Великий Новгород  новости
Февраль 16th, 2019
M. Hiller 1, S. Holst 1, T. Welfonder 1, A. Weber 2, M. Koschenz 2 TRANSSOLAR Energietechnik GmbH
1 CuriestraKe 2, D — 70563 Stuttgart, tel.: +49 711 / 67976 — 0, fax: +49 711 / 6797611 welfonder@transsolar. com, http://www. transsolar. com
2 EMPA, Abteilung Energiesysteme/Haustechnik, CH8600 Dubendorf
In the planning process and evaluation of innovative energy concepts simulation of buildings and systems gets more and more important. With the internationally well known software program TRNSYS [1] those simulations can be accomplished with a very high complexity.
The paper describes the Program TRNFLOW [2] which is a new addon for TRNSYS for the simulation of natural ventilation. Also the main new features of the TRNSYS Version 16 released in Mai 2004 will be described.
Coupled Airflow Simulation — Current Situation
In order to achieve sustainable buildings new energy systems have been generated using natural effects to renew the air and lead away the heat. Examples are passive night cooling, double facades, solar chimneys, inside courtyards and so on. In these systems the mutual impacts of thermal and air flow behavior are very distinctive. Thus for numerical building simulation programs an integral approach is inevitable.
Already in 1993 in the frame of the IEA project Annex 23 the EMPA has developed a coupling of the multizone air flow model COMIS [3] with the thermal building and system simulation program TRNSYS and this was presented at the TRNSYS Userday 1994. The selfcontained program COMIS was modified to TRNSYS Type 157 which can be linked to the thermal building model Type 56 within the TRNSYSDeck via in — and outputs. The input information of the air flow model are read in by Type 157 from the standard COMIS Input File (CIF). The TRNSYS solver iterates the results of the two models until they match. Meanwhile the coupling has been successfully applied by several projects and simulation tasks. However thereby it was pointed out that the coupling is not very userfriendly and also requires a laborious handling. As the mutual classification of the data is made by hand the inputs tend to be incorrect. This errorproneness is also increased by the redundancies of the two input files. Furthermore it was proven that the TRNSYS solver is not always the perfect solution for such a system. The solver possibly has to be supported by additional convergence promoting Types what makes the handling again more difficult. Yet the need of an integral approach concerning thermal building dynamics and natural air exchange was clearly necessary. Therefor with TRNFLOW an improved version including a deepened integration of the two models (thermal and airflow) has been developed.
Multizone Air Flow Model in TRNFLOW and COMIS
Multizone air flow models idealize the building as a network of nodes and airflow links. A node represents a room volume which a set of state variables can be assigned to. Cracks, window joints and openings, shafts as well as ventilation components like inlets and outlets, ducts and fans represent the links. Boundary conditions and thereby also input factors are: State variables of the air in the zones Local wind pressures
The pressure pZ is a free parameter in the node which is evaluated according to the continuity equation (mass flow balance in the node = 0 ). This results in nZ equations where nZ represents the number of zones.
The relation between mass flow rate m and pressure difference dp and the zone pressure pZ is not linear. Therefor an iterative process is used to solve the system of equations. The mass flow rates per link and all dependent factors such as air exchange rates, air age etc. are calculated of the resulting zone pressures pZi. The calculation is static without an explicit consideration of the timestep. In principle calculating a condition based on a new time is independent of the previous timestep.
The total hemispherical reflectivity R(A) and transmission T(A) for singlelayer on glass samples show a quasizero absorption of the films, confirming the previous results from laser reflectometry and spectroscopic ellipsometry methods. The theoretical curves of Figs. 2 and 3 are calculated using the experimental optical constants determined by spectroscopic ellipsometry on singlelayer samples. In our model, we suppose homogenous layers and sharp interfaces.
Figure 2 represents the total hemispherical reflectivity of Ti02/Si02 multilayers formed by three and five alternating layers of Ti02/Si02. The thicknesses of the layers are indicated in table 2. We observe a reflectivity peak in the visible range. The peak position determines the color of the multilayer film. The dotted lines indicate the theoretical reflectivity. We observe a good agreement between the experimental and the theoretical values. The reflectivity peak position, its maximum value and its FWHM depend on the layer thicknesses and on the number of layers. In general, at one wavelength the reflectivity peak maximum increases with increasing layer number.
