An evaluation study of miniature dielectric crossed compound 1 parabolic concentrator (dCCPC) panel as skylights in building 2 energy simulation

13 The potential of miniature dielectric crossed compound parabolic concentrator (dCCPC) 14 panel as skylights for daylighting control has drawn a considerable research attention in the 15 recent years, owing to its feature of variable transmittance according to the sun position, 16 but the viability of using it as skylights in buildings has not been explored yet 17 comprehensively. This paper aims to study the feasibility of utilizing miniature dCCPC panel 18 as skylight in different locations under various climates in terms of energy saving potential 19 besides its daylighting control function. The transmittance of dCCPC panel varies at every 20 moment according to the sky condition and sun position. Due to this specific property, this 21 study novelly implemented a polynomial formula of the dCCPC transmittance in the 22 Grasshopper platform, from which EnergyPlus weather data can be called to calculate the 23 hourly transmittance data of dCCPC skylight panel throughout the whole year. An hourly 24 schedule of transmittance is generated according to the hourly sky condition determined by 25 the daylight simulation through Radiance and Daysim, and is then input to EnergyPlus 26 simulation to predict the energy consumption of a building with dCCPC skylight. Fourteen 27 locations around the world are therefore compared to find the most appropriate place for 28 using miniature dCCPC panel as skylights. The energy saving in cooling, heating and lighting 29 with use of dCCPC skylight panel are investigated and compared with low-E and normal 30 double glazing. The results show that the dCCPC skylight panel can reduce cooling load by 31 mitigating solar heat gain effectively although its performance is affected by several criteria 32 such as sky conditions and local climates. It is generally more suitable for the locations with 33 longer hot seasons, e.g., Log Angeles, Miami, Bangkok and Manila, in which dCCPC could 34 provide up to 13% reduction in annual energy consumption of building. For the locations 35 having temperate and continental climates like Beijing, Rome, Istanbul and Hong Kong, a 36 small annual energy saving from 1% to 5% could be obtained by using dCCPC skylight panel. and air The results highlight a very good agreement and high-level matching between simulation and measured results. In the study provided by Dahanayake


44
The energy consumption in buildings takes more than one-third of total global energy 45 consumption (Lowry, 2016). The electricity required by artificial lighting is one of the main 46 parts of the energy demand for buildings. In the solar heating & cooling (SHC) programme in 47 2015 held by the international energy agency, it was stated that the lighting energy took 19% 48 (2900 TWh) of the total global electricity consumption approximately, and it is estimated to 49 reach 4250 TWh by 2030 under current policies (Attia et al., 2017, SHC, 2015. Daylighting 50 design is a popular choice in modern building design with the considerations of energy 51 saving, visual comfort and hence occupant health. The combination of direct sunlight and 52 diffuse skylight are regarded as daylight whose quality and intensity varies depending on the 53 location, season, time, weather, sky condition and so forth. With an appropriate daylighting 54 design, about 40% lighting energy could be saved (Dubois and Blomsterberg, 2011), and this 55 could even reach 70% with the proper designs of space type and control type (Ahadi et al.,56 2017). As a passive solar energy application, daylighting is accompanied with solar heating 57 which can reduce the heating load in winter to some extent. It was also found by many 58 researchers that daylight is good for human health by curing medical ailments and reducing 59 psychological sadness related to the seasonal affective disorder (Hraska, 2015, Wong, 2017, 60 Liberman, 1990. In a survey conducted by Hourani et al. (Hourani and Hammad, 2012), 61 more than 80% of the working staffs were willing to sit by windows and similar results were 62 obtained from the student and patient groups. Daylight also results in the better perception 63 and higher productivity for occupants (Sivaji et al., 2013, S. R. Kellert et al., 2008). 64 As one type of the nonimaging optics, compound parabolic concentrator has been 65 attempted to be utilized in building facade for daylighting application in the past decades. 66 Walze et al. (Walze et al., 2005) proposed two kinds of smart windows with the 67 microstructure of two dimensional (2D) compound parabolic concentrator (CPC) array on 68 the surface, which focused on preventing unnecessary solar radiation and improving light-69 guiding abilities. Yu et al. (Yu et al., 2014) investigated the feasibility of 2D dielectric CPC in 70 daylighting control as it is used as a skylight and found that the transmittance of the 71 stationary CPC varies with the sun positions, which is lower at noon and larger in the 72 morning and afternoon. Li  skylight panel on the energy performance of a building will be investigated in this paper to 97 evaluate its implication and suitability in actual applications. 98 As is known, the transmittance of a dCCPC panel varies with sky condition and sun position, 99 which means that it would not be a constant value for different time points. A polynomial 100 formula for their relationship has been obtained in our previous study (Tian and Su, 2018a  constructions provided by EnergyPlus, respectively. The default constructions may not be 174 the best selections for the purpose of energy saving for building, but can be considered as 175 the constructions with average performances that are more suitable for analysing the effect 176 of skylights in different climates. Similarly, the heating and cooling load in simulations are 177 calculated by using the ideal loads air system template, which aims to focus on the variation 178 of thermal load caused by skylights rather than different air-conditioning systems. The 179 heating set point is 21°C and cooling set point is 24°C. It is important to mention that, the 180 control types of artificial lighting for all models are the same, which is auto dimming and it 181 will be switched off when there is no occupancy in the room. The sensor points of lighting 182 and lighting control are located in a 13 5 array detecting the illuminance level of working 183 plane. The set point of lighting is 500lux. Shading and glare control are not considered for 184 windows and skylights. 185

