Solar Glare Hazard Analysis Tool (SGHAT)

Technical Reference Manual

Clifford K. Ho, Cianan A. Sims, Julius Yellowhair, and Evan Bush
Sandia National Laboratories
(505) 844-2384, [email protected]
SAND2014-18360 O
March 2015

Contents

1. Requirements ........................................................................................................................... 2
2. Introduction ............................................................................................................................. 2
3. Assumptions and Limitations .................................................................................................. 2
4. Determination of Glare Occurrence ........................................................................................ 3
4.1 Sun Position.................................................................................................................... 3
4.2 Reflected Sun Vector...................................................................................................... 5
4.3 Scattering and Subtended Beam Angle .......................................................................... 5
4.4 Beam Projection onto PV Array Plane ........................................................................... 6
4.5 PV Single-Axis Tracking ............................................................................................... 7
4.6 PV Dual-Axis Tracking .................................................................................................. 8
5. Determination of Ocular Impact.............................................................................................. 9
5.1 Ocular Hazard Plot ......................................................................................................... 9
5.2 Direct Normal Irradiance (DNI) ................................................................................... 10
5.3 Reflectance ................................................................................................................... 11
5.4 Slope Error ................................................................................................................... 12
5.5 Subtended Beam Angle ................................................................................................ 12
6. Annual Energy Production .................................................................................................... 13
7. Other Formulations................................................................................................................ 13
7.1 Flight Path Calculations ............................................................................................... 13
8. Acknowledgments ................................................................................................................. 15
9. References ............................................................................................................................. 15

1|Page
1. Requirements
 Use of this software requires the latest version of one of the following free web browsers:
Mozilla Firefox or Google Chrome.
 The Solar Glare Hazard Analysis Tool (SGHAT) can be accessed by registering at
www.sandia.gov/glare.

2. Introduction
With growing numbers of solar energy installations throughout the United States, glare from
photovoltaic (PV) arrays and concentrating solar systems has received increased attention as a
real hazard for pilots, air-traffic control personnel, motorists, and others. Sandia has developed a
web-based interactive tool that provides a quantified assessment of (1) when and where glare
will occur throughout the year for a prescribed solar installation, (2) potential effects on the
human eye at locations where glare occurs, and (3) an estimate of the maximum annual energy
production.

The Solar Glare Hazard Analysis Tool (SGHAT) employs an interactive Google map where the
user can quickly locate a site, draw an outline of the proposed PV array, and specify observer
locations or paths. Latitude, longitude, and elevation are automatically recorded through the
Google interface, providing necessary information for sun position and vector calculations.
Additional information regarding the orientation and tilt of the PV panels, reflectance,
environment, and ocular factors are entered by the user.

If glare is found, the tool calculates the retinal irradiance and subtended angle (size/distance) of
the glare source to predict potential ocular hazards ranging from temporary after-image to retinal
burn. The results are presented in a simple, easy-to-interpret plot that specifies when glare will
occur throughout the year, with color codes indicating the potential ocular hazard. The tool can
also predict relative energy production while evaluating alternative designs, layouts, and
locations to identify configurations that maximize energy production while mitigating the
impacts of glare.

This Technical Reference Manual describes the theory and models used in SGHAT.

3. Assumptions and Limitations

Below is a list of assumptions and limitations of the models and methods used in SGHAT:

 The software currently only applies to flat reflective surfaces. For curved surfaces (e.g.,
focused mirrors such as parabolic troughs or dishes used in concentrating solar power
systems), methods and models derived by Ho et al. (2011) [1] can be used and are
currently being evaluated for implementation into future versions SGHAT.
 When enabled, PV array single- or dual-axis tracking does not account for backtracking
or the effects of panel shading and blocking.

