respectively, by means of Equations (3) and (4). An average value was determined from two-dimensional luminance distributions, $L_{\text{v}}(x,y)$, by the operating software. All of these data are presented in Figure 1 as percentage deviations with regard to the calibration values $L_{\text{v}}=12.0\pm 0.4\text{kcd}{\text{m}}^{-2}$ and $L_{\mathrm{B}}=3.3\pm 0.1\text{W}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$ (coverage factor $k=2$).

Figure 1. Percentage deviations of measurement results at three distances with regard to (a) the luminance and (b) the blue-light hazard (BLH) weighted radiance of the SRS 8. Note: The instruments labeled according to Table 1 are grouped in terms of their measurands on the upper abscissa. The gray hatched areas reflect the expanded calibration uncertainties with coverage factor $k=2$. $\delta L_{\text{B}}$ = percentage deviation of measured/calculated to calibrated BLH weighted radiance; $\delta L_{\text{v}}$ = percentage deviation of measured/calculated to calibrated luminance; $E(\lambda )$ = spectral irradiance; $E_{\text{v}}$ = illuminance; $L_{\text{B}}$ = BLH weighted radiance; $L(\lambda )$ = spectral radiance; $L_{\text{v}}$ = luminance; $L_{\text{v}}(x,y)$ = two-dimensional luminance distribution; SRS 8 = Bentham Halogen Spectral Radiance Standard.

Figure 1(a) shows that most of the devices were able to determine $L_{\text{v}}$ within the expanded uncertainty (gray hashed area) or that they were close to it for distances of $50.4$ and $100.7\text{cm}$. The mean value of the absolute (positive) deviations, $|\delta L_{\text{v}}|$, is given by 2.6 ± 1.7%, which is an excellent outcome. Examining the $\delta L_{\text{v}}$ data reveals that the percentage deviations are much higher at $201.6\text{cm}$, reaching up to $14\mathrm{\%}$, and that the $E_{\text{v}}$ and $E(\lambda )$ instruments detect too much, except for the VL-TEC that needed to be recalibrated afterwards, in contrast to the $L_{\text{v}}$, $L(\lambda )$ and $L_{\text{v}}(x,y)$ devices. Both findings can be explained with the detection of reflected light from the surrounding. At a distance of $201.6\text{cm}$ to the SRS 8, the $E_{\text{v}}$ value becomes very small (<10 lx) approaching the lower end of the detectors’ sensitivity ranges. As a result, the relative contribution of detected reflections to the actual values increases leading to higher $\delta L_{\text{v}}$ values. Radiance and luminance are virtually unaffected by light from the surrounding.

Percentage deviations with regard to the calibrated BLH weighted radiance, $\delta L_{\text{B}}$, are depicted in Figure 1(b), and show findings more or less consistent with those for $\delta L_{\text{v}}$ whereby individual $\delta L_{\text{B}}$ values are higher up to $\pm 10\mathrm{\%}$ at $50.4$ and $100.7\text{cm}$. For both distances, the mean of $|\delta L_{\text{B}}|$ yields 4.2 ± 3.1% and therefore is close to the $\delta L_{\text{v}}$ average. A comparison of both PD-BLH instruments (identical types) demonstrates the effect of either varying calibrations and/or the operators’ handling.

Summarizing the inter-comparability test, it can be stated that all participating instruments can be compared among each other within percentage deviations of $\pm 10\mathrm{\%}$ for sufficiently high illuminance levels.