Skylights model description 188
In order to investigate the effect of dCCPC panel on building energy performance, three 189 types of skylight panels as listed in Table 1

Location 212
In order to investigate the performance of dCCPC skylight panel in different locations and 213 climates, 14 cities are chosen for energy simulation of the example office building. The 14 214 cities in Table 2   the whole simulation of this study. 247 In our previous study (Tian and Su, 2018a), a multiple nonlinear regression model, as shown 248 in Eq. (1), has been proposed to correlate the transmittance of a horizontal dCCPC with the 249 altitude and azimuth angles and sky clearness factor, and the coefficient of determination 250 (R 2 ) is up to 0.944. However, when a dCCPC panel is used as skylights, its tilt angle should be 251 adjusted according to the local latitude to maximise solar utilization. In order to fit this 252 regression model, the equivalent altitude and azimuth angles and equivalent sky clearness 253 factor with reference to a tilted surface are proposed and applied to calculate the 254 transmittance of dCCPC used in the building energy simulation under given sky conditions in 255 this study, as expressed in Eq.
(2). This section introduces how to calculate those and an 256 example of the whole process of calculating the transmittance of dCCPC in a specific 257 moment is given. 258 (1) 259 Where is altitude; is azimuth; is sky clearness factor; is the transmittance of 260 dCCPC; are regression coefficients; is the transmittance of dCCPC under 261 overcast sky. 262 (2) 263 Where is equivalent altitude (expressed in radian measure), ° °; is 264 relative equivalent azimuth (expressed in radian measure), ° °, and 265 ° when the incident plane to the entry aperture of dCCPC is parallel to either side of its 266 square entry aperture; is equivalent sky clearness factor. 267

Description of coordinate system 268
For the purpose of calculating the equivalent altitude and azimuth angles of dCCPC, a 269 coordinate system is applied as illustrated in Fig. 5. The south, east and zenith directions are 270 represented by x, y and z axis respectively. The incident sunlight is denoted by vector .

271
The actual altitude and azimuth are indicated by and . To obtain the best result of 272 controlling daylight by dCCPC, the dCCPC would be tilted to the south when it is applied in 273 the northern hemisphere. The entry aperture (top surface) of dCCPC, which is also the 274 interface between air and dielectric material, is denoted by the plane ABCD. The plane ABCD 275 is south-tilted by from the horizontal plane, which stands for the tilt angle of dCCPC, and 276 which is also the angle between the surface normal line NN' of the plane ABCD and the z axis. : equivalent north direction of the plane 293 ABCD; : equivalent solar azimuth angle. 294

Calculation of equivalent altitude angle 295
It is assumed that the lengths of the vector and are 1. The coordinates of point S and 296 N can be expressed by: 297 and ; 298 The vector and can be defined as: 299 and 300 Then the angle between and , that is, the incident angle , can be calculated by: 301

Hence, the incident angle is 302
And the equivalent altitude of tilted dCCPC is: 303 π

Calculation of equivalent azimuth angle 304
For the right triangle SOE with hypotenuse SO, 305 In addition, because and are two parallel vectors, the vector of can be expressed 306 as: 307 The vector can be calculated by: 308 Where 309

