For monitoring the photovoltaic performance of the system, a separate prototype with a hybrid absorber with polycrystalline silicon cells and a reflector was constructed.

The optical efficiency g(a) is defined as the ratio between the performance of the concentrating module and a vertical module of the same area as the concentrating aperture. It was determined through outdoor measurements. The short circuit current Isc of the concentrator module was monitored as a function of the angle of incidence p in the meridian plane. The optical efficiency (figure 6), was then derived according to

Isc -1000

I1000 • • G ■ cos(fl)

where 11000 is the short circuit current of the bare module at an irradiance of 1000 W/m2 at normal incidence, Cg is the geometrical concentration of the concentrator system, p is the angle of incidence of beam irradiance, and G is the global intensity perpendicular to the sun.

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

The measurements were performed during high irradiance and with a diffuse fraction of around 10%. The concentrator accepts all irradiance for solar altitudes exceeding 15° in the meridian plane, which means that the diffuse optical performance of the concentrator will be similar to that of a module tilted 20° with a correction for reflectance losses. This further means that the optical acceptance of diffuse irradiance will be around 70% of the beam efficiency. For this reason, the global intensity can be used in Eq (3) without significantly increasing the error of the model.

The optical efficiencies are functions of the projected angle of incidence in the transversal plane (i. e. the north south vertical plane) and the transmission of the glazing is given as a function of the conventional angle of incidence. Ray tracing represents the theoretical optical efficiency of the system at 85% reflectance. The graph labeled Optical efficiency — Isc in figure 6 contains contributions from measurements with corrections from ray tracing. The difference between measured values and ray tracing at 15°<0T<60° is due to resistive losses in the cells when the reflector is effective. The cells on the prototype absorber did not cover the whole width of the absorber, which meant that for angles above 40° the reflected beam partly missed the cell. The angulars above 40° are instead generated by ray tracing. The transmission of the glazing has also been included in the graph as it was used in the calculations of the annual output.

A simulation software, MINSUN (Chant and Hakansson 1985), estimated the annual output of electricity using the optical efficiencies at different angles of incidence. The model used to describe the incidence angle dependence of the system in MINSUN is defined by Eq. (4)

Пар, — RT (@T )fL (A )

RT describes the behaviour of the reflector as dependent of QT and fL the transmission of the window glass as dependent of 0, . QT is the projected angle of incidence in the transversal plane and в, is the conventional angle of incidence relative to the glass normal.

This model has previously been shown to describe the optical performance of an asymmetric compound parabolic reflector system such as this one well (Brogren et al, 2004).

The simulations show a 93% increase in electrical output for the concentrator module relative to the vertical reference module, which means that one square meter of this window annually would deliver 79 kWh of electric energy. The annual performance is 43% higher than that of an identical module tilted 20°.

The active area of the tested measured prototype covers only 87% of the total glazed area, which this has to be taken into consideration when an economical comparison is made with other systems. It is however possible to increase the active area of the window in a future full scale installation.

L. J. Yebra1, M. Berenguel2, M. Romero1, D. Martinez1, A. Valverde1

1L. J. Yebra, M. Romero, D. Martinez, A. Valverde, CIEMAT-Plataforma Solar de Almerla,
Apdo. 22, E 04200, Tabernas, Almerla, Spain, Phone: +34 950 387923, Fax: +34 950
365015, E-mail: luis. yebra@psa. es

2M. Berenguel, Universidad de Almerla. Departamento de Lenguajes y Computacion. Area de
Ingenierla de Sistemas y Automatica, Ctra. Sacramento s/n, La Canada, E 04120, Almerla,
Spain, Phone: +34 950 015683, Fax: +34 950 015129, E-mail: beren@ual. es

This work overviews some of the main activities and research lines that are being carried out within the scope of the specific collaboration agreement between the Plataforma Solar de Almerla-CIEMAT (PSA-CIEMAT) and the Automatic Control, Electronics and Robotics research group of the Universidad de Almerla (TEP197) titled “Development of control systems and tools for thermosolar plants" and the projects financed by the MCYT DPI2001-2380-C02-02 and DPI2002-04375-C03. The research is directed by the need of improving the efficiency of the process through which the energy provided by the sun is totally or partially used as energy source, as far as diminishing the costs associated to the operation and maintenance of the installations that use this energy source. The final objective is to develop different automatic control systems and techniques aimed at improving the competitiveness of solar plants. The paper summarizes different objectives and automatic control approaches that are being implemented in different facilities at the PSA-CIEMAT: central receiver systems and solar furnace. For each one of these facilities, a systematic procedure is being followed, composed of several steps: (i) development of dynamic models using the newest modeling technologies (both for simulation and control purposes), (ii) development of fully automated data acquisition and control systems including software tools facilitating the analysis of data and the application of knowledge to the controlled plants and (iii) synthesis of advanced controllers using techniques successfully used in the process industry and development of new and optimized control algorithms for solar plants. These aspects are summarized in this work.