3.2. Demanding measurement conditions

While the instrument comparison described in the previous section was intended to verify the devices’ inter-comparability, the second test not only increases measurement demands regarding inhomogeneities of the emitting surface and the optical output power but also applies all RA approaches described in Section 2.5. For this, Robe’s Spiider was chosen with luminance levels in the order of $\text{Mcd}{\text{m}}^{-2}$ (mega candela per square meter). As can be seen in the iso-luminance distribution depicted in Figure 2(a), each of the eighteen 40 W and the centered 60 W RGBW LED have different $L_{\text{v}}(x,y)$ values. In addition, most of the front side does not emit light so that the $11$ and $\text{100}\text{mrad}$ FOVs shown (green color) do not correspond to the actual extent of the illuminating surface. This strongly varying light emission distribution is reflected in the large standard deviations of the mean luminance values of 2.1 ± 6.7 $\text{Mcd}{\text{m}}^{-2}$ ($\text{100}\text{mrad}$) and 18 ± 11 $\text{Mcd}{\text{m}}^{-2}$ ($\text{11}\text{mrad}$) determined by the Opticam software.

Figure 2. (a) Iso-luminance distribution of a Robe Spiider measured with an Opticam at a distance of $2.58\text{m}$. (b) Blue-light hazard (BLH) weighted radiances at $2.58\text{m}$ with their mean of $(2.8\pm 0.2)\text{kW}\text{m}{}^{-2}\text{sr}{}^{-1}$ shown as the dashed line and gray hatched area (not considering the LDM-BLH and Opticam values). Note: The color scale in panel (a) (not shown) ranges up to a maximum of $\text{35}\text{Mcd}\text{m}{}^{-2}$ (white). The green circles represent 11 and $\text{100}\text{mrad}$ fields of view (FOVs). $E(\lambda )$ = spectral irradiance; $E_{\text{v}}$ = illuminance; $L_{\text{B}}$ = BLH weighted radiance; $L(\lambda )$ = spectral radiance; $L_{\text{v}}(x,y)$ = two-dimensional luminance distribution.

Figure 2(b) summarizes the measurement results obtained at a distance of $2.58\text{m}$ (equal to $\alpha =100\text{mrad}$). Spectral radiance, $L(\lambda )$ with $\gamma =100\text{mrad}$, and irradiance, $E(\lambda )$, were convoluted with $B(\lambda )$ according to Equation (1). (Il)Luminance was converted to BLH weighted (ir)radiance by means of Equations (5) and (6) with $K_{\text{B,v}}^{\text{Planck}}=0.91\text{W}\text{kl}{\text{m}}^{-1}$ for $5600\text{K}$ (see Appendix 1). $E_{\text{B}}$ was subsequently converted to $L_{\text{B}}$ via Equations (3) and (4).

It is striking that the LDM-BLH $L_{\mathrm{B}}$ value of 28.8 kW m⁻² sr⁻¹ strongly deviates from the results of all other devices although it matched the expanded $L_{\text{B}}$ uncertainty of the SRS 8 in Figure 1(b). A discussion with the experienced operator revealed that the most likely explanation is incorrect handling of the photometer that is coupled to the LDM-BLH and whose FOV parameter must be set either to $1.7$, $11$ or $\text{100}\text{mrad}$. If the emitted light is detected within $\gamma =100\text{mrad}$ but is mistakenly associated with a $\text{11}\text{mrad}$ setting, such a large deviation can occur.

The $L_{\text{B}}$ mean (not considering the LDM-BLH and the Opticam) is given by 2.8 ± 0.2 $\text{kW}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$, equal to a standard deviation of $\pm 7\mathrm{\%}$ and comparable to the previous findings for $\delta L_{\text{v}}$ and $\delta L_{\text{B}}$. Note that the averaged $L_{\text{B}}=1.9\pm 6.1\text{kW}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$of the imaging Opticam was not considered because of its large standard deviation. The median for all $L_{\text{B}}$ values is 2.9 kW m^–2 sr^–1.

Close attention must be paid to the HCT and VL-3701 with respect to the alternative illuminance approach described in Section 2.5.3. Despite the complex $L_{\text{v}}$ distribution of the Spiider and the approximation of $K_{\text{B,v}}$ with $K_{\text{B,v}}^{\text{Planck}}$, the measured and converted illuminances are as accurate as the other results.