2|Page
  SGHAT does not rigorously represent the detailed geometry of a system; detailed
features such as gaps between modules, variable height of the PV array, and support
structures may impact actual glare results. However, we have validated our models
against several systems, including a PV array causing glare to the air-traffic control tower
at Manchester-Boston Regional Airport and several sites in Albuquerque, and the tool
accurately predicted the occurrence and intensity of glare at different times and days of
the year.
 SGHAT assumes that the PV array is aligned with a plane defined by the total heights of
the coordinates outlined in the Google map. For more accuracy, the user should perform
runs using minimum and maximum values for the vertex heights to bound the height of
the plane containing the solar array. Doing so will expand the range of observed solar
glare when compared to results using a single height value.
 SGHAT does not consider obstacles (either man-made or natural) between the
observation points and the prescribed solar installation that may obstruct observed glare,
such as trees, hills, buildings, etc.
 The variable direct normal irradiance (DNI) feature (if selected) scales the user-
prescribed peak DNI using a typical clear-day irradiance profile. This profile has a lower
DNI in the mornings and evenings and a maximum at solar noon. The scaling uses a
clear-day irradiance profile based on a normalized time relative to sunrise, solar noon,
and sunset, which are prescribed by a sun-position algorithm [2] and the latitude and
longitude obtained from Google maps. The actual DNI on any given day can be affected
by cloud cover, atmospheric attenuation, and other environmental factors.
 The ocular hazard predicted by the tool depends on a number of environmental, optical,
and human factors, which can be uncertain. We provide input fields and typical ranges of
values for these factors so that the user can vary these parameters to see if they have an
impact on the results. The speed of SGHAT allows expedited sensitivity and parametric
analyses.

4. Determination of Glare Occurrence

Determination of glare occurrence requires knowledge of the following: sun position, observer
location, and the tilt, orientation, location, extent, and optical properties of the modules in the
solar array. Vector algebra is then used to determine if glare is visible from the prescribed
observation points.

4.1 Sun Position

The sun position algorithm[2] calculates the sun position in two forms: first as a unit vector
extending from the Cartesian origin toward the sun, and second as azimuthal and altitudinal
angles. The algorithm relies on the latitude, longitude and time zone offset from UTC in order to
determine the position of the sun at every time step throughout the year.

First, we calculate the solar time:

3|Page
 𝑡𝑠𝑜𝑙𝑎𝑟 = 4(𝐿𝑠𝑡 − 𝐿𝑙𝑜𝑐 ) + 𝐸 + 𝑡𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑

Where:

𝐿𝑠𝑡 = 𝑡𝑧𝑜𝑓𝑓𝑠𝑒𝑡 ∗ 15

𝐸 = 229.2(0.000075 + 0.001868 ∗ cos 𝐵 − 0.0320077 ∗ sin 𝐵 − 0.014615 ∗ cos 2𝐵

− 0.04089 ∗ sin 2𝐵)

Lst is the local standard meridian, Lloc is the given longitude and E is the equation of time, in
minutes.

The solar time can then be used to calculate the Hour angle, 𝜔:

𝜔 = ∆𝑡𝑛𝑜𝑜𝑛 ∗ 15

Where ∆𝑡𝑛𝑜𝑜𝑛 is the difference between solar time and solar noon.

Once the declination,𝛿 is known, the solar zenith and azimuthal angle of the sun can be found:

284 + 𝑛
𝛿 = 23.45 ∗ sin (360 ∗ )
365

𝜃𝑧 = cos −1(cos 𝜑 ∗ cos 𝛿 ∗ cos 𝜔 + sin 𝜑 ∗ sin 𝛿)

cos 𝜃𝑧 sin 𝜙 − sin 𝛿

𝛾𝑠 = 𝑠𝑖𝑔𝑛(𝜔) |cos−1 ( )|
sin 𝜃𝑧 cos 𝜙

Where:

 n is the day of the year (1 to 365)

 𝜃𝑧 is the sun zenith angle (subtract from 90 to get the altitude angle, 𝜃𝑎 )

 𝜑 is the given latitude

 𝛾𝑠 is the sun azimuthal angle

The sun altitude and azimuth can be converted to unit vector components as follows:

⃗⃗𝑠𝑖 = cos 𝜃𝑎 ∗ sin 𝛾𝑠

⃗⃗𝑠𝑗 = cos 𝜃𝑎 ∗ cos 𝛾𝑠

𝑠𝑘 = sin 𝜃𝑎
⃗⃗⃗

4|Page
4.2 Reflected Sun Vector
Once the sun position is known for each time interval a simple vector reflection equation [4] can
determine the reflected sun vector, based on the normal vector of the PV array panels:

𝑥1′ − 𝑥0 = 𝒗 − 2(𝒗 ∙ 𝒏
̂ ) ̂𝒏

Figure 1 illustrates this vector reflection graphically.