Figure 3 shows the total hemispherical reflectivity of Al203/Si02 multilayers formed by an increasing number of alternating layers having the same thicknesses. The dotted lines indicate the theoretical reflectivity increasing with the layer number. This evolution shows the same tendency as previously presented results obtained by using the simplified model of constant refractive indexes [17].
The thicknesses of the individual layers of Fig. 3 are indicated in table 3. The peak position is relatively constant and its maximum value increases by increasing the number of alternating layers. The disagreement between the experimental and calculated values for
the ninelayers samples can be explained by the long deposition time and an eventual change of the deposition conditions.
Wavelength [nm]
Within the framework of the IEA Task 27 "Performance of Solar Facades" we have tested a series of different combinations of solar protection and glazings. These systems include internal, integrated and external shading devices. The characteristic data are given in the following section.
External blind systems
• three devices with identical complex lamellae
• white, white perforated and brown lamellae (see Figure 1) external blinds
• 90 mm width, a distance of the lamellae of 80 mm
• combination with a lowe glazing (pos. 2, 16mm Argon), g=48%, U=1.3 W/(m2K)
Internal systems
• white lamella type blinds using slats 25mm white, distance 22mm
• textile roller blind silver (outside) and white (inside)
• combination with a lowe glazing (pos. 2, 16mm Argon), g=48%, U=1.3 W/(m2K)
• combination with a lowe glazing (pos. 2, 16mm Argon), g=35%, U=1.1 W/(m2K)
Integrated systems
• white lamella type blinds using slats 15mm white, distance 13mm
• textile roller blind grey (both sides)
• integrated in lowe glazing (pos. 2, 27mm air), g=47%, U=1.5 W/(m2K)
• integrated in lowe glazing (pos. 2, 27mm air), g=32%, U=1.4 W/(m2K)
The heat transfer governing equations for steady state laminar natural convection in cavities are the mass, momentum and energy conservation equations in x, y and z axis for an incompressible fluid [Versteeg, 1995]. These equations can be expressed in conservative form:
Conservation of mass:
where T0 is the reference temperature, and is calculated by the mean temperature distribution of the exterior glass surface and its result is averaged with the temperature of the isothermal wall, so:
Hx, Hy y Hz are the lengths of the edge surfaces of the cubic cavity, Hgx is the thickness of the glass and Tci it is the temperature of the wall 2.
The boundary conditions for the momentum equation are:
u(0,y, z)= v(0,y, z)= w(0,y, z)= 0 u(Hx, y,z)= v(Hx, y,z)= w(Hx, y,z)= 0
u(x,0,z)= v(x,0,z)= w(x,0,z)= 0 (6)
u(x, Hy, z)= v(x, Hy, z)= w(x, Hy, z)= 0 u(x, y,0)= v(x, y,0)= w(x, y,0)= 0
u(x, y,Hz)= v(x, y,Hz)= w(x, y,Hz)= 0
The boundary conditions for the energy equation are:
Wall 1
— k“ (X’0’Z ) = ЧГ3 (X’0’ Z ) (7)
dy
Wall 2
T(0,y, z)= Tci (8)
Wall 3
— ka T(x, Hy, Z ) = ЧГ3 ( Hy, z)
dy
Wall 4
d — dT
— ka (Ях ’У’z)=~kg (Hx, y,z)h qr4(Hx, y, z)h Sa, f
Wall 5
— ka (x’ y ’ Hz ) = qr5 (x y ’ Hz )
dz
Wall 6
дт
— k a —(x’ y,0) = qr6(x ’ y,0)
dz
where qr1(x,0,z), qr2(x, y,0), qr3(x, Hy, z), qr4(Hx, y,z), qr5(x, y,Hz) and qr6(x, y,0) are the energy flux from the radiative exchange between the wall surfaces, Saf is the absorbed energy by the solar control coating and Tg(x, y,z) is the glass temperature for Hx<x< Hx+Hx2, where
d—
Hx is the thickness glass. The temperature gradients (Hx, y,z) in the glazing were
dx
evaluated by using the heat conduction model.