A simple calculation shows the interesting result, that the cooling capacity in the ice storage is at similar level as in a lead battery based on both volume and weight.

One supplier of lead battery informs, that a 50 Ah, 12 Volts battery has the weight of 13,6 kg. The dimesions are 0.24*0.175*0.175 meters. The energy content of 50 Ah can be calculated as a specific energy content of 0.159 MJ/kg or 294 MJ/m3.

The cooling system will have a COP-value (coefficient of performance) of about 1.3 (Danfoss BD35F, CECOMAF-data for -15 °С, 2000 RPM). This will result in a specific cooling capacity of 0.206 MJ/kg or 382 MJ/m3. For the ice storage: the specific cooling capacity is identical to the melting heat of ice, which is 0.333 MJ/kg or 333 MJ/m3.

The conclusion is, that the specific cooling capacity of ice is 62 % higher compared to lead battery on basis of weight and 13 % smaller compared with lead battery based on volume. In reality, the ice storage outperforms the lead-acid battery, because the allowed daily cycling is less than the nominal 50 Ah, which corresponds to 100% depth of discharge.

To evaluate the influence of the rooftop surface cooling on the indoor heat load, we conducted a dynamic heat load calculation by the thermal circuit network method (Ishida et al., 1987). Furthermore, to evaluate the influence of the rooftop surface cooling on the atmospheric heat load, we conducted an unsteady heat conduction calculation of a one­dimensional multi-mass system on the roof using the backward relaxation method, which made it possible to predict more detailed data such as a cross-sectional temperature distribution. For the evaporation and the reflection of solar radiation on building outer surface, we used new sol-air temperature taking evaporation into consideration. Details of the SAT* is given in the following:

Qa= Qr-Qv-Qie Eq. (1)

Since Qr= Qs-Qi, Eq. (1) can be rewritten as follows:

A.(<30/<3z) = Qs-Qi-Qv — Qie Eq. (2)

Here,

Qv= ac(0s-0a) Eq. (3)

Qs= (1-p)Js Eq. (4)

equation (fs=a*0s+b) of 0S,

Qie = wTaw(fs-fa) = a* wTa, w(0s-0a) + w*l*aw(a*0a +b-fa) Eq. (6)

Organizing Equations (2)~(6), the following equation can be obtained:

-A.(<30/<3z) = at[(0a+0e*) — 0s] Eq. (7)

Here, however, at and 0e* will be

at = (1-K*CC)* e*ar+ ac+ a*w*l*aw Eq. (8)

0e* = [(1-p)Js — (1-K*CC)(1- Br) *eaTa4 — wTaw(b-fa)] / at Eq. (9)

In this study, (0a+0e*) is handled as the sol-air temperature taking evaporation into consideration (SAT*, hereafter). Fig. 5 shows the results for which SAT* was calculated using the hottest day the maximum air-conditioning load was generated (August 5) from the standard weather data of Tokyo. SAT* dynamically changes depending on the solar reflectance (p, hereafter) rather than the evaporation rate (w, hereafter). If the value is identical, the SAT* with a value of p that is greater than that of w takes precedence. Here, based on the concept of above SAT*, we propose the equivalent solar reflectance (p*, hereafter) to evaluate the effects of evaporation and the reflection of solar radiation. If equivalent outside air temperatures with evaporation in mind are equal, we considered the

effects of evaporation and solar reflection to be equal. In the following, we conducted a numerical analysis with p* as the main parameter.

Fig. 6, 7 shows the floor plan of the building and the rooftop slab structure in the indoor heat load calculation. For the subject, we envisioned an office building; the plan for the room was made based on standard office simulations of AIJ (Architectural Institute of Japan). The rooftop surface temperature can vary greatly depending not only on the surface finish but also the insulated state of the rooftop slab. For this reason, a calculation was made for the case with no insulation (thermal resistance: 0.51m2K/W) on the rooftop slab.