Thus, 310
The vector is the equivalent north direction on the plane ABCD and the length of it is 311 assumed to be 1. The coordinates of point M is: 312 The angle is the equivalent azimuth angle on the plane ABCD, which is defined by 313 irradiance; is a constant and equals 1.041 for in radians; is equivalent solar zenith 342 angle in radians. The values of , and could be obtained as shown in Table 3. 343 The equivalent sky clearness factor is 3.98 according to Eq. (18). Therefore, the 344 transmittance of dCCPC can be calculated by Eq.
(2) and the value of transmittance is 0.72. 345 In addition, the transmittance obtained by Photopia simulation is 0.75, which provides a 346 good agreement with the calculated result. All of the values obtained in example calculation 347 are summarized in Table 3 below. 348  transmittance of dCCPC-DB and dCCPC-lowE varies as time goes on: the transmittance is 373 higher in the morning and afternoon, and it becomes lower at noon. The total solar heat 374 gain from skylight is affected by the transmittance significantly. For DB, the solar gain 375 becomes higher from morning to noon, and then drops down in the afternoon. For dCCPC-376 DB, the solar gain also goes higher from morning to noon and decreases in the afternoon, 377 but the solar gain is reduced at 11am, 12pm and 1pm due to the low transmittance at noon. 378 For dCCPC-lowE, the total solar gain is less than 10kWh for all the time and has similar 379 tendencies with dCCPC-DB. In terms of hourly solar gain, dCCPC-DB reduces more than half 380 of the solar gain compared with DB. The solar gain by dCCPC-lowE is about a quarter of 381 dCCPC-DB owing to the lower transmittance and SHGC. The solar gain also affects the total 382 thermal load. In Birmingham on 22 nd Jun, only cooling load is required. In Fig. 8, it is 383 important to note that the thermal load here indicates cooling load because only cooling is 384 required in this day. It can be seen that the demand of cooling starts from 11am and 385 becomes high in the afternoon. Due to the less solar gain through dCCPC-DB and dCCPC-386 lowE, the cooling load of using these two skylights are less than that of using DB except 7pm. 387 The reason is that at 19:00, outdoor illuminance becomes low and artificial lighting is 388 required for dCCPC-DB and dCCPC-lowE. Lighting causes more thermal load so that the 389 thermal load of DB is smaller at this time. For 12pm, 1pm and 2pm, when the solar gain from 390 dCCPC-DB and dCCPC-lowE are much less than DB, more than 1/3 of cooling requirement are 391 saved by dCCPC-DB and dCCPC-lowE compared to DB. The total cooling load savings of 392 dCCPC-DB and dCCPC-lowE are 14.5% and 30% respectively for the whole day of 22 nd Jun 393 comparing with double glazing (DB).

Monthly and annual thermal load 401
Based on the annual weather data and detailed model settings, the results of cooling and 402 heating load of the example building are obtained and compared in this section. Fig. 9(a) and 403 Fig. 9(b) illustrates the data of monthly cooling and heating loads when the building utilizes 404 double glazing (DB), double glazing with dCCPC layer (dCCPC-DB) and low-E double glazing 405 with dCCPC layer (dCCPC-lowE) as skylights. This radar chart is provided aiming to provide a 406 comprehensive idea of how dCCPC-DB and dCCPC-lowE affects cooling and heating loads 407 comparing with DB, that is, increase or decrease or stay same for different locations in 408 different seasons. The quantity of thermal load variations were given in Fig. 10  Birmingham and Helsinki, it can be found that the savings on heating load are not as much as 420 on cooling load, even the heating load after using dCCPC window is more than that of using 421 double glazing in some months. For the locations in which building needs cooling and 422 heating, like Los Angeles, Rome, Beijing, Shanghai and Istanbul, similar results are obtained.