Andreas Gombert, Benedikt Blasi, Wolfgang HoRfeld, Volker Kubler, Michael Niggemann, Peter Nitz, Gunther Walze, FraunhoferInstitut fur Solare Energiesysteme ISE, Heiden — hofstr. 2, 79110 Freiburg, Germany
Jorg Mick, Institut fur Mikrosystemtechnik, AlbertLudwigsUniversitat, GeorgesKohler — Allee, 79110 Freiburg, Germany
An overview of known methods to modify the optical properties of solar energy materials by using microstructured surfaces is given. Applications for micro structures in solar energy components are wavelengthselective absorbers, heat mirrors, light traps for PV cells, wavelengthselective concentrators for solar radiation, daylighting components, antireflective zeroorder gratings, and radiation emitters with selective optical properties based on grating resonances. This paper addresses the design and the whole experimental process chain from the microstructure origination on large areas to the replication. The need for costeffective production technologies and durable structured materials is emphasised.
Introduction
A wide variety of solar energy systems from PV cells to buildings exists. All the very different systems have in common that they demand sophisticated optical solutions for an efficient transport, collection and conversion of the solar radiation. Modifying the optical properties of surfaces or planar devices by coatings and microstructures is often used in order to optimise the radiation power management of solar energy systems. Publications in which diffractive structures were proposed to finetune the optical properties date back to the 1970’s [1 — 4].
In fact, advances in diffractive optics like the first approaches to solve the problem of dou — blyperiodic gratings were driven by scientists having solar applications in mind, e. g. Mc Phedran and Maystre [2]. Since the fundamental work of the mentioned authors, diffractive structures in solar energy systems were published for a variety of components in solar energy systems, e. g. [5 — 15]. In Table 1, the quoted publications are classified according to the components for which the diffractive structures were proposed. Additionally, the required optical properties and the proposed structures are listed.
The advantage of considering periodic microstructures is the possibility to model their optical properties rigorously, i. e. by solving the Maxwell equations [16]. Thus, the optical properties of periodically microstructured surfaces can be simulated with rather high accuracy. This is not the case for aperiodic structures.
The challenge of using microstructures is the requirement on very precise manufacturing technologies. Such technologies exist in the field of microelectronics or microsystems technology but are in general not suitable to structure large areas homogeneously. Thus, many of the published approaches are difficult to realise especially due to the mismatch of dimensions between the microstructures and the areas which have to be microstructured in solar applications.
From the technical point of view, homogeneous origination of precise microstructures on large areas is the most difficult step in the process chain. Because of its ability to share the high origination cost with a large number of products, microreplication is very promising from the commercial point of view. Microreplication is a suitable process for some of the applications listed in Table 1 but not for all of them.
At Fraunhofer ISE, we have picked up the idea of using periodic microstructures in solar energy applications at the beginning of the 1990s. Since then we have been working on the design and the manufacturing techniques for the following optical components: antireflective surfaces, light traps for PV cells, sun protection systems, and wavelength — selective radiation emitters. By using interference lithography we were able to originate microstructures on areas of up to 4800 cm2.
In Fig. 3 the trend of S varying with d is shown, for the two cases of copper wall (CW) and brick wall (BW). For d<0.045 m the flow is laminar with Reynolds numbers inferior to 2500; for d>0.060 m the flow is turbulent with Reynolds numbers superior to 3500: the laminar — turbulent transition zone is pointed out in dashed outline. In the case of turbulent flow the roughness value of the ventilation duct has been assumed to be equal to 0.005 m.
The brick outer slab turns out to be more convenient, from an energy point of view, than the copper one.
The following figures all refer to a ventilation duct being 0.04 m in thickness.