Then, after changing the thickness of the insulation material in five stages, a calculation was conducted for the case of internal insulation (thermal resistance: 0.87~5.88m2K/W) on the rooftop slab. Tab. 2 shows various input conditions to calculate the indoor heat load. Air­conditioning was set to be on between 8:00 and 18:00 and the fixed temperatures for air­conditioning and heating were set at 26°C and 22°C respectively. Inside the building, the heat generation of 25W/m2 from lighting and equipment was established. The convection heat-transfer rate on the indoor side and outdoor side was 11.1 and 25.0W/m2K respectively.

Numerical code AGLA allows the simulation of muti-layered facades including advanced elements, such as transparent insulation (TIM) and liquid-based collectors-accumulators im­plemented as layers of a facade. It is based in the one-dimensional discretization of all the domain in the predominant direction (horizontal for heat transfer from outdoors to indoor building) and vertical for heat transfer occured in air channel (at a double envelope facade). Empirical heat transfer coefficients are introduced to solve convective heat exchange. Ther­mal radiation is solved applying the radiosity method.

Outdoor conditions are introduced as mean monthly values of horizontal global radiation, minimum and maximum air temperatures, wind velocity and direction, and relative humidi­ty. From these data, instantaneous values are calculated considering for solar radiation an isotropic diffuse model. Outdoor conditions over the surface of the facade are calculated, and all the heat transfer mechanisms are solved by means of a transient implicit algorithm. Thermal balances are performed for each component, and global heat transfer balances must be satisfied for the whole facade in order to pass to solve the following time step. The period of simulation is typically of one year, and time step varies from five to ten minutes depending on the case to simulate.

Solar collectors including transparent insulation are treated as multi-layered semi-trans­parent walls, TIM is solved by means of a discrete ordinate method [10]. A multi-node model is applied to accounts for stratification in the water accumulator [4]. More detail about the mathematical models implemented in AGLA code can be found in [6].

Discomfort glare is one of the most important issues in the building planning process that is yet to be fully understood. Nowadays it is well known (Reinhart & Voss, 2002), that people tailor their sun shading devices to their visual needs. This means avoiding direct sunlight and glare at their workplaces, but also providing themselves with a view to the outside. The energetic impact of this typical behaviour is often underestimated. Modern architecture often applies highly glazed facades, so that a good sun protection is needed. In the planning phase, minimal total solar energy transmission values (g — values) referring to closed shadings are often used for the calculations — even if the user opens the shading for the view and the g-value could change significantly. The result from this is that overheating in summer occurs or the cooling energy required is much higher than expected.

As a conclusion from this, the main visual aspects of direct sun and glare should be taken into account in the planning phase of a building.

To avoid the direct sun on a workplace is only a geometrical problem and is quite easy to be solved. Much more difficult is the glare aspect.

Facility description:

The Plataforma Solar de Almeria gathers the most extended experience in testing volumetric absorbers. More than 20 different materials and geometrical configurations have been qualified since 1986 (Leon, 1996; Tellez, 2002). A versatile test bed allows both the operation of up-to 250 kW solar receivers under real conditions of concentrated solar flux and air mass flow outside the receiver aperture (Tellez, 2003). Some metallic volumetric absorber modules were tested in the Solair Receiver, in the Central Receiver System (CRS) facility at the PSA, Spain. Small solar receivers or prototypes (about 200 kWth ) were tested at the top of the CRS tower. The facility receives the radiation from a
primary field with 91 heliostats of 39.3 m2 of reflecting area and a secondary field in the north with 20 larger heliostats. The nominal average reflectivity of the field is 92%.

The CRS facility has a CCD camera system for measuring the concentrated solar radiation flux. In this system, a Lambertian target is installed over a moving bar. It passes in front of the receiver aperture, parallel to its surface, intercepting the solar irradiation when a flux measurement is required. Two kinds of lambertian targets were used during the tests: a plate of 0.65 m x 0.26 m, area enough to cover the central cups and give the image of the region of interest in only one recorded image, and a Lambertian strip (long and narrow). The image of the concentrated solar focus is given by the superposition of several recorded images. The image of the radiation diffused by the target is obtained and the gray-level value of each pixel is converted to a radiation value by means of the calibration (Ballestrin, 2002). Then, the total power can be integrated and the calculation of the rest of the magnitudes of interest, such as peak irradiance or distribution, is possible.

A calorimetric test-bed for the evaluation of small atmospheric-pressure volumetric receivers is installed at the top of the 43 m metallic tower. The test-bed consists of different sections to conduct the hot air from the absorber to the blower through the water cooled heat exchanger. Every section has the necessary instrumentation to measure the temperature, pressure and flow rate. The test-bed was also implemented with an air — recirculation circuit, to recover the exhaust warm air and to improve the total efficiency.

