PRC Krochmann Sky Scanner Characterisation (parts 1 & 2)

Transcription

1 PRC Krochmann Sky Scanner Characterisation (parts 1 & 2) Pierre Ineichen * University of Geneva - GAP 1231 Conches - Switzerland * ISES Member FIRST DRAFT IEA Task XVII expert meeting December June 1992

2 PRC Krochman Sky Scanner Caracterisation - part #1 Pierre Ineichen University of Geneva - GAP/CUEPE 158 rte de Florissant Conches - Switzerland December 1991 Instrument The PRC Krochmann Sky Scanner is a photometer for the sky luminance distribution measurement. The sky luminance is measured in 145 equal solid angles with an aperture angle of 10 ; the zenith luminance Lvz is recorded 6 times. Stability during the scan The sky scanner offers the possibility of taking 150 measurements of the same point of the sky dome. We made the measurement for different azimuths and elevations and reported the results on figure 4a, b and c. On these figures, one can see that the fluctuation of the measured values is within 0.2%. For the first point measured in the meridian plane of the sun (a=0 or a=180 ), the measurement error can reach 1-2% Figure 1 Extract of figure 4b. Precision of the measurements. During the scan, the sky scanner measures six times the zenith luminance : scan # 145 to 150, the position of measurement #1 is in the meridian plane of the sun. We reported on figure 5a & b the relative difference between measurements #145 to 150 and the mean zenith luminance value (Lvz mean = Σ Lvz( )/6). These representations are splitted for different values of sun height and clearness index Kt, different values of sky clearness (epsilon) and sky brightness (delta), and for 2676 series of measurements. For example, on figure 5a, the points reported on the upper left graph are morning and evening measurements (low sun height) and for overcast meteorological conditions (low clearness index) Each point on these graphs is a mean relative difference between measured Lvz value (pts ) and Lvz mean; the bars are plus and minus one root mean square difference. Figure 5b gives the same representation but for different sky clearness and brightness - 1 -

3 values. For example, the upper right graph is a bright overcast sky, the lower left graph is a very clear blue sky. Depending on meteorological conditions, the relative mean difference between the measured points ( ) and Lvz mean can reach 4-6%, the root mean square difference up to 10%. These differences occur principally under intermediate and bright luminous skies. In order to understand why such a dispersion occurs, we consider the value measured just before and just after the point 145 to 150 during the scan. Figure 2 shows the path of the scan. The sequences are the following : first measurement : nd measurement : rd measurement : th measurement : th measurement : th measurement : Figure 2 Numbering of the measurements during the scan. Figures 6a & b show the sequences for each point ( ), with the previous and the next points. All the points are normalized to Lvz mean, with plus and minus one root mean square difference bars. Figure 3 Extract from figure 5a & 5b for a particular value of h and Kt'

4 The h, Kt, epsilon and delta distributions are the same as for figures 5a & b If one looks at the previous graph (extracted from figure 5a & 6a), one can see that the value preceding the point #145 has a high luminance value and that the point #145 is overestimated. The situation is the inverse for the point #146. We made the same representation for different ambient temperatures, and for clear sky conditions (Kt > 0.5) : The temperature s effect is not significant. The high and low temperatures correspond to different seasons and consequently to different meteorological conditions, even if Kt is greater than 0.5. The dispersion is higher for low temperature values. Conclusion It seems that the system measurements+electronics is influenced by the previous value, and underestimates or overestimates the actual measurement. This effect is certainly not due to the measuring cell : it is a silicium cell, which has a short response time. Notation Kt = Kt / [1.031 exp(-1.4/( /m)) + 0.1] Epsilon = [(Dh + I)/Dh z3]/[ z3] z = sun zenith angle Delta = Dh m / Io Dh = horizontal diffuse I = normal beam m = optical air mass - 3 -

5 Figure 4a Static measurement for different height and azimut of the sky scanner.the height is in degrees over the horizon, the azimuth in degrees from north and the luminance in [cd/m2]

6 Figure 4b Static measurements for different heights and azimuts of the sky scanner.the height is in degrees over the horizon, the azimuth in degrees from north and the luminance in [cd/m2]

7 Figure 4c Static measurements for different heights and azimuts of the sky scanner.the height is in degrees over the horizon, the azimuth in degrees from north and the luminance in [cd/m2]

8 Figure 5a Relative difference between point # and Lvz mean for different sun height and Kt values. Figure 5b Relative difference between point # and Lvz mean for different delta and epsilon value

9 Figure 6a Relative difference between point # and Lvz mean, with the preceding and following point for different sun height and Kt value. Figure 6b Relative difference between point # and Lvz mean, with the preceding and following point, for different delta and epsilon value

10 PRC Krochman Sky Scanner Caracterisation - part #2 Pierre Ineichen, Benoît Molineaux University of Geneva - GAP/CUEPE 158 rte de Florissant Conches - Switzerland June 1992 Introduction In a first note, we analyzed the response time of the PRC Sky Scanner: the scan duration was too short with regards to the electronic. This problem was solved by the manufacturer by changing the operational amplifier in the head of the scanner in January 1992 (the time for a scan is 40", the time from point #1 to point #150 is 27"). Instrument The PRC Krochmann Sky Scanner is a photometer for the sky luminance distribution measurement. The sky luminance is measured in 145 equal solid angles; the zenith luminance Lvz is recorded 6 times. Position of the measured point. During our sky scanner data analysis, we found out that the position of the step motors was not correct. The worst results were obtained in December Figure 1 shows two scans, measured on December 6 and December 8, Figure 1 : Sky scan for December 6 and December 8, 1991 On the first scan, one can see the calculated sun position (the scanner calculates the sun position and skips the measurement in this direction, the value is put to [cd/m 2 ]), and the effective measured sun position (the values are saturated at [cd/m 2 ]): it seems that in this case, the horizontal axis motor is missaligned of