3.3. BLH risk assessment field tests

Changing from rather close measurement distances (with $\alpha \ge 11\text{mrad}$) and laboratory conditions to (almost) real-life situations, two field tests were performed as described in Section 2.4. Each spotlight was assessed via radiance, irradiance and illuminance measurements. The results are presented in Table 3.

Table 3. BLH assessment results for the spotlights of the in-laboratory exposure scenario and of the workplace.

		Radiance		Irradiance		Illuminance
Spotlight	Instrument	$L_{\text{B}}$	t	$E_{\text{B}}$	t	$E_{\text{v}}$	t	$t_{\text{res}}$
		($\text{kW}\text{m}{}^{-2}\text{ sr}{}^{-1}$)	($min$)	($\text{W}\text{m}{}^{-2}$)	($min$)	($\text{lx}$)	($min$)	($min$)
ARRI L7-C	(i)	2.2	7.7	0.30	11.0	355	8.6	4.4
	(ii)	1.9	8.7	0.28	11.5	350	9.6	4.9
Robe MegaPointe	(i)	(90)	($11\text{s}$)	7.7	$13\text{s}$	6900	$14\text{s}$
	(ii)	(76)	($13\text{s}$)	7.5	$13\text{s}$	6930	$14\text{s}$
Robe Spiider	(i)	3.5	4.8	1.6	4.1	1200	6.1	1.5
	(ii)	6.5	2.6	0.92	7.3	1150	4.8	1.2
Ultralite PAR 56 (A)	(i)	0.48	35	0.05	41	183	36	32
	(ii)	0.33	51	0.07	28	190	42	37
Ultralite PAR 56 (B)	(i)	0.33	51	0.07	27	120	53	49
	(ii)	0.34	49	0.04	45	250	32	29
Martin Quantum Profile	(i)	–	–	2.0	$50\text{s}$	2300	$44\text{s}$
	(ii)	(21.5)	($47\text{s}$)	2.1	$49\text{s}$	–	–	–
	(iii)	(20.4)	($49\text{s}$)	1.9	$52\text{s}$	2260	$45\text{s}$
	(iv)	–	–	2.1	$48\text{s}$	1978	$52\text{s}$
	(v)	(15.7)^a	($64\text{s}$)	1.5	$67\text{s}$	–	–	–
	(vi)	–	–	1.7	$57\text{s}$	–	–	–
	(vii)	–	–	1.8	$56\text{s}$	–	–	–
	(viii)	(22.0)	($46\text{s}$)	2.3	$43\text{s}$	–	–	–
Robe Robin 600+	(i)	–	–	0.17	23	140	23	9.5
	(ii)	0.9	19	0.16	25	–	–	–
	(iii)	0.8	21	0.18	23	159	20	8.3
	(iv)	–	–	0.19	21	191	17	6.9
	(v)	–	–	0.20	20	–	–	–
	(vi)	–	–	0.21	19	–	–	–
	(vii)	–	–	0.23	17	–	–	–
	(viii)	1.1, 0.033^b	15, Ø^c	0.26	15	–	–	–
Showtec Stage Blinder 2	(i)	–	–	0.055	30	150	28
	(ii)	(0.29, 0.20)^d	(39)^e	0.061	27	–	–	–
	(iii)	(0.20^f, 0.006^b)	(50^e, Ø^c)	0.056	30	165	26
	(iv)	–	–	0.069	24	165	26
	(v)	(0.36)^g	(47)	0.047	35	–	–	–
	(vi)	–	–	0.090	19	–	–	–
	(vii)	–	–	0.058	29	–	–	–
	(viii)	(0.31^h, 0.008^b)	(54, Ø^c)	0.063	26	–	–	–

^a Experimental $K_{\text{B,v}}=0.73\text{W}\text{kl}\text{m}{}^{-1}$ (AvaSpec-1) multiplied by $L_{\text{v}}=21.6\text{Mcd}\text{m}{}^{-2}$ (Canon).

^b$\gamma =100\text{mrad}$.

^cNo MPED or t > 10,000 s.