At the front of the test-bed, the Solair 200 kWth receiver is currently installed (Tellez, 2003). This is a prototype, designed to fit the dimensions of the CRS installation, which is about

0. 9 m aperture diameter. This receiver is based on the same concept that the High Temperature Receivers (HitRec I & II) (Hoffschmidt,1999; Tellez, 2002): a modular receiver of absorber cups held in the rear part by the structure of a double membrane, which serves to re-circulate the air. The Solair receiver improves the material construction of the HitRec, and changes the shape of the cups from hexagons to squares with sides of 125 mm length. The 36 absorber modules are relatively free to move or thermally expand, thanks to a gap between adjacent modules. The gaps are also used to inject the returned air over the absorber entrance at the same time that it cools the stainless steel construction. In addition, they make easier the replacement of the modules.

Tested absorbers:

The Solair receiver concept is not directly linked to one specific absorber material. Any material (ceramic or metallic) can be placed into a cup of the correct dimensions and installed in the receiver. Thus, the collaboration with the German manufacturer of metal catalyst carriers Emitec enabled to test several metallic monoliths of corrugated foil, which were kindly provided by Emitec. The number of modules tested for this work amounts to 31: five square cups and twenty-six cylindrical absorbers (18 cylinders with 8 cm diameter and 8 units with a diameter of 11 cm).

The modules had different lengths, ranging from 20 mm to 58 mm in the case of the cylindrical absorber modules. Moreover several cell densities were tested, varying between 400 and 800 cells (or channels) per square inch (cpsi). All the tests for the different absorber modules were carried out during two test campaigns at the PSA: the first in July 2003 and the second in the winter of 2003-04. The first test campaign was aimed to measure the air outlet temperature distribution. Thus, the metallic square cups of 125×125 mm2 were instrumented with 7 thermocouples type K, which were symmetrically placed at three different radial distances in a cross section located at 2 cm from the absorber exit.

The set of 8-cm diameter cylindrical absorber modules were adapted to the ceramic square cups with an alumina plate adaptor. Due to the smaller area of the absorbers, only 5 sensors were placed at the absorber exit. These thermocouples were located also symetrically at two different radial distances.

In the second campaign, besides the thermocouples placed at the absorber exit, some additional thermocouples were inserted inside the absorber. The positions for the thermocouples inside the absorber were planned regarding the results of the model of Hoffschmidt (1996). These results state that the channels in the absorber centre transfer the heat radially to the edge of the absorber module, resulting in different axial temperature distributions for a channel in the absorber axis or near the absorber edge. In order to measure these axial temperature differences, in the 11-cm diameter cylindrical modules, 4 thermocouples were inserted through central channels at different downstream distances. Other 3 thermocouples were put near the edge.

The last two square metallic absorber cups tested were instrumented during manufacturing (see fig.1) with 5 inner thermocouples about the absorber centre at 1, 3, 10, 30 and 50 mm from the entrance, respectively (see thermocouples 1-5 at fig.1). Other 3 thermocouples were installed at a distance of 30 mm from the central axis, at 50 mm, 30 mm and 10 mm from the entrance (see thermocouples 6-8). Behind the absorber outlet section, other 7 themocouples were installed: one at the centre, two at 30 mm from the centre at both sides and four at 60 mm in the diagonal lines. One of these cups had a cell density of 600 cpsi and the other 500 cpsi. The absorber length inside the metallic square cup was 7 cm.

Planning of the tests:

The experimental goal was to perform a parametric study of the absorber temperature response with the variation of the most relevant system inputs: the flux of the concentrated radiation onto the absorbers and the cooling air mass flow rate. The usual test procedure was to begin the test with a certain number of heliostats and the maximum mass flow rate (blower power set to 100%). After approximately one hour, when the whole receiver had heated up and achieved a steady temperature, the incoming radiation was measured for those conditions. Then, the mass flow rate was reduced by decreasing the blower power by 5% or 10%. About twenty minutes later, when the cups outlet air temperatures had
achieved a new steady value (provided that no clouds disturbed the system), a new flux measurement was taken. Furthermore, the mass flow rate was lowered again, repeating the process for several values of the blower power. When the weather conditions were good and no clouds obliged to wait until having stable direct radiation and high temperatures again, five or six different values of the air flow rate were used (reaching the 70% or 75% of the blower power). If there was still some time before ending the working day, the mass flow rate was increased up to a value already tested before, in order to check repeatability. Some other specific tests were also made, like changing the air return rate or decreasing radiation and mass flow rate at the same time.