In Fig. 4 the trend of S varying with G for the two examined walls, CW and BW, is reported. Two values have been considered for the indoor air temperature: T=24°C and T=26°C. The percentage energy saving S distinctly increases as G and the indoor air temperature Ti rise. In Fig. 5 the trend of S varying with G is shown for the wall BW. The following values have been considered for the friction factors on the inlet and outlet sections: Xin=0.5 and Xou=1; Xin=1 and Xou=2; Xin=2 and Xou=4; Xin=4 and Xou=8. The graphs clearly show the convenience to reduce, as much as possible, the head losses occurring on the inlet and outlet sections.
Figure 3 — Variation of S with d (m) for the two walls CW (solid line) and BW (dashed line). The laminarturbulent transition zone is pointed out in dashed outline.
Figure 4 — Variation of S with G for two values of the indoor air temperature: Tj=24°C (solid line) and Ti=26°C (dashed line).
The trend of the mean heat flux Q coming into the room through the ASW varying with G is reported in Fig. 6, for the two examined walls, CW and BW. The two cases of T=24 and T=26 °C have been considered. The trend of the mean heat flux Q0 (obviously the same for the two walls) coming into the room when the ventilated duct is closed (dotted line) is also reported for comparison. The difference (Q0Q) and, therefore, the reduction in summer thermal loads, achievable by using a ventilated wall, increases as G and Ti rise.
In Fig. 7 the trend of Q varying with the solair temperature Te, for the wall BW, is reported for three values of the air temperature in the shade: T0=24°C, 26°C and 28°C. The trend of Q0 (dotted line) is reported for comparison. Obviously, it results that Q=Q0 for Te=T0 (without solar radiation). The ventilation convenience increases as Te rises as well as it increases, for a given value of Te, as T0 decreases.
The Fig. 8 refers to winter and the two examined walls. In this figure the trend of Q varying with G for two outdoor air temperatures in the shade is shown: T0=0°C and 7°C. The trend of Q0 (dotted line) is reported for comparison. Notice that, in this case, the wall showing less heat losses is the copper one; it happens as a consequence of the fact that the thermal resistance RB of the wall CW, with copper outer slab, is higher than the resistance RB of the wall BW, with brick outer slab (see Tab. 1). The graphs clearly show that, in winter, the ventilation always determines a rise in heat losses.
Conclusions
The ASW can meet, if well designed, the aesthetic and formal requirements of contemporary architecture, and also contribute to reduce energy consumption in buildings. The examined graphs clearly show that the use of ASW can determine a remarkable reduction in summer thermal loads; the duct is, obviously, required to be, as much as possible, free from any obstacle and the head losses to be reduced on the inlet and outlet sections for the abovestated reduction in summer thermal loads to occur. Hence the necessity of an accurate design of the inlet and outlet openings.
The energy saving achievable using the ASW distinctly increases as insulation increases; for a given value of the insulation and of the outdoor air temperature in the shade, the
reduction in the summer thermal load increases sensibly as the temperature provided for the indoor environment increases. In the examined situations the brickfaced wall (BW) has turned out to be more convenient than the copperfaced one (CW), from an energy point of view. In any case, it seems to be not convenient to consider air duct thicknesses inferior to 45 cm.
In winter, remarkable rises in heat losses can occur, leaving the duct open, especially connected with remarkable values of G. This leads to advise closing the duct in winter. But, considering that in winter the values of G are usually moderate, it would be advisable reducing the ventilation, e. g. with selfregulating dampers at the duct inlet and outlet sections, in order to drain the humidity due to possible infiltrations or condensation phenomena.
Acknoledgements
This research was supported by Italian Ministry of Education, University and Scientific Research (MIUR) and by University of Pisa within the National Relevant Interest Project (PRIN 20032005): "Energy and environmental diagnosis on existing buildings: research methodologies, determination of qualification parameters and technicoeconomic assessments”.
1.
The solar radiation daily variation corresponding to the typical clear days characterized by a sunshine fraction a > 0,9 and a nebulosity index Kd < 0,2 [18], The variation of temperature, pressure and the total are obtained by establishing a mass and a thermal balance of the volume elements of the porous medium discretised on equal thickness and to evaluate the equations of heat and mass transfer in each slices separately.
