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 temperature-depending nor gasification type/rate — depending behaviour. In addition, there is no clear difference in degradation between the O2, N2 or non-aeration 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 de-ionized water according to draft proposal ISO/CD 12592,2 (Brunold et al., 2000) all the samples exhibited Arrhenius-type temperature- and time-dependent degradation (Konttinen and Lund, 2003). Complexity of the simulated acid rain test method including multiple variables makes it difficult to determine the reasons for non-Arrhenius 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 delta-distribution-shaped 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.

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.

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 in-situ laser reflectometry and confirmed by ex-situ 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: quasi-zero 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.

Within the European project ALTSET (Angular-dependent Light and Total Solar Energy Transmittance) the angular properties of so-called complex glazings (i. e. glazings with special optical properties) and shading devices have been investigated. The scope of the project was

the comparison between direct calorimetric measurements and modeling from layer properties

— the important experimental factors for good interlaboratory comparison of results

— error analysis and evaluation procedure

— conclusions for and development of a testing procedure.

The results showed that using different approaches for apparatus design one can reach uniform results for well-defined reference conditions and the project team has developed a test procedure for this purpose. Solar calorimetry could be shown to be an indispensible methodology for all complex glazings like diffusing glazings, transparent insulation,

shading lamellae and switchable glazings and can be used for validation of optical-thermal glazing models. But also for conventional glazings the accuracy is comparable with standard testing techniques. [ 4]

In a German national project REGES this approach has been extended for shading devices (internal and external devices), and recommendations for testing have been given. [ 6]

The principle and advantage of solar calorimetric testing is that glazing plus eventually shading devices, the complete fagade structure, can be tested in a "black box” approach, i. e. no additional information on the layer properties or the internal heat transport (ventilation) is needed. There is no modelling involved which is always a problem for more complex fagade constructions. Only a general parameter model is being used which is parameterized using the information from the experiment and parallel optical measurements.

The second important advantage is that angular dependence can be measured with these test procedures.

Figure 2: View of the solar calorimetric test Figure 3: Test frame for internal solar

cabin of Fraunhofer ISE, with an external shading devices shading device mounted in front of the calorimetric plate

The radiative exchange between the walls surfaces is calculated by using the net radiation method [Siegel, 1992], in which the net radiative heat transfer is:

qrk (x, y, z) = Jj (x, y, z)- qk (x, y, z) for k & j (12)

where Jj is the radiosity and qk is the irradiation on the wall surface given by:

qk(x, y,z)= J Jj(x, y,z)dFk-J k&j and k & 2

Aj

Jj (x, y, z) = £k&Tk {x y>z) + (l -£k ))k (x, y >z) for k & J’

k and j are the wall numbers, in which the heat flux was calculated. dFk-j is the differential view factor that indicates the fraction of energy that leaves from a wall k and strikes wall j. For walls 1, 3, 4, 5 y 6 the thermal emittance is ek&1 and for wall 2, ek=1.

The optical properties of the micro-structured surface and also the theoretical models to describe them depend very much on the relation between the period Л of the grating and the wavelength X of the incident radiation. Therefore, a classification of gratings defined by the period-to-wavelength relation is very helpful (Fig. 1).

effective medium theories rigorous diffraction theories rigorous diffraction theories photonic band structure calculation rigorous diffraction theories extended sea ar diffraction theories

geometrical optics (ray tracing)

If Л << X then the microstructured regions can by regarded as an effective media. They lead to a modification of reflection and transmission at the boundary air to material but not to a modification of the propagation direction of the radiation (Fig. 1). Only the zero-order diffracted wave is propagating and the wavelength dependence of the optical properties is small in this case. The effective refractive indices of this effective media depend on the refractive indices of the two media in the structured region and on the volume fractions of each of the media. Such subwavelength gratings can be used for antireflective surfaces or for polarisation sensitive devices.

If Л = X then resonance effects dominate and result in a strong wavelength dependence of the optical properties. It is possible to achieve high diffraction efficiencies in a specific dif­fraction order just due to the fact that only few diffracted waves propagate. The optical properties of such gratings have in general to be modelled by using rigorous diffraction theory, i. e. by solving Maxwell equations numerically [16]. Gratings in the resonance re­gion have mainly been used for spectral filtering but also for radiation deflection due to the high diffraction efficiencies which can be achieved.