Figure 1 - Vector reflection over normal vector of plane. Source:

mathworld.wolfram.com/Reflection.html

4.3 Scattering and Subtended Beam Angle

The reflected sun vector defines the axis of a conical beam representing the actual beam of
sunlight. This sunbeam is translated to extend from the OP toward the PV array. The aperture of
this conical sun beam is equivalent to β, the subtended beam angle, which is the sum of the sun
shape and the scattering caused by slope error:

𝜃𝑠𝑢𝑛 𝑎𝑛𝑔𝑙𝑒
𝛽 = 2∗( + 2 ∗ 3 ∗ 𝜃𝑠𝑙𝑜𝑝𝑒 𝑒𝑟𝑟𝑜𝑟 )
2

5|Page
4.4 Beam Projection onto PV Array Plane
The beam is projected onto the PV array in several steps: first, points lying on the edge of the
beam in a conical section orthogonal to the axis are calculated. This conical section is arbitrarily
defined to be 1 meter from the cone apex (the OP).

These 30 points are calculated by randomly generating two coordinates and solving for the third
using the following equation:

𝑣𝑎𝑥𝑖𝑠 ∙ 𝑣𝑟𝑎𝑑𝑖𝑢𝑠 = 0

This equation states that the cone axis is orthogonal to the radius vectors of the conical section
upon which the 30 conical points lie.

Next, conical edge vectors are defined by subtracting the cone apex (the OP) from the cone
points. This collection of vectors extends from the OP toward the PV array plane. These vectors
define the conical sun beam. At their center, or the axis of the cone, is the reflected sun vector
calculated in 4.2.

These conical vectors are then intersected with the PV array plane. This cone-plane intersection
will be an elliptical conical section defined by 30 co-planar points.

These intersection points are calculated using line-plane intersection equations [5]:

(𝒑𝟎 − 𝑰𝟎 ) ∙ 𝒏
⃗
𝑑=
𝑰∙𝒏⃗⃗
(𝑥, 𝑦, 𝑧) = 𝑑𝐈 + 𝑰𝟎

Where:

 ⃗⃗ is the PV array panel normal vector

𝒏

 I is one of the vectors extending from the OP to the PV array plane, which define the
conical sun beam.

 I0 is a point on the vector (the OP)

 p0 is a point on the PV array plane

 d is the distance from the OP to the intersection point, and

 (x, y, z) define the intersection point for this vector.

The n intersection points found using the above equations define the elliptical conical section of
the sun beam cone as it intersects the PV array plane.

6|Page
Glare is present and viewable from the OP if any of the PV array vertices lie within this co-
planar ellipse. This is determined using an optimized points-in-polygon algorithm from
matplotlib [6].

4.5 PV Single-Axis Tracking

Single-axis tracking allows for the PV panels to rotate over one dimension in order to track the
apparent movement of the sun over time. This rotation is modeled using the normal vector of the
⃗⃗ . The components of 𝒏
PV array panels, 𝒏 ⃗ are calculated using the following:

 𝛽𝑡 – Tracking axis tilt where 0º is parallel with flat ground and 90º is perpendicular to the
ground, facing the horizon.

 𝜇 – Panel offset from the tracking axis.

 𝜌 – Tracking, or rotation, angle designating the rotation of the panel at a given time.
Clockwise and counter-clockwise over the tracking axis (see below).

 𝛾 – Orientation of the tracking axis. Clockwise from due south (0º).