423
The skylights with dCCPC layer can reduce cooling load in summer, and these reductions are 424 quite much in some specific months and locations, for example, the July, August and 425 September in Los Angeles, the July and August in Rome and Istanbul. Generally speaking, 426 dCCPC and low-E coating can reduce cooling load effectively, but the low SHGC can also lead 427 to the increase of heating load in cold seasons. Balances should be found to save the total 428 energy consumption on both cooling and heating for building. The annual thermal load for the sum of cooling and heating loads in the example building is 437 summarized in Fig. 10, in which the effects of the skylights with dCCPC layer on the total 438 thermal load are illustrated. The cities are arranged by climate category firstly. The climates 439 are ordered from low to high altitude. In each climate type, the cities are ordered by the 440 time percentage of clear sky from long to short. As is known from Fig. 9(a) and Fig. 9(b), the 441 effects of dCCPC is mainly on reducing cooling load by preventing solar heat gain. On the 442 contrary, it will also result in increasing heating load. Thus, after combining the variations on 443 heating and cooling load, it provides different results compared to the result of either 444 cooling or heating shown in Fig. 9(a) and Fig. 9(b). It was found that the thermal loads have 445 slightly decreases (1%-3%) for the cold locations, like Helsinki, Kiruna and Aberdeen, which 446 may be not suitable for using dCCPC. For the locations having cold winter, such as Beijing 447 and Birmingham, heating takes more than half of the total thermal load, the reduction in 448 thermal load by dCCPC are quite low (< 5%). In these locations, cold seasons are long and 449 solar gain from window are expected to be as much as possible in winter to reduce heating 450 load. It is important to point out that Lhasa is an exception among cold locations in which 451 the thermal load of building is decreased after using dCCPC. Although most of the time 452 during the whole year in Lhasa is cold, the clear sky takes about 65% of daytime during the 453 whole year so that the annual solar radiation reaches 7.2GJ/m 2 which is extremely strong 454 (Wu et al., 2015). Form the annual cooling load, it can be seen that using dCCPC-DB and 455 dCCPC-lowE reduces 10% and 24% cooling load respectively compared to using traditional 456 double glazing. They also lead to reductions in heating load in winter time. The reason is 457 because the dCCPC layer causes lower transmittance of skylights so that more artificial 458 lighting is required. The thermal energy from lighting offsets some requirements for heating. 459 For the locations having long hot seasons, the window with dCCPC provides outstanding 460 performance of reducing total thermal load. Use of dCCPC-lowE reduces up to 23% of annual 461 thermal load compared with DB for Los Angeles, from 10% to 14% for Hong Kong, Rome, 462 Miami, Bangkok and Manila. The reduction in heating and cooling load by dCCPC-DB also 463 ranges from 5% to 10% for these locations. 464 465 Fig. 10. Annual thermal load of the example building with dCCPC-DB, DB and dCCPC-lowE as 466 skylights, respectively 467

Energy consumption of artificial lighting 468
Although dCCPC provides effective daylight control, when it is integrated with standard or 469 low-E double glazing, its transmittance is smaller than that of traditional double glazing. 470 Thus, more artificial lighting may be required to guarantee the indoor illuminance level. The 471 annual electricity demand of artificial lighting is demonstrated in Fig. 11, together with the 472 percentage of relative difference of lighting consumption between using dCCPC-DB and DB 473 as skylights. Because the difference in the amount of annual lighting energy consumptions 474 between using dCCPC-DB and dCCPC-lowE for each city is quite small and less than 3%, the 475 percentage difference of using dCCPC-lowE is not shown in Figure. It can be seen that the 476 lighting energy consumption is increased by about 6% when using the skylights with dCCPC 477 layer in general, except for Beijing. It has been discussed that dCCPC has the advantage of 478 diffusing incident light. When the sun is in lower position, traditional double glazing cannot 479 provide a relatively large bright-area, but the dCCPC could lit larger space through diffusing. Kong has the opposite condition with Beijing: during the time when sun is low, more of the 488 sky conditions in Hong Kong is likely to be overcast, and light is prevented by dCCPC causing 489 much more demand on lighting. It is also important to mention another exception of Lhasa. 490 Lhasa has strong direct sunlight and long-time clear sky conditions (about 65%). Although 491 the outdoor illuminance will be extremely high sometime, e.g. 90klux, it is still rare case. 492 Thus, dCCPC performs low transmittance, e.g. 0.3-0.4, during these time periods so that 493 much more lighting is needed. However, shading requirement is not considered in this 494 simulation. But it can be speculated that the normal double glazing can provide extreme 495 bright indoor environment as well as the very high indoor illuminance level in Lhasa, and 496 shading should be a necessary requirement to provide a comfort visual environment. The 497 energy consumed by artificial lighting should be larger than the results presented under such 498 circumstances. 499 Fig. 11. Annual lighting energy consumption of the example building with dCCPC-DB, DB and 501 dCCPC-lowE as skylights, respectively 502 The energy consumption of a building mainly consists of electricity usage of artificial lighting, 503 electricity usage of equipment and energy consumption of heating and cooling system. As 504 discussed in previous sections, dCCPC can reduce total thermal load but increase lighting 505 usage, and the variation of lighting caused by dCCPC can also lead to the change of thermal 506 load. It is important to investigate the interactions among different energy usage sectors. In 507 the energy simulations in this study, it is assumed that all of the systems and schedules are 508 same. Thus the electricity usage of equipment is assumed to be same for different locations. 509 The lighting and heating/cooling energy demands are the only two aspects that should be 510 considered to evaluate the performance of using the dCCPC skylights. Fig. 12 shows the 511 comparisons of the total energy consumptions of lighting, cooling and heating when utilizing 512 DB, dCCPC-DB and dCCPC-lowE as skylights. It can be found that for the locations with long 513 hot seasons such as Los Angeles, Miami, Bangkok and Manila, a considerable reductions of 514 up to 13% (dCCPC-lowE) and 8% (dCCPC-DB) occur in total energy consumption. A small 515 reduction of 1%-5% can be obtained by utilizing dCCPC for the locations having temperate  Table 4 demonstrates the values of simulation and measured results of dCCPC skylights 534 under different sky conditions. It was found that almost all of the deviations between 535 experiment and simulation results are smaller than 10%, only the deviation at 10:50am are 536 about 16% which may be caused by the occasional experimental error. The root-mean-537 square-error (RMSE) of the two data sets are 3.33% and 2.89% respectively, which are quite 538 small and can prove the precision and reliability of the transmittance prediction model.  was validated in an outdoor experiment with a good accuracy, and also those building 592 energy simulation software packages have proved accurate enough, therefore, 593 incorporation of the proposed mathematical model in the building energy simulation 594 software can offer a cost effective way to evaluate the viability of dCCPC skylights in 595 buildings. It will be ideal to be followed by the field test of dCCPC skylights in a real building, 596 but due to the resource constriction, it is a regret that a corresponding experiment was 597 unable to be implemented in the current study. However, it is expected and recommended 598 to proceed this field test in a further work. 599 The key findings of this paper can be summarized into following points: 613 1) In general, dCCPC panel as skylights can reduce cooling load due to effectively 614 mitigating solar heat gain. However, it also causes increases of heating load and 615 artificial lighting energy consumption. The energy performance of a building with dCCPC 616 skylights is also related to the local climate conditions such as solar irradiation and 617