If Л >> X then many diffracted orders propagate. The distribution of the diffraction efficien­cies depends very much on the structure profile. For very large ratios Л/X the optical properties of the surface-relief grating can derived by means of geometrical optics be­cause this is an approximation which holds for X ^ 0. The grating then represents an array of prisms, lenses, etc. which can be modelled to some extend by using ray tracing meth­ods. For ratios Л/X even as large as 100, the ray tracing method is not sufficiently accurate in many cases. Then, extended scalar approaches or rigorous diffraction theory has to be applied. Surface-relief gratings with a large ratio Л/X are particularly suited for light re­directing elements.

A. Wagner, Building Physics and Technical Building Services, University of Karlsruhe; S. Herkel, Fraunhofer ISE, Freiburg; G. Lohnert, sol id ar Planungswerk — statt, Berlin; K. Voss, Building Physics and Technical Building Services, University of Wuppertal.

Within the funding programme1 "Solar optimised building — SolarBau" projects of commercial buildings are subsidised, if the predicted primary energy de­mand for all technical building services does not exceed 100 kWh m-2a-1. Main objective of the demonstration buildings is the combination of high workspace quality with a low energy consumption. An accompanying research programme is evaluating the buildings with a two-year data acquisition campaign. The pa­per summarises some of the results and experiences.

Towards lean building concepts

In contrast to the established low energy and passive house standards in the sector of domestic dwellings there is only little consciousness of the energy consumption of commercial buildings — neither by their planners nor by owners and users. Numerous office buildings of the eighties and nineties show a very high energy consumption due to the fact that they have been designed without any respect to the interdepend­ence between outdoor and indoor climate. As a result the thermal and visual comfort in office rooms can only be guaranteed by extensive technical building services for heating, ventilation, air-conditioning and lighting (HVACL). High investment costs and a space demand of about 20 — 30 % of the building volume for HVACL equipment characterise a large amount of commercial buildings. The electricity consumption is dominated by HVACL facilities and not by office equipment.

Despite internal heat gains caused by the electric energy consumption, there is still a high demand of heating energy due to the high proportion of glazing and high air ex-

German Ministry for Economy and Labour / Deutsches Bundesministerium fur Wirtschaft und Arbeit

change rates. The left graph in figure 1 qualitatively shows the energy consumption of a normal office building as a function of the ambient temperature. The base load is caused by office equipment and the idling consumption of building services facilities. The waste heat associated with this base load affects the position of the balance temperature.

Figure 1: Qualitative end energy consumption of a conventional office building (left) and a lean office building (right). The dependence of the total consumption (HVACL and office equipment) on the ambient temperature is shown based on daily average values.

A higher common consciousness of resources, an increasing awareness of operation costs of buildings and the preference of users towards individual control of the indoor climate have led to a new trend in architecture: buildings with moderately glazed fa­cades, a high amount of daylight at the workspaces and the option of natural ventila­tion through windows that can be opened. However, a combination of integrated measures for "passive cooling" is a pre-requisite to ensure summer comfort without actively cooling or dehumidifying the inlet air.

Due to the reduced HVAC equipment these "lean" building concepts show a different performance (figure 1, right graph). Energy efficient office equipment, lower air change rates and a higher daylight autonomy reduce the base load and better insula­tion results in a lower balance temperature. Above this temperature the indoor condi­tions remain within the comfort range only by passive cooling measures. Although the indoor climate will vary more than in a completely air-conditioned building, this does not necessarily affect the perceived comfort negatively. Only extreme outdoor condi­tions may lead to short periods of discomfort.

T. Bostrom, Division of Solid State Physics, Department of Engineering Sciences,

Uppsala University, Sweden.

E. Wackelgard, Division of Solid State Physics, Department of Engineering Sciences,

Uppsala University, Sweden.