^d$\text{11}\text{mrad}$ $L_{\text{B}}$ for the left and the right part.

^e$L_{\text{B}}$ was summed up (or doubled), converted to $E_{\text{B}}$ and used in conjunction with Equation (7b) for D = 12 cm.

^fMean of both $\text{11}\text{mrad}{L}_{\text{B}}$ values.

^gExperimental $K_{\text{B,v}}=0.34\text{W}\text{kl}\text{m}{}^{-1}$ (AvaSpec-1) multiplied by $L_{\text{v}}^{\text{max}}=1.06\text{Mcd}\text{m}{}^{-2}$ (Canon).

^hMaximum $\text{11}\text{mrad}{L}_{\text{B}}$ value.

Note: For each method, the measurand, $L_{\text{B}}$ ($\gamma =11\text{mrad}$), $E_{\text{B}}$ or $E_{\text{v}}$, is shown on the left and the calculated MPED on the right. The more restrictive $t_{\text{res}}$ from the worst-case variant of the illuminance method is also listed. Note that some MPED are in units of seconds. Repeated measurements with different devices are indicated in the second column: $L_{\text{B}}$ – (i) PD-BLH-1, (ii) PD-BLH-2, (iii) Specbos-ACC, (v) Canon combined with AvaSpec-1, (viii) Flame; $E_{\text{B}}$ – (i) Spectis, (ii) VL-TEC, (iii) Specbos, (iv) ILT, (v) AvaSpec-1, (vi) AvaSpec-2, (vii) XD45, (viii) Flame; $E_{\text{v}}$ – (i) VL-3701, (ii) HCT, (iii) L100, (iv) XD45. $L_{\text{B}}$ and associated t given in parentheses do not meet the $\alpha \ge 11\text{mrad}$ condition according to Equation (7a). $\alpha$ = angular subtense of the source; D = front side inner diameter; $E_{\text{B}}$ = blue-light hazard (BLH) weighted irradiance; $E_{\text{v}}$ = illuminance; $\gamma$ = field of view (FOV); $K_{\text{B,v}}$ = BLH efficacy of luminous radiation; $L_{\text{B}}$ = BLH weighted radiance; $L_{\text{v}}$ = luminance; t = maximum permissible exposure duration (MPED); $L_{\text{v}}^{\text{max}}$ = maximum luminance; PAR = parabolic aluminized reflector; $t_{\text{res}}$ = restrictive maximum permissible exposure duration.

For sources with $\alpha \ge 11\text{mrad}$, MPED values based on $L_{\text{B}}$ were calculated via Equation (7a). Irradiance was converted to $L_{\text{B}}$ values by means of Equations (3) and (4) and subsequently compared with the radiance BLH EL of 1 MJ m⁻² sr⁻¹ t⁻¹. For the alternative $E_{\text{v}}$ approach, $K_{\text{B,v}}^{\text{Planck}}$ was taken from Appendix 1 with the CCTs presented in Table 2. Based on the restrictive assumption of an $\text{11}\text{mrad}$ FOV, the angular subtense of 5 from the 8 spotlights is larger than $\gamma$, and the third line of Equation (8) must be used.

The beam diameters of the MegaPointe and the Quantum Profile were measured to be approximately equal to $6$ and $\text{10}\text{cm}$ yielding angular subtenses of $5.4$ and $\text{6}\text{.9}\text{mrad}$, respectively. For these sources, $L_{\text{B}}$ and $E_{\text{B}}$ values were compared with their respective EL. The illuminance RA method can be applied with the first line of Equation (8) that is also equal to the worst-case approach with the most restrictive MPED, $t_{\text{res}}$. A detailed discussion of the Showtec Stage Blinder 2 as well as of all MPEDs is presented in the following sections.