In each slice, the transfer of heat is obtained by applying the first principle of thermodynamic for an open system by taking into account the fractions corresponding to the adsorbed fluid, the gas and the solid:
d(^U) + ^ qs hs — ^ qe he = Ф + E (2)
Su = Vc [(1 — s)psUs + (є — a)pgUg + a paUa] (3)
The combined of equations (3) and (4), enable us to obtain the general equation of heat and mass transfer in a layer, equation (5), these equations are written in the case of :
— cylindrical elements:
rn 4 4 , idT P d[(sa)pg]
TOC o "15" h z [(1 S)pscs + (є — a)pgcg +apaca] —
dt pg St
d(apa) P d(apa) d2T 1 d)T
AHads (T, P) =2e ( +
dt pa dt dr2 r dr
These equations in the porous medium are completed by the initial and boundary conditions:
— Initial condition:
— T(r,0) = Ta ( r = 0,…, R ) (7)
— T(i, j,0) = Ta ( i = 1,., n ) and ( j = 1,., m ) (8)
Ta is the ambient temperature before the sunrise All the reactor is a constant temperature.
— Boundary conditions:
The boundary conditions to the center of the porous medium is a condition of symmetry;
(ffr) _ 0 , (ffr) _ (ffT) _ 0 (9)
(~&r ) r=°_0 (Ж) x=i, у — ~оу) x, y=i — 0 w
The thermal balance of the metallic wall is given by the following equations;
— Cylindrical tube without fins
pac VacCac — TvUacPsDe —UlDe(Tac—Ta) — hinDi(Tac — T) (10)
— Cylindrical tube with fins
Caofitc Vc— = Tv Oca Ps (De + 2 ШУ — U (Dev + 2 Qi)(TacTa ) — k 7t D, (Tac~T)
TOC o "15" h z dt 2
In this equation we take account the efficiency of the fins into consideration [19]
Q =tanh (m £) (19) and m = VUl/2acs (11)
m t
I is the wide of the fin
— Rectangular tube
Qj m
pac Vac Cac— = Tvaac Psu — Ul Sr (Tac ~ Ta) ~ 4 ^ hi AY (Tac ~ Tnj) —
» ‘ (12)
2 hi AY (Tac — Tnl) — 4 £ hi AX (Tac — Tim) — 2 hi AX (Tac — T 1m)
i
The obtained equations from a system of non linear differential equations that are solved by the implicit finite difference method [20].
The efficiency of the machine is characterized by the thermal coefficient of performance; COP and a solar performance coefficient COPs, deduced from the characteristic points of the obtained cycle using the following relations;
Tmax
Qc ^ mi Cpi dT + Qdes
Index i relates to ammonia, the activated carbon and the metal tube.
Qdes is the quantity of energy necessary to the desorption of the quantity Am [20];
5. Results:
The numerical simulation of the modelled reactor, under ambient temperature and solar radiation recorded in Tetouan, enable to describe aspects of heat and mass transfer inside the porous medium. The results gives the characteristic parameters of the functionning machine.
The numerical results obtained under real conditions of ambient temperature and solar irradiation relative to typical clear days of each season, allow the evaluation of the considered reactors performances from the cycle characterising points. The adsorption temperature is equal to the ambient one, the evaporation temperature is zero and the condensation temperature corresponds to the ambient temperature related with the beginning of ammonia desorption inside the condenser.
Figure 3 shows the variation of the thermal performance coefficient COP versus the normal and finned tubes diameter for the studied typical days. We observe that for each case there is a maximum value corresponding to a given tube diameter representing the optimum values.
Hence, under the applied functioning conditions the optimum COP value (diameter) are variable and depends strongly on solar radiation and on the ambient temperature. The same remarks are observed for the variation of the daily cycled mass versus the diameter figure 4a, considering a collector of a 1 m2 of surface composed a number of equal tubes. The total cycled mass corresponds to the sum of the desorbed quantities by each cylindrical tube. We note that the optimum values are higher for the rectangular reactors compared to the cylindrical ones figure 4b corresponding to the amount of the activated carbon used and thus to the offered volume to the reacting mixture.