The data acquisition system (DAS) of the calorimetric test-bed with the Solair receiver has 138 signal inputs. Due to large amount of data that produces one single record, these data were recorded every five seconds. The acquisition system was implemented with the sensors installed inside the cups with metallic absorbers, but these signals were recorded every second, in order to appreciate any unstable behaviour, even with a very small frequency. Another temperature was recorded in the implemented DAS: the ambient temperature just under the receiver, given by a thermocouple out of the focus of the incoming radiation. It is usually assumed in the literature that the air temperature at the absorber entrance coincides with the ambient temperature, disregarding the possible heating up of the air around the receiver due to radiation losses.

The liquid desiccant system has been fully operational since April 2003, and its operation was monitored throughout the summer of 2003 (27 May through November 6).Numerous experimental runs were conducted prior to this period, to test various components and instruments and to make final adjustments in the control system. As it turned out, weather during the spring of 2003 in Haifa was relatively dry, and the desiccant system could not perform in a meaningful way until the end of May.

The complete set of data collected during the monitoring period contains 30955 lines, each representing a record of the system’s performance at a particular time. Normally, data was taken at one-minute intervals. Each record contains 28 instrument readings of temperatures, humidities, flowrates, pressures, etc. at various parts of the system. A computerized data acquisition system was used, showing the location and type of the various sensors and measuring points in the system. A log book of events was maintained during the monitoring period and all irregular events were listed. These include electric power interruptions, failed equipment etc.

For the purpose of the following discussion, three typical days were selected from the large volume of data — one for July, one for August and one for September. These are days during which the system operated rather smoothly, representing what may be expected ultimately after all the various operational and control problems have been fixed. Characteristic measured data representing the total monitoring period is as follows:

Heating water temperature Heating water flow rate Cooling water temperature Cooling water flow rate Air flow rate through desorber Air flow rate through absorber

Maximum solution concentration Minimum DPT reached

Air flow rate through desorber/absorber fans [m3/sec] Average air density through absorber/desorber Desorber fan power @ 32 Hz (40Hz)

Absorber fan power @ 20 Hz (40 Hz)

Air/Air Heat exchanger power

Solar collectors circulation pump power

Solution pumps power @ absorber/desorber sides

Hot water pump power

Cooling tower fan/pump power

Total parasitic power

Figure 2 describes the variation of three air humidities as functions of time for the same selected typical day (21 August 2003): outside air (Win), desorber outlet air (Woutl) and absorber outlet air (Wout2). Note that the absorber outlet air humidity is also that of the supply air to the conditioned space. As evident, the outside air humidity has remained approximately constant during the whole day, at about 16 g/kg, with a slight increase toward the evening. The absorber outlet humidity was equal to that of the outside air when the liquid desiccant system was put in operation at about 10:00, and was reduced to about 8 g/kg within 20 minutes. The machine was able to keep this humidity steady throughout the day. The desorber outlet air humidity, which is the control parameter, shows considerable variations. The control system shut the desorber off about 5 minutes after the start of operation, when the temperature of the hot water did not reach the set minimum limit; it turned the desorber back on and then turned it off when its outlet air humidity went below 30 g/kg; this is an indication that the solution in the desorber becomes too concentrated in LiCl, which may lead to crystallization. The same sequence continues several times during the day. The on-off cycling of the desorber makes it possible to maintain the supply air humidity at the desired and steady values.

In view of the ambiguity often encountered in the literature regarding the role of parasitic losses in the performance of desiccant systems (and other heat pumps and HVAC systems) we have introduced four types of coefficient of performance:

1) COP1 is a strictly thermal COP, not including parasitic losses.

2) COP2 includes in the denominator the sum of (solar) heat input and parasitic losses.

3) COP3 is the same as COP2, but the parasitic losses in the denominator are converted to their equivalent heat value, assuming power plant electricity generation at 40% efficiency.

4) COP4 is the ratio of latent heat removed from the process air to the electric power equivalent of the input energy, consisting of (solar) heat and parasitic.

Figure 3 describes the four COP’s as functions of time, for the data of 21 August 03, discussed previously. Clearly, COP4 shows the highest values and COP3 the lowest, based on the above definitions. Except for local fluctuations, all COP’s seem relatively

steady at periods of operation. The on-off operation of the desorber is clearly reflected in the chart.