4. Discussion

During the in-laboratory exposure scenario, problems arose with regard to both PAR 56 spotlights. Frequently switching them on and off presumably led to a varying filament temperature, resulting in a different light emission that is also reflected in the MPED. At the workplace, the $L_{\text{v}}$ distribution method (v), based on the bracketing technique, failed for the Robe Robin 600+ (Table 3). A real-time recording revealed that the emission of the spotlight’s LEDs were rotating and thus flickering. Furthermore, the HiPoSisAs partners planned to assess a Martin Atomic 3000 flashlight (xenon discharge lamp), not presented in Table 2, but its pulsed emission yielded noisy spectra being close to zero, caused by a mismatch of the integration time, pulse length and spectral averaging. Special equipment will be necessary for a RA of the Atomic 3000.

4.1. Showtec stage blinder 2

Although photobiologically safe, the Showtec Stage Blinder 2 resulted in the longest discussion among the HiPoSisAs partners. As can be seen from its luminance distribution shown in Figure 3, the spotlight consists of two separate sources with their filaments being rotated by ${90}^{\circ }$. One RA approach, carried out by the PD-BLH-2 instrument (ii), measured $L_{\text{B}}$ with $\gamma =11\text{mrad}$ for each lamp individually. Assuming that light from both parts is focused on one retinal site, their $L_{\text{B}}$ sum is converted to one $E_{\text{B}}$ value and compared to the irradiance BLH EL, but using an averaged diameter of $12\text{cm}$ (substituting both emitting areas by a single circular one) instead of $8.5\text{cm}$. The resulting MPED is given as $39min$. A similar approach was taken by the Specbos-ACC (iii) applying the doubled mean $\text{11}\text{mrad}$ $L_{\text{B}}$ value of both lamps (MPED of $50min$). The Flame (viii) spectroradiometer determined the maximum $\text{11}\text{mrad}$ $L_{\text{B}}$ value from the brightest of the two sources, yielding an MPED of $54min$.

Figure 3. Two-dimensional luminance distribution, $L_{\text{v}}(x,y)$, of the Showtec Stage Blinder 2 recorded by the Canon camera. Note: The $L_{\text{v}}$ scale is given as $\text{Mcd}\text{m}{}^{-2}$. The white double arrow represents the spotlight’s diameter of D = 8.5 cm. The red circle visualizes $\gamma =11\text{mrad}$. D = front side inner diameter; $\gamma$ = field of view (FOV); $L_{\text{v}}$ = luminance; $L_{\text{v}}(x,y)$ = two-dimensional luminance distribution.

The latter two instruments also followed the RA approach based on IEC 62471:2006. Measurements with $\gamma =100\text{mrad}$ provided $L_{\text{B}}$ values well below $\text{100}\text{W}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$ so that no MPED must be considered, and the Stage Blinder 2 can be regarded as BLH safe. $L_{\text{v}}(x,y)$ recorded by the Canon camera (v) was used to identify the maximum luminance, $L_{\text{v}}^{\text{max}}=1.06\text{Mcd}{\text{m}}^{-2}$, of both parts. In combination with an experimentally determined $K_{\text{B,v}}=0.34\text{W}\text{kl}{\text{m}}^{-1}$ from the AvaSpec-1 (v), this maximum luminance was converted to $L_{\text{B}}=0.36\text{kW}{\text{m}}^{-2}\text{ s}{\text{r}}^{-1}$ via Equation (5) yielding a MPED of $47min$.

4.2. Comparison of maximum permissible exposure durations

In order to compare all MPED values presented in Table 3, mean values with standard deviations (absolute and percentage) were calculated. Over-restricted MPED, $t_{\text{res}}$, values were not considered.