Figure 2 - PV panel with single-axis tracking. The panel normal is displayed as N. Source:
http://www.powerfromthesun.net/Book/chapter04/chapter04.html

The components of 𝒏 ⃗ are first calculated in a Cartesian coordinate system of (b, r, u) where r
aligns with the tracking axis, b is perpendicular to it and extends to the horizon, and u is
perpendicular to r vertically (see Figure 2). Note that 𝛽𝑡 , 𝜇, 𝛾 are all inputs provided by the user.
7|Page
 cos(𝜃𝑎 ) sin(𝛾𝑠 − 𝛾)
𝜌 = tan−1 { }
sin(𝜃𝑎 − 𝛽𝑡 ) + [1 − cos(𝛾𝑠 − 𝛾)] sin(𝛽𝑡 ) cos(𝜃𝑎 )

⃗⃗ 𝑏 = sin 𝜌
𝒏

⃗ 𝑟 = sin 𝜇
𝒏

⃗⃗ 𝑢 = cos 𝜌
𝒏

These components are converted back to the standard Cartesian system:

⃗⃗ 𝑧 = 𝒏
𝒏 ⃗ 𝑢 cos 𝛽𝑡 + 𝒏
⃗ 𝑟 sin 𝛽𝑡

⃗ 𝑢 sin(𝛾) sin(𝛽𝑡 ) + 𝒏
⃗⃗ 𝑒 = −𝒏
𝒏 ⃗ 𝑟 sin(𝛾) cos(𝛽𝑡 )
⃗ 𝑏 cos(𝛾) + 𝒏

⃗ 𝑢 cos(𝛾) sin(𝛽𝑡 ) − 𝒏
⃗ 𝑛 = −𝒏
𝒏 ⃗⃗ 𝑟 cos(𝛾) cos(𝛽𝑡 )
⃗ 𝑏 sin(𝛾) + 𝒏

Vector components are calculated for the panels at each time step.

4.6 PV Dual-Axis Tracking

Dual-axis tracking implies the PV panels face “toward” the sun at every time step. Again, the
panel normal varies every minute. Because the variance occurs in two dimensions, the sun vector
(extending from the origin toward the sun) can be used as the panel normal:

⃗ =𝒔
𝒏 ⃗

8|Page
5. Determination of Ocular Impact
Determination of the ocular impact requires knowledge of the direct normal irradiance, PV
module reflectance, size and orientation of the array, optical properties of the PV module, and
ocular parameters. These values are used to determine the retinal irradiance and subtended
source angle used in the ocular hazard plot [1, 3].

5.1 Ocular Hazard Plot

The ocular impact of viewed glare can be classified into three levels based on the retinal
irradiance and subtended source angle: low potential for after-image, potential for after-image,
and potential for permanent eye damage [1, 3]. The following log-log plot illustrates these three
areas of glare intensity:

Figure 3 - Glare hazard plot illustrating the ocular impact as a function of retinal irradiance
and subtended source angle [1, 3].

The subtended source angle represents the size of the glare viewed by an observer, while the
retinal irradiance determines the amount of energy impacting the retina of the observer. Larger
source angles can result in glare of high intensity, even if the retinal irradiance is low.

The boundary between the “yellow” and “red” regions, signifying glare that transitions from
causing an after-image to causing permanent eye damage, can be quantified with the following
equations:

9|Page
 0.118
𝐸𝑟,𝑏𝑢𝑟𝑛 = for 𝜔 < 0.118 rad
𝜔

𝐸𝑟,𝑏𝑢𝑟𝑛 = 1 for 𝜔 ≥ 0.118 rad

The second boundary, between the low potential for after-image (green) and potential for after-
image (yellow) areas, adheres to the following equation:

3.59 × 10−5
𝐸𝑟,𝑓𝑙𝑎𝑠ℎ =
𝜔1.77

5.2 Direct Normal Irradiance (DNI)

The variable direct normal irradiance (DNI) feature (if selected) scales the user-prescribed peak
DNI using a typical clear-day irradiance profile. This profile has a lower DNI in the mornings
and evenings and a maximum at solar noon. The scaling uses a clear-day irradiance profile based
on a normalized time relative to sunrise, solar noon, and sunset, which are prescribed by a sun-
position algorithm [2] and the latitude and longitude obtained from Google maps.