The high values of the COP in April and October can be explained considering the fact that ammonia adsorption takes place before the sun rise in a uniform temperature porous medium, equal to the ambient temperature but less than that in July. So, the adsorption is very important, the choice of typical clear days characterised with high solar radiation allows to heat to the maximum values the reactors and thus the COP is a function of the considering temperature and that the maximum heating of the absorbent permits an important heat adsorption.
The variation of the maximum temperature at the center of the porous medium is a decreasing function of its width, it is had has the thermal conductivity of the porous medium and the thermal capacity of the whole of the reactor figure 5. We notice that the finned tubes improves the thermal exchanges between the metallic walls and the porous medium, consequently the maximum temperature attained is greater allowing an important desorption for the finned reactor with regard the same diameter normal tube, figure 5a. Figure 6 shows the evolution of temperature at the center of the porous medium versus the time for the three optimum widths reactors. For the cylindrical tube, we compare the temperature variation inside the tubes with a similarly diameter for the finned and normal reactor. The studied cycle begin the morning where all the reactor is at the ambient temperature and finish at midnight marking the start of a new cycle relatively to the temperature recorded at LT.
The rectangular reactor heating duration is higher, owing to the important volume of the fixed bed containing the mixture, than the cylindrical cases.
Figure 7 shows the pressure evolution inside the reactors versus time, causes by the temperature variation. The temperature elevation during the heating phase of the closed reactors causes an increases in the gas pressure until it becomes just larger than the condensation pressure which corresponds to the saturation pressure at the temperature condensation, then the desorption of ammonia into the condenser starts at a constant pressure and the heating of the fixed bed continues until the temperature reaching the maximum value. The reactors are closed and both temperature and pressure decrease.
At the pressure value of 4,2 bar the reactors are opened and the adsorption phenomena of ammonia vapour start with a cooling product quantity.
These evolution of temperature and pressure is represented in a Clapeyron diagrams, corresponding to the variation of Log P versus the temperature figure 8, and shows the daily thermodynamic cycle characterised by two isosters and two isobars representing four phases relatively to the heating or the cooling of the reactors.
In figure 9, we show the daily evolution of ammonia total mass, both adsorbed and gaseous, inside two cylindrical tubes having the same diameter in the two optimal cases. At the beginning the temperature is the same inside both of the tubes implying that their respective ammonia masses are also the same. During the heating of the closed reactor, condensation pressure inside the finned reactor is reached before the tube without fins, causing the opening of valve V1 and hence ammonia desorption. This desorption is important considering the temperature elevation and the values of 2,42 kg and 1,55 kg are collected for the unit area respectively for the normal and finned reactor.
Inside a 1m2 surface captor, 1 m long and 1 m wide, equals to the multiplication of the number of tubes by their external diameter. The total desorbed mass represents the sum of all the desorbed amounts in each tube. The non desorbed mass is the total fluid mass inside the reactor which the variation during a cycle is showed in figure 10.
We gives in table I the values of the computed amounts and those of the parameters under which the reactors functioning for the typical clear days of July, of which can be compared the three reactors. The obtained optimal geometry of each reactor presents an evaluation of the parameters that characterises the functioning conditions, the efficiency of the machine and the computed both provided and useful energy.
6. Conclusion
In this work, the aims is to present a model and an optimisation of solar adsorption cooling machine using ammonia / activated carbon couples, that allows a design according to the real functioning condition. The prediction of the performance of the solar refrigerator require the knowledge of various parameters, which characterise the daily thermodynamic cycle. The optimisation is based on heat and mass transfer in the porous medium consider the collected mass, the thermal and solar performance coefficient, allow to give an idea of the transitory evolution of temperature, pressure and ammonia concentration inside the reactors. The efficiency of each reactors are enhanced and the preferential adsorber depends on the desired role to generate (the useful cooling quantity).
A presentation of temperature and adsorption ammonia quantity inside the reactor that develop solar radiation is carried out in this paper. Thus, the simulation has been performed using some assumption will be applied to an experimental test.