Regarding the in-laboratory exposure setting (upper part of Table 4) most MPED values agree among each other except those of the irradiance and illuminance RA approach for the ARRI L7-C and for the MegaPointe, respectively. Potential detector misalignments, different calibrations or a $K_{\text{B,v}}^{\text{Planck}}$-to-$K_{\text{B,v}}$ mismatch (for the $E_{\text{v}}$ method) can be reasons for this. Percentage deviations of the radiance and irradiance RA approaches for the Spiider are higher than $40\mathrm{\%}$ and thus do not meet the maximum $30\mathrm{\%}$ measurement uncertainty given by EN 14255-2:2005 [22], whereas the $E_{\text{v}}$ method is more accurate. However, the validity of this finding is not sufficient taking into account only two different instruments for each RA method. The irradiance and illuminance RA approach for the PAR 56 (B) also have percentage deviations larger than $30\mathrm{\%}$. In general, increasing the number of instruments for the workplace RA (lower part of Table 4) improves the statistics. Except for the Stage Blinder 2, all MPED values agree among each other within their percentage deviations.

Table 4. Mean MPED calculated on the basis of the data presented in Table 3.

Spotlight	Units	Radiance	Irradiance	Illuminance
ARRI L7-C	Minutes	$8.2\pm 0.7$, $9\mathrm{\%}$ (2)	$11.2\pm 0.4$, $3\mathrm{\%}$ (2)	$9.1\pm 0.7$, $8\mathrm{\%}$ (2)
Robe MegaPointe	Seconds	$12\pm 1$, $12\mathrm{\%}$ (2)	$13.1\pm 0.2$, $2\mathrm{\%}$ (2)	$14.2\pm 0.1$, $1\mathrm{\%}$ (2)
Robe Spiider	Minutes	$3.7\pm 1.6$, $44\mathrm{\%}$ (2)	$5.7\pm 2.2$, $40\mathrm{\%}$ (2)	$5.5\pm 0.9$, $17\mathrm{\%}$ (2)
Ultralite PAR 56 (A)	Minutes	$43\pm 11$, $26\mathrm{\%}$ (2)	$35\pm 9$, $25\mathrm{\%}$ (2)	$39\pm 4$, $10\mathrm{\%}$ (2)
Ultralite PAR 56 (B)	Minutes	$50\pm 2$, $4\mathrm{\%}$ (2)	$36\pm 12$, $34\mathrm{\%}$ (2)	$42\pm 15$, $35\mathrm{\%}$ (2)
Martin Quantum Profile	Seconds	$51\pm 8$, $16\mathrm{\%}$ (4)	$53\pm 7$, $14\mathrm{\%}$ (8)	$47\pm 4$, $8\mathrm{\%}$ (3)
Robe Robin 600+	Minutes	$18\pm 3$, $16\mathrm{\%}$ (3)	$21\pm 3$, $16\mathrm{\%}$ (8)	$20\pm 3$, $15\mathrm{\%}$ (3)
Showtec Stage Blinder 2	Minutes	$47\pm 6$, $13\mathrm{\%}$ (4)	$28\pm 5$, $18\mathrm{\%}$ (8)	$27\pm 1$, $6\mathrm{\%}$ (3)

Note: Standard deviations are given in absolute units as well as in percentage. Number of individual measurements considered for each mean are presented in parentheses. MPED = maximum permissible exposure duration; PAR = parabolic aluminized reflector.

Overall, it is notable that the alternative illuminance RA approach has the lowest standard deviation for six of the eight assessed spotlights. In accordance with previous findings [19], it can be concluded that the $E_{\text{v}}$ method is applicable for a BLH workplace RA of the tested white-light sources just as well as the radiance or irradiance approach.

5. Conclusion

The idea of a simple and practical BLH RA approach arose from demanding radiance measurements with expensive equipment, whereas luxmeters are easy to handle, cost-effective, usually available at workplaces with high-power spotlights and thus known by persons commissioned with performing a RA. Combining a single $E_{\text{v}}$ measurement, a pre-calculated tabular $K_{\text{B,v}}$ value for a blackbody radiator (for given CCT) and some solid angle considerations yielded three simple equations for the MPED. It was the aim of the present work to evaluate the applicability of this alternative illuminance RA method.