𝐷𝑁𝐼 = cos(1 − 𝑡𝑠 )

Here 𝑡𝑠 represents the normalized time relative to solar noon. Normalization is based on the
amount of time between sunrise or sunset and solar noon.

The DNI scaling profile was determined by fitting empirical DNI data to the cosine function, as
illustrated in Figure 4. Note that DNI on any given day can be affected by cloud cover,
atmospheric attenuation, and other environmental factors.

10 | P a g e
 Actual summer DNI
1000 Predicted summer DNI
Actual winter DNI
Predicted Winter DNI
800
DNI (W/m^2)

600

400

200

0
0:00 4:48 9:36 14:24 19:12 0:00
Time of day

Figure 4 - Fit functions modeling normalized DNI vs. hour. Cosine was chosen to profile
empirical data.

5.3 Reflectance
Panel reflectivity can be varied for each time step to account for the position of the sun relative
to the array. Smooth glass and light textured glass with and without Anti-Reflection coating,
along with deeply textured glass were analyzed to derive accurate functions for computing
reflectivity based on sun incidence angle [7].

Table 1 contains the fit functions for panel reflectivity.

Table 1 - Reflectance fit functions for PV cover types.

Fit Function Defined over
PV Glass Cover Type Fit Function Defined over 0    60 60 <  < 90
Smooth Glass without Anti- 2 0.1019x
y = 1.1977E-5 x – 9.5728E-4 x + 4.410E-2 y = 6.2952E-5 e
Reflection Coating
Smooth Glass with Anti-Reflection 2 0.1051x
y = 1.473E-5 x – 9.6416E-4 x + 3.2395E-2 y = 4.7464E-5 e
Coating
Light Textured Glass without Anti- 2 0.0994x
y = 1.5272E-5 x – 1.1304E-3 x + 4.305E-2 y = 7.3804E-5 e
Reflection Coating
Light Textured Glass with Anti- 2 0.0994x
y = 1.4188E-5 x – 1.0326E-3 x + 3.9016E-2 y = 7.0179E-5 e
Reflection Coating
2 0.0834x
Deeply Textured Glass y = 6.8750E-6 x – 6.5250E-4 x + 2.10E-2 y = 4.1793E-5 e

11 | P a g e
5.4 Slope Error
The slope error can be automatically correlated with the selected surface type by checking
correlate slope error with module surface type. If unchecked, the slope error can be entered
manually. If checked, the following table is referenced for determining the appropriate slope
error and corresponding value of 𝛽, the subtended beam angle:

Table 2. Slope error and beam spread vs. surface type

Average
Average Standard Standard
Beam
PV Glass Cover Type RMS Slope deviation of deviation of
Spread
Error (mrad) slope error beam error
(mrad)
Smooth Glass without
6.55 87.9 4.43 53.3
Anti-Reflection Coating
Smooth Glass with Anti-
8.43 110 2.58 30.9
Reflection Coating
Light Textured Glass
without Anti-Reflection 9.70 126 2.78 33.3
Coating
Light Textured Glass with
9.16 119 3.17 38.0
Anti-Reflection Coating
Deeply Textured Glass 82.6 1000 N/A N/A

5.5 Subtended Beam Angle

The glare analysis must account for the actual visible area of the PV array when viewed from the
observation point. For example, less viewable area will be apparent when viewing an array with
panel tilt of 0 degrees on a flat surface from the side than when viewing it from above in an
aircraft.

To account for this, the analysis replaces the solar beam angle computed in Section Error!
Reference source not found. with an array-limiting beam angle if the latter is a smaller value.
This represents the physical situation where the sun beam “overflows” the PV array from the
viewer’s perspective, and thus less glare is possible.

1 4 ∗ 𝐴 ∗ |cos 𝜃𝑟𝑒𝑓−𝑝𝑣𝑎 |
𝜃= √
𝑑 𝜋

where:

 A is area of PV array

 d is distance between observer and array

 θref-pva is angle between reflected sun vector and PV array normal

12 | P a g e
6. Annual Energy Production
SGHAT can also predict the annual energy produced by a PV array based on its capacity and
configuration. First the rated capacity is scaled at each time step based on the angle between the
panel normal and the position of the sun. E.g. when the sun vector is perpendicular to the panel
normal, the power produced will be the rated capacity.