Table II. Comparison of the operating parameters and results of each reactors

SHAPE * MERGEFORMAT
Figure 4. Total daily condensed mass versus the tube dimensionless — January April July 0 October
SHAPE * MERGEFORMAT
Absorptance changes were generally larger and occurred faster at lower pH values (Fig. 3). Changes in emittance were mainly the opposite, i. e. larger at lower pH values (Fig 4). The resulting PC values (Eq. 1.) were almost in all cases within the acceptable limit at pH
3.5, distributed both side of the limit at pH 4.5, and were generally above the acceptable limit at pH 5.5 (Fig. 5).
The majority of the samples exhibited neither specific temperaturedepending nor gasification type/rate — depending behaviour. In addition, there is no clear difference in degradation between the O2, N2 or nonaeration or the rate of aeration at any pH level. It seems that the pH level is the major determinant regarding to the degradation rate. Unfortunately, there was large deviation especially in the absorptance results at pH 5.5 exposure times between 0.5h and 4h.
In previous condensation tests for similar samples with deionized water according to draft proposal ISO/CD 12592,2 (Brunold et al., 2000) all the samples exhibited Arrheniustype temperature and timedependent degradation (Konttinen and Lund, 2003). Complexity of the simulated acid rain test method including multiple variables makes it difficult to determine the reasons for nonArrhenius type behaviour. The most likely reason is uncontrolled movement of the acid rain solution causing irregular chemical reactions. Futhermore, the primary assumption of the combined effect of gas feeding and natural convection being sufficient for moving the solution seems to be inadequate. The amount of reactants in the solution is quite small (Table 1). Therefore small variations in solution composition can have caused different results as well.
At one side the indoor temperatures are important boundary conditions for the multi zone air flow model and should therefore not be defined on the basis of a user’s guess. On the other side the indoor air temperatures calculated by the thermal model strongly depends on the exchange of air between the zones as well as the outside. To link the two models and mutually use the results is the obvious consequence. In TRNFLOW the multi zone air flow model of COMIS is completely integrated into the thermal model of Type 56. This means that the exchange of data between the thermal and the air flow model is made internally and no longer by inputs and outputs. The proper classification of air flows (infiltration, ventilation, couplings) and temperatures to the air flow node resp. the thermal zones and the appropriate other model is automatically carried out by the program.
The input files of both models are kept in the existing formats (BUI, CIF) but are created by only one user interface witch is a TRNFLOW Version of PREBID. Air flow model data depending on time, like wind velocity or window opening factors are defined as inputs or schedules. Outputs like air flows or zone pressures are declared as outputs by means of
new NTYPES and can be written into an output file using a printer type or processed otherwise. The standard COMIS Output File (COF) is optionally also available.
In order to take account of the solar spectrum, a multilayer sample is characterized by its solar transmission Tsol, its solar reflectivity Rsoi defined respectively by the following relations:
_ JT(A) Iso, (A) dA
J Iso. (A) dA
We note here Isol the intensity of the solar spectrum AM1.5. The integration range is given by the limits of the solar spectrum. The visible reflectance Rvis is determined from the photopic luminous efficiency function V(l), the standard illumination D6s(A) and the hemispherical reflectivity R(A):
R _ fR(A) • Р65(Л) • V(A)dA vis f 065(Л) • V(Л)dЛ
For the theoretical case of a deltadistributionshaped reflectivity, Schuler et al. [18] introduced a merit factor M defined as the ratio of the visible reflectance Rvis and the solar reflectivity Rsoi. M is then large for a high visible reflectance or low solar energy losses and consequently describes the energy efficiency of the visual perception.
Numerical simulations allow optimizing the reflectivity and transmission of the multilayer films as a function of the film thicknesses, the refractive indexes and the number of alternating layer. They show a correlation between the difference of the refractive index of the two materials. For example, a lower refractive index difference increases the optimal thicknesses of the individual layers and the layer number, but the solar transmission is high. The simulation optimization results based on the experimental optical constants of single layers will be published elsewhere [19].