In order to make a statement regarding inter-comparability, first, the applied instruments were tested with respect to $L_{\text{B}}$ and $L_{\text{v}}$ values of a comparably low-power halogen radiance standard. The determined values were found within or near to the expanded uncertainties ($\pm 4\mathrm{\%}$) for close distances. Otherwise, they deviated up to $\pm 15\mathrm{\%}$ as the consequence of a small illuminance (E_v < 10 lx) challenging the detectors’ lower dynamic ranges. In a second step, the light emission of a complex high-power spotlight was detected. Most measurement results including those from the alternative illuminance RA approach were presented by means of $L_{\text{B}}$ and were found within a standard deviation of $\pm 7\mathrm{\%}$. Finally, two exposure settings were assessed, a semi-in-laboratory setting and a workplace setting, with eight spotlights in total. Not all of the RA methods could meet the $30\mathrm{\%}$ maximum measurement uncertainty criterion given by EN 14255-2:2005 for every spotlight. However, most of the standard deviations fulfilled this requirement, and those from the illuminance RA approach were the smallest (<20%) for six of the eight spotlights. Concluding these findings, the alternative $E_{\text{v}}$ method is at least as accurate as the radiance or irradiance RA approach; thus, its practicality can be stated for a BLH RA of white-light sources at workplaces. For this, it is recommended to use luxmeters that have a certificate of calibration.

For a continued assessment of the alternative illuminance RA method, a larger number of different spotlights should be considered to characterize potential problems, e.g., regarding sources with a specific color where the experimental $K_{\text{B,v}}$ value does not agree with the $K_{\text{B,v}}^{\text{Planck}}$ value. During the preparation of this article, the Austrian Workers’ Compensation Board (AUVA) and Seibersdorf Laboratories were conducting detailed measurements on selected spotlights (on site and in the laboratory) with the aim to compare results obtained by using simplified methods for photochemical and photothermal retinal hazards with those from a highly accurate ‘artificial eye’ set-up for radiance determination. Their findings will provide further information about the practicality of this work’s alternative illuminance RA approach.

Acknowledgements

The authors thank PEROSH for providing the framework within the HiPoSisAs project took place. The regular support of PEROSH but also that of the participating occupational safety and health institutes (AUVA, BAuA, CIOP-PIB, IFA, INAIL, INRS, Unisanté) and cooperative partners (DGUV, GL Optic, JETI, OSRAM, Seibersdorf Laboratories) allowed the experimental evaluation of the proposed illuminance BLH RA approach on a European level.

Disclosure statement

No potential conflict of interest was reported by the authors.

Appendix 1.

Calculated values for $K_{\text{B,v}}^{\text{Planck}}$

Table A1. CCT-dependent BLH efficacy of luminous radiation for a Planckian radiator calculated according to Equation (6).

$T_{\text{cp}}$	$K_{\text{B,v}}^{\text{Planck}}$	$T_{\text{cp}}$	$K_{\text{B,v}}^{\text{Planck}}$	$T_{\text{cp}}$	$K_{\text{B,v}}^{\text{Planck}}$
2.0	0.13	4.0	0.58	6.0	0.98
2.2	0.16	4.2	0.63	6.2	1.01
2.4	0.21	4.4	0.67	6.4	1.04
2.6	0.25	4.6	0.71	6.6	1.07
2.8	0.30	4.8	0.76	6.8	1.10
3.0	0.34	5.0	0.80	7.0	1.13
3.2	0.39	5.2	0.84	7.2	1.16
3.4	0.44	5.4	0.87	7.4	1.19
3.6	0.49	5.6	0.91	7.6	1.21
3.8	0.53	5.8	0.94	7.8	1.24
–	–	–	–	8.0	1.26

Note: $T_{\text{cp}}$ is given in multiples of $1000\mathrm{K}$; $K_{\text{B,v}}^{\text{Planck}}$ in $\text{W}\text{kl}\text{m}{}^{-1}$. $K_{\text{B,v}}^{\text{Planck}}$ = blue-light hazard (BLH) efficacy of luminous radiation for a Planckian radiator; $T_{\text{cp}}$ = correlated color temperature (CCT).