- IAM
- fraction of the direct-beam light that enters the module relative to normal incidence, from 0 to 1
- θ
- angle of incidence between the sun beam and the module normal, in degrees (0° to 90°)
- Γ
- fitting parameter, greater than 1. The larger it is, the more slowly the IAM falls as the angle grows
- It is exactly 0 at θ = 90°.
- It approaches 1 at normal incidence. At θ = 0° the value is 1 − Γ⁻⁹⁰, about 0.9999 for Γ = 1.10984.
- It has a single parameter, and the calculation is one power and one subtraction.
Where the equation comes from
A PV module is directional: light that arrives at a steep angle yields less. The human eye is too, a behavior known as the Stiles–Crawford effect, which is described by an exponential with a single coefficient. The equation was inspired by it. In the paper it is fitted to IAM measurements from the literature, alongside the physical, ASHRAE, Sandia and Martin–Ruiz models.
Γ values fitted in the paper
Average of measurements from nine laboratories on three cell technologies: mono-Si, BSi RIE and BSi ADE.
Average of outdoor measurements of commercial modules from five manufacturers: Canadian, JA Solar, Q-Cells, Yingli and First Solar.
Compare it with other models
The ASHRAE (b₀) and Martin–Ruiz (aᵣ) parameters are the ones fitted in the same paper. The dots are the measured laboratory average (Table II of the paper). Drag over the chart or use the angle control.
- Goncalves-Malagón
- 0.9561
- Martin–Ruiz
- 0.9554
- ASHRAE
- 0.9492
Show the values as a table
| θ (°) | Goncalves-Malagón | Martin–Ruiz | ASHRAE | Measured (lab) | SD (%) |
|---|---|---|---|---|---|
| 0 | 0.9999 | 1.0000 | 1.0000 | 1.000 | 0.02 |
| 5 | 0.9999 | 0.9999 | 0.9998 | 1.000 | 0.04 |
| 10 | 0.9998 | 0.9998 | 0.9992 | 1.000 | 0.06 |
| 15 | 0.9996 | 0.9995 | 0.9982 | 1.001 | 0.08 |
| 20 | 0.9993 | 0.9990 | 0.9967 | 1.000 | 0.17 |
| 25 | 0.9989 | 0.9983 | 0.9947 | 1.001 | 0.16 |
| 30 | 0.9981 | 0.9972 | 0.9921 | 0.999 | 0.27 |
| 35 | 0.9968 | 0.9956 | 0.9888 | 0.999 | 0.26 |
| 40 | 0.9945 | 0.9930 | 0.9845 | 0.995 | 0.39 |
| 45 | 0.9908 | 0.9890 | 0.9789 | 0.991 | 0.42 |
| 50 | 0.9845 | 0.9827 | 0.9718 | 0.983 | 0.54 |
| 55 | 0.9739 | 0.9724 | 0.9622 | 0.971 | 0.63 |
| 60 | 0.9561 | 0.9554 | 0.9492 | 0.952 | 0.77 |
| 65 | 0.9261 | 0.9270 | 0.9306 | 0.921 | 0.96 |
| 70 | 0.8756 | 0.8790 | 0.9022 | 0.871 | 1.26 |
| 75 | 0.7905 | 0.7971 | 0.8545 | 0.790 | 1.66 |
| 80 | 0.6473 | 0.6565 | 0.7582 | 0.660 | 2.63 |
| 85 | 0.4061 | 0.4148 | 0.4677 | 0.469 | 5.34 |
| 90 | 0.0000 | 0.0000 | 0.0000 | — | — |
Fit quality in the paper
Coefficient of determination (R²) of each model when fitted to the measured averages, as reported in Table III of the paper.
| Model | Laboratory | Field |
|---|---|---|
| Simplified physical (K = 4 m⁻¹) | 0.984 | — |
| Complete physical | 0.997 | — |
| ASHRAE (b₀) | 0.840 | 0.973 |
| Martin–Ruiz (aᵣ) | 0.985 | 0.993 |
| Sandia (6 coefficients) | 0.995 | 0.997 |
| Goncalves-Malagón (Γ) | 0.988 | 0.994 |
According to Table III of the paper, the complete physical model fits best in the laboratory (0.997), but it needs the optical properties of each layer. Among the one-parameter models, the authors report the highest R² for Goncalves-Malagón in both datasets; in the field its margin over Martin–Ruiz is small (0.994 and 0.993). These are the values reported by the authors, and they depend on the fitting procedure and angle range used in the paper.
How to use it
Use the equation from 0° to 90° and zero from 90° on. The examples already handle this and use the laboratory Γ.
import numpy as np
def iam_goncalves_malagon(aoi_deg, gamma=1.10984):
"""IAM = 1 - gamma**(aoi - 90), aoi in degrees; 0 from 90 degrees on."""
aoi = np.abs(np.asarray(aoi_deg, dtype=float))
iam = 1.0 - gamma ** (aoi - 90.0)
return np.where(aoi >= 90.0, 0.0, np.maximum(iam, 0.0))function iamGoncalvesMalagon(aoiDeg, gamma = 1.10984) {
const aoi = Math.abs(aoiDeg);
return aoi >= 90 ? 0 : Math.max(0, 1 - Math.pow(gamma, aoi - 90));
}=IF(theta>=90,0,1-GAMMA^(theta-90))If you have measurements of your own module, fit Γ to them. The paper's values come from the averages described above.
What else the paper does
Besides proposing the model, the paper uses the IAM to calculate the annual angular losses (AAL) and the energy gain of single- and two-axis solar trackers, at locations across several latitudes and climate zones.
For horizontal surfaces, the annual angular loss ranged from about 3% (low latitudes) to 6% (high latitudes). With two-axis trackers, the lowest loss was about 1.3%. The energy gain calculated with the IAM is higher than the one calculated without it.
Where it is already used
In the ElétricPro photovoltaic generation simulator, Goncalves-Malagón is the default IAM model, with a selector for ASHRAE and an editable Γ.
Limits and care
- It is an empirical model. The Γ values in the paper were fitted to the datasets described above.
- In the paper the IAM is applied to the direct (beam) component of the radiation. For diffuse sky light and ground-reflected light the IAM has to be integrated over the hemisphere, for example with the numerical method of Marion (2017).
- The formula does not hold outside 0° to 90°. At 90° or more, use IAM = 0.
How to cite
If you use the model, please cite the paper:
GONÇALVES, Ailton; RATIVA, Diego; GÓMEZ-MALAGÓN, Luis Arturo. Model-Based Assessment of the Incident Angle Modifier on the Annual Angular Losses and Gain of PV Modules in Tracking Systems. IEEE Journal of Photovoltaics, v. 14, n. 1, p. 185–193, jan. 2024. DOI: 10.1109/JPHOTOV.2023.3323802.A. Gonçalves, D. Rativa, and L. A. Gómez-Malagón, “Model-based assessment of the incident angle modifier on the annual angular losses and gain of PV modules in tracking systems,” IEEE J. Photovoltaics, vol. 14, no. 1, pp. 185–193, Jan. 2024, doi: 10.1109/JPHOTOV.2023.3323802.@article{goncalves2024iam,
author = {Gon{\c{c}}alves, Ailton and Rativa, Diego and G{\'o}mez-Malag{\'o}n, Luis Arturo},
title = {Model-Based Assessment of the Incident Angle Modifier on the Annual Angular Losses and Gain of {PV} Modules in Tracking Systems},
journal = {IEEE Journal of Photovoltaics},
volume = {14},
number = {1},
pages = {185--193},
year = {2024},
month = jan,
doi = {10.1109/JPHOTOV.2023.3323802}
}