Table 2 shows the solar transmission, the solar reflectivity, the relative visible reflectance and the merit factor M = Rvis/Rsoi in the case of the Ti02/Si02 multilayers. We indicated the experimental and calculated values. We see that for a given number of alternating layers, it is always possible to obtain either a high solar transmission or a high relative visible reflectance by adapting the thicknesses of both oxide materials. In order to obtain the best compromise between the energy losses by reflectivity and the visual effect, both parameters have to been optimized. Samples a and c show that the merit factor is not a sufficient indicator and one has to take into account the absolute Rsol. In fact, in these examples, the solar transmission is low and results in a uselessly high visible reflectance.
dTi02 [nm] 
dSi02 [nm] 
Tsol exp theo 
Rsol exp theo 
Rvis exp theo 
Rvis/Rsol exp theo 

2L 
27 
195 
88.1 
87.6 
12 
12.4 
19.8 
20.1 
1.65 
1.62 

3L 
a 
30 
122 
77.8 
77.2 
22.1 
22.8 
39.4 
39.1 
1.78 
1.71 
63 
73 
66 
67.3 
33.7 
32.7 
64.1 
58.7 
1.90 
1.80 

b 
18 
160 
85.8 
84.7 
14.3 
15.3 
24.2 
25.2 
1.70 
1.65 

5L 
c 
35 
148 
70.7 
69.9 
29.2 
30.1 
60.5 
61.1 
2.10 
2.00 
d 
14 
155 
85.5 
85.9 
14.2 
14.1 
23.3 
24.8 
1.83 
1.75 

e 
19 
130 
82.9 
82.9 
16.9 
17.1 
23.3 
22.2 
1.37 
1.30 
Table 2. Measured parameters (thicknesses, solar transmission and reflectivity, visible reflectance and merit factor) of TiO2/SiO2 multilayers combined with the same theoretical parameters 
Table 3 shows the solar transmission, the solar reflectivity, the relative visible reflectance and the merit factor M in the case of the Al203/Si02 multilayers. The solar transmission is slightly decreasing by increasing layer number, but stays at a high level superior to 89%, which is comparable to the solar transmission of uncoated glass (92 %). As mentioned above, this is due to the small refractive index difference between Si02 and Al203. The relative visible reflectance and hence the factor M increases.
The result shows that the prepared coatings can meet the requirements for obtaining different reflected colors. More efforts are needed to improve at the same time the solar transmission and the visible reflectance by considering other oxides and by optimizing the layer thicknesses.
dAl2O3 [nm] 
dSiO2 [nm] 
Tsol exp theo 
Rsol exp theo 
Rvis exp theo 
Rvis/Rsol exp theo 

3L 
a 
83 
95 
90.5 
90 
9.8 
10 
12.7 
13.5 
1.3 
1.34 
5L 
b 
83 
92 
89.9 
89.6 
10.2 
10.4 
15.2 
16.4 
1.5 
1.58 
7L 
c 
80 
91 
89.7 
89.1 
10.3 
10.9 
16.7 
20 
1.63 
1.84 
9L 
d 
80 
90 
89.4 
88.8 
10.7 
11.2 
18.7 
21.7 
1.74 
1.93 
Table 3. Measured parameters (thicknesses, solar transmission and reflectivity, visible reflectance and merit factor) of Al2O3/SiO2 multilayers combined with the same theoretical parameters 
4. CONCLUSION
In this work, colored glass to cover solar collectors has been obtained by alternative deposition of dielectric layers with high and low refractive indices. The stoichiometry was first checked by XPS. The deposition rate has been controlled by insitu laser reflectometry and confirmed by exsitu ellipsometry for complex systems with several layers. The optical properties of individual oxides of titanium, silicon and aluminium have been determined. A Cauchy dispersion model is adequate for extracting the refractive and extinction index in the case of reactive magnetron sputtering deposition.
The reflectivity and the solar transmission depend on the thicknesses and the number of the alternative dielectric layers. The fabricated multilayers fulfilled the fixed requirements: quasizero absorption, reflectivity peak in the visible, solar transmission above 85% up to 89% and an acceptable visible reflectance.
More effort will be directed to study the lifetime of the multilayer coatings by aging tests in orderto investigate theirapplicabilityfor architectural integration in buildings.
ACKNOWLEDGEMENTS
The authors wish to thank Dr. M. Ley for helpful discussions and R. Steiner for the technical support. This work is supported by the Swiss Federal Office of Energy and the Swiss National Science Foundation.