A summary of the calculated data representing the total monitoring period is as follows: Average heat balance error (desorber side) 10%

Average heat balance error (absorber side) 20%

(where the Temperature uncertainty of the PT100 sensors is 0.2oC and the Relative Humidity uncertainty is 2% )

Heat transfer effectiveness:

Desorber solution/water heat exchanger Absorber solution/water heat exchanger Solution/solution heat exchanger Desorber side air/air heat exchanger Mass transfer coefficients:

0.3 kg/sec 0.9 kg/sec

81%

CONCLUSION

The objective of this project has been to construct a solar-driven liquid desiccant system for cooling, dehumidification and air conditioning — to test the concept, identify problems, carry out preliminary design optimization and measure performance. The prototype system, built at the Technion campus in Haifa, is designed to air-condition a group of offices on the top floor of the Energy Engineering Center. The design process involved initial measurements to determine unknown parameters, along with extensive performance simulations. With many unknown factors, the initial design of the system underwent several changes during the development period. The characteristic performance of individual components, analyzed theoretically in the simulation, was studied experimentally. Measurements have provided much-needed realistic data about heat and mass transfer coefficients. Important information was obtained about practical design aspects of the key components — dehumidifier and regenerator — as well as quantitative
data about their performance. The final prototype, including controls, has been fully operational since April 2003. The system functioned well, with 12 kW dehumidification capacity, and its performance was monitored throughout the summer of 2003 — from the end of May till the beginning of November. The data analysis indicates a thermal COP of about 0.8, with parasitic losses on the order of 10%.

The COP calculations performed on the monitoring data have yielded satisfactory results, particularly with regard to the thermal COP. By more elaborate design in the future, it is anticipated that parasitic losses could be minimized and better overall COP’s could be achieved.

The thermodynamic cycle of an absorption chiller is characterised by the working fluids used and the temperatures of generation (desorption), condensation, evaporation and ab­sorption. These temperatures are determined by the external temperatures of the heating, the cooling and the cold water and by the design of the heat exchangers (desorber, con­denser, evaporator, absorber). In high capacity absorption chillers flooded tube bundle heat exchangers are used as generator. This causes a relatively high difference between the temperature of the external heating water and the solution temperature reached inside the desorber. The limit for the solution temperature is the heating water outlet temperature. For the low capacity absorption chiller that was developed a special heat exchanger de­sign is used for the main components (desorber, condenser, evaporator, absorber). They consist of spiral copper tubes over which the solution (generator, absorber) or water (evaporator) is dripped. In the generator there is a cross current flow of the solution trick­ling down the tubes and the heating water inside the tubes. The limit for the solution tem­perature is the hot water inlet temperature.

To reach a high efficiency a plate heat exchanger is used as solution heat exchanger.

The design parameters of the small capacity absorption chiller are:

— Condensation temperature of 40 °C with a cooling water inlet into the chiller of 32 °C and an outlet temperature of 38 °C. The cooling water flows in series first through the ab­sorber and then through the condenser. The absorption temperature is 36 °C.

— Evaporation temperature of 13 °C with a cold water inlet temperature of 21 °C and a cold water outlet temperature of 15 °C.

—

Temperature of the weak solution after generation of 77 °C with a hot water temperature of 90 °C/80 °C inlet/outlet.

Figure 2 shows the configuration of the absorption chiller and the location of the state points of the cycle. Table 1 contains the results of the calculation of the absorption chiller cycle.

At the Kimbell Art Museum direct sunlight penetrates only at the west-facing entry lobby, which has a large span of clear glass. To protect the entrance of direct sunlight on this fagade, Kahn extended one module of the vaulted roof and used an array of trees to filter sunlight. Few sculptures are displayed in the lobby, where much of the social and gathering activities take place. In the rest of the museum, sunlight penetration occurs only in small areas next to the narrow openings at the end of the vaulted ceiling, mainly in areas toward the south and west of the building. These discreet sun patches move slowly throughout the day without ever reaching the display areas. The movement of these bright surfaces gives to visitors and museum workers a subtle indication of the sun movement, highly appreciated by them. Figure 7 shows a view of the south wall of the central gallery, where soft reflected light bounces off the metal reflective panels onto the vaulted ceiling. Illuminance levels on the display areas throughout the museum do not exceed 50 lux.

The Amon Carter Museum

Direct sunlight was observed in three different galleries of the Amon Carter Museum: the East-facing gallery (entry lobby), the South-facing (2nd floor), and North-gallery adjacent to atrium (2ndfloor),