𝑃 = 𝑃𝑟𝑎𝑡𝑒𝑑 ∗ cos 𝜃𝑖𝑛𝑐

Next, this power is multiplied by the scaled time increment to get the energy produced in that
time interval (kWh).
𝑡
𝐸𝑡 = 𝑃 ∗
60
Finally, the energy produced over the year is summed to get the maximum annual energy
produced by the array, assuming clear sunny skies each day.

𝐸𝑚𝑎𝑥 = ∑𝐸𝑡

7. Other Formulations
7.1 Flight Path Calculations
The flight path OP coordinates are computed based on the latitude and longitude of the selected
threshold, and the specified direction of the flight path.

First, the distance covered by one degree of longitude is calculated, based on the latitude.

𝑑 = 𝑟𝑎𝑑𝑖𝑢𝑠𝑒𝑎𝑟𝑡ℎ ∗ cos(𝑙𝑎𝑡𝑡ℎ𝑟𝑒𝑠 )

Next, the distance between each point in terms of latitude and longitude is computed based on
the direction and distance, in meters, of a degree of latitude or longitude. This is the Δlatitude
and Δlongitude between each OP.

402.3 ∗ cos 𝜃
∆𝑙𝑎𝑡 =
111325
402.3 ∗ sin 𝜃
∆𝑙𝑛𝑔 =
𝑑
Here d is the distance covered by one degree of longitude, 402.3 is ¼ mile in meters, and 𝜃 is the
flight path direction.

13 | P a g e
Finally, compute each OP position by augmenting the coordinates of the previous OP by Δlat
and Δlng.

The flight path heights of each OP are calculated based on the threshold height above ground,
glide slope and threshold elevation. The tangent of the glide slope, multiplied by the distance
from the threshold yields the height above the threshold of the given OP. Since each OP
elevation is queried automatically from Google, this height value is augmented by the elevation
difference between the OP and the threshold elevation.

ℎ𝑒𝑖𝑔ℎ𝑡 = 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 ∗ tan(𝜑) + ℎ𝑒𝑖𝑔ℎ𝑡𝑡ℎ𝑟𝑒𝑠 + (𝑒𝑙𝑒𝑣𝑎𝑡𝑖𝑜𝑛𝑡ℎ𝑟𝑒𝑠 − 𝑒𝑙𝑒𝑣𝑎𝑡𝑖𝑜𝑛𝑂𝑃 )

where 𝜑 is the glide slope.

14 | P a g e
8. Acknowledgments
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia
Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S.
Department of Energy’s National Nuclear Security Administration under contract DE-AC04-
94AL85000.

9. References
[1] Ho, C.K., C.M. Ghanbari, and R.B. Diver, 2011, Methodology to Assess Potential Glint
and Glare Hazards From Concentrating Solar Power Plants: Analytical Models and
Experimental Validation, Journal of Solar Energy Engineering-Transactions of the Asme,
133(3).
[2] Duffie, J.A. and W.A. Beckman, 1991, Solar engineering of thermal processes, 2nd ed.,
Wiley, New York, xxiii, 919 p.
[3] Ho, C.K., Relieving a Glaring Problem, in Solar Today2013, American Solar Energy
Society: Boulder, CO. p. 28 - 31.
[4] Weisstein, Eric W. "Reflection." From MathWorld - A Wolfram Web
Resource. http://mathworld.wolfram.com/Reflection.html
[5] http://en.wikipedia.org/wiki/Line-plane_intersection
[6] http://matplotlib.org/1.2.1/api/path_api.html#matplotlib.path.Path.contains_points
[7] Yellowhair, J. Reflectance Measurements on the Photovoltaic Sample Panels, October 7,
2013.
[8] Stine, W. B. and Geyer, M, 2001, Power From the Sun,
http://www.powerfromthesun.net/book.html

Additional references can be found at www.sandia.gov/glare.

15 | P a g e