Conclusion and recommendation 600
temperature. 618 2) The dCCPC skylight is more suitable for the cities having long summer time, such as 619 Bangkok, Manila, Miami, and Los Angeles. The reduction of thermal load is up to 23% 620 and the total energy saving could reach 13%. 621 3) The dCCPC skylight is more effective under clear sky conditions. For example, Los 622 Angeles (23% reduction of thermal load) is the best choice for using dCCPC due to its 623 longest period of clear sky among the cities with long hot seasons. 624 4) For the cities with continental climates, only the place with prevalent clear sky is 625 appropriate for using dCCPC skylight. For instance, in Beijing, Rome, Hong Kong and 626 Shanghai, dCCPC could decrease the annual thermal load by 3% to 10%. Considering the 627 lighting energy consumption, the total energy saving ranges from 1% to 5% in these 628 cities. 629 5) The dCCPC skylight is not suitable for the cities with long cold seasons, e.g. Aberdeen, 630 Birmingham, Helsinki and Kiruna. The reduction of solar gain by dCCPC leads to more 631 energy consumption in heating load and artificial lighting. Using dCCPC in these cities 632 leads to 1%-5% increase of total annual energy consumption. 633 6) In terms of optical properties, dCCPC is recommended for all locations for the purpose 634 of glare control, especially for the cities with strong solar radiation. 635 The further work about dCCPC is suggested to be proceeded in the following aspects. Firstly, 636 different shading devices should be considered and glare analysis are recommended to be 637 taken to evaluate the dCCPC effects on indoor visual environment comparing with 638 traditional glazing, and then the energy analysis in this study could be updated by 639 considering various shading devices. Secondly, an experiment implemented in a real building 640 was highly recommended to verify the simulated effect of dCCPC skylight on building energy 641 and visual environment. Thirdly, considering the great potential of utilizing dCCPC as 642 skylights in diffusing direct sunlight and energy saving of building, the asymmetric dCCPC is 643 suggested for investigating its feasibility in daylighting control as vertical building facade. 644 Finally, the economic analysis of dCCPC could be taken to evaluate its viability in practical 645 application.             Table  Table 3 Term Calculation formula Value of example Step No.