11 On the second scan, the error seems to be on the vertical axis motor and the right half part of the scan is probably a measurement of the luminance below the horizon. These are very bad measurements and it is easy to eliminate them; but generally, the errors are not so obvious. To isolate the poor measurements, we analyzed the 6 zenith luminance measurements [1]. For each scan, we calculate the mean zenith luminance as the average of the six zenith measurements' points (Lvzm = Σ Lvz( )/6). We report on Figures 2a and 2b the relative difference between each zenith luminance (pt 145 to 150) and Lvzm for each scan (one measurement every two minutes). We choose data for February 2 and 9; both are clear days (Gh 400 [W/m 2 ], Dh 160 [W/m 2 ], Bn 500 [W/m 2 ] at 13h, solar time). The full scale in Figure 2a is ±30% and in Figure 2b ±10%. Figure 2a : February 2, Variation of the relative zenith luminance versus measurement point

12 At noon, the distance between the sun and the zenith is minimum and the amplitude of the relative error is maximum. This is due to the wrong position of the measurement direction. Also at noon, the sky scanner head turns itself by 180 in azimuth and this explains the change in the error pattern. Figure 2b shows that this lack of precision in the position of the motors is not systematic, and that we have measurements with very low relative differences. Figure 2b : February 9, Variation of the relative zenith luminance versus measurement point The zenith's luminance variation in the morning is due to clouds in the first part of the day. If clouds are crossing the zenith region of the sky during the 30 seconds measurement period, we will have such variation in the luminance

13 Aperture angle of the Sky Scanner We took measurements of the aperture angle of the PRC Sky Scanner using two different experimental setups. First we used the sun simulator of the "Ecole Polytechnique Fédérale de Lausanne" [2]. It is a 8m. high, 5mx5m room; the walls are covered with black velvet, the floor with black carpeting. The light source is a 2500 Watts arc-light with a fresnel lens to concentrate the beam. The homogeneity of the beam is better than 3.5% on a one square meter surface. Our tests were done in the center of the beam, within a 1 cm 2 surface; the photometers were mounted on a two axes' computer driven table. The aperture angle of the source is 2.8 Secondly, we used a 1000 Watts halogen light of 7.5 cm of diameter and situated it at 4.5m from the scanner. This setup gives an aperture angle of We made the measurements in different ways: scanner on the floor, source at the zenith and variation of the height of the scanner head, scanner on the table, tilted at 30 and variation of the height of the scanner head, scanner on the table, tilted at 30, scanner at fixed position and variation of the tilt angle of the table, scanner in horizontal position, source at 183 of azimuth and 3 of elevation, variation of the scanner azimuth, and source at 154 of azimuth and 20 of elevation, variation of the scanner azimuth and elevation. For each condition, the results were very similar, and we obtained the following sensitivity of the scanner (source at 154 of azimut and 20 of elevation) : 120.0% 100.0% Relative luminance Relative sensitivity 80.0% 60.0% 40.0% 20.0% Cumulated integral % M eas urement direction with regard to the s ource One can see that the sensitivity curve is not symmetric. An error of ±1 in the scanner head position can result in a ±5% variation of the measured signal. The aperture angle is 7.5, the limit angle is 5 and the slope angle cannot be defined

14 Situated at 9 from the zero position (zero = direction of the source) is a small peak of sensitivity, which can reach 2% : 2.5% Relative sensitivity 2.0% 1.5% 1.0% 0.5% R elative s ens itivity 0.0% M eas urement direction with regard to the source This looks unimportant, but if we consider a point situated at 15 from the sun position, on a clear day, one can measure a luminance of 20'000 [cd/m 2 ]. 9 from this point, in the sun direction, the luminance reaches 60'000 [cd/m2]. This means that 2% (of 60'000 [cd/m 2 ]) additional luminance to the 20'000 [cd/m 2 ] represents 6% of the measurement. The extreme case is at 9 from the sun direction where we have luminance values of 40'000 [cd/m 2 ]; here a 2% of the "sun luminance" represents more than 15%. If we consider that the modeling of the sky luminance distribution is critical at small zeta angles (near the sun), this is not a negligible problem. We have here a parasite luminance that can reach 15% of our measurement, certainly due to an internal reflection in the sky scanner optic. Conclusion In the actual configuration, it is difficult to do acquisition with a good precision. The uncertainity of the measurement's position is to high. With regard to the IDMP Guide which specifies an aperture angle of 10, the measurements aperture is here to limited. It is also difficult to evaluate the consequences of the peak of sensitivity situated at 9 from the measurement direction. References [1] PRC Krochman Sky Scanner Characterisation - part #1. P. Ineichen. GAP - CUEPE, University of Geneva. December [2] Cosine dependence of PRC Krochmann photometers. P. Ineichen, Benoît Molineaux. GAP - CUEPE, University of Geneva. June [3] Aperture angle of Normal Beam Photometers (NBP). P. Ineichen, Benoît Molineaux. GAP - CUEPE, University of Geneva. June