G. Westin, Division of Inorganic chemistry, Department of Materials Chemistry, Uppsala University, Uppsala, Sweden.

A promising novel solution-chemistry method to fabricate spectrally selective solar absorber coatings has been developed. The objective is to create highly efficient, flexible, inexpensive and durable absorbers for solar thermal applications using simple techniques. The selectively absorbing film consists of a composite with nickel nano­particles embedded in a dielectric matrix of alumina.

The AR material should have the following properties: the proper refractive index, low thermal emittance, dense, flexible and long term stable. The AR materials tested were silica, alumina and mixtures of silica-titania. The refractive indexes of the above mentioned materials range from 1.4 (silica) to about 2.1 (50/50 molar ratio silica/titania mixture). Besides increasing the normal solar absorptance, asoi, it is equally important that the AR layer is long term stable in order to create a successful solar selective coating. The AR coatings were synthesized using different solution-chemical methods and deposited on the absorber surface by spin coating. Prepared samples were subjected to an accelerated lifetime test. In the test procedure the temperature of the environment was set to 40°C and the relative humidity to 95 %. Samples made with alumina as AR coatings failed the ageing test. The other materials, silica and silica — titania mixtures proved to be very resilient. Samples that were coated with these AR materials showed no visible degradation of the sample surface even after 600 hours of testing.

Absorbers without an AR layer typically attain a normal solar absorptance of 0.80 and a normal thermal emittance of 0.03. Of the samples made with durable films a 70/30 silica/titania mixture showed the greatest increase of the asol value, 0.91, while the thermal emittance remained unaltered.

Introduction

The most efficient thermal solar collectors for hot water production use a spectrally selective surface that absorb and convert solar radiation into heat. There are already high performing selective surfaces but there are a few difficulties with some of them, such as the long-term durability, moisture resistance, adhesion, scratch resistance, cost and complicated production techniques. In order to make thermal solar collectors more accepted and widespread, the price per unit has to decrease. The most costly component of a thermal solar collector is the spectrally selective surface.

The main aim was to investigate the durability of spectrally selective absorbers produced by a newly invented solution-chemical method. This work is a continuation of a preceding study
where spectrally selective absorbers were produced using a novel solution-chemical technique [1]. Advantages with this technique are that it is simple and easy to control, the coating can be manufactured under ambient pressure conditions, the chemicals involved are environmentally friendly and it is low in material consumption. Furthermore there exist several methods like spin-, flow-, spray — and dip-coating to coat a surface with a liquid medium. The method seems promising and could hopefully reduce production costs for absorbers and hence make them less expensive and more available. The focus in this part of the thesis has been set on the durability properties of anti reflection treated absorbers. The optical characteristics of produced samples before and after the accelerated ageing testing were investigated.

The used absorber belongs to a group of absorbers called metal-dielectric composite/metal tandem, which normally consists of metal embedded in a dielectric matrix applied on a metal surface [2]. The absorbing layer, spin coated on an aluminum substrate, consists of nickel particles embedded in an aluminum oxide matrix. The composition of the absorbing layer is 65 volume percent nickel and 35 volume percent alumina and the thickness is about 100 nm. The metal particles are between 5 — 10 nm in size. A major advantage with a composite is that it offers a high degree of flexibility. By varying the choice of particle, particle size, particle orientation and shape, film thickness and particle concentration in the film, innumerable combinations can be created. Thus spectral selectivity can easily be achieved. By applying the coating on a poor thermal emitter, in this case aluminum, a low thermal emittance value, stherm, is obtained. The normal solar absorptance value, asoi, for the absorbing layer is about 0.80 and the normal thermal emittance value 0.03.

The AR material should have the following properties: the proper refractive index, poor thermal emitter, dense, flexible and durable. A correct refractive index is required in order to obtain as high solar absorptance as possible. At the same time the AR material should be a poor thermal emitter not to increase the stherm value. Lastly the material has to withstand accelerated ageing tests in order to be successful and should therefore be dense and elastic.

Five different AR coatings were studied, alumina, silica, hybrid silica, and two compositions of silica — titania. Silica is well known to be a very resilient but static material. In order to make silica more flexible an organic compound can be incorporated into the structure and then the resulting material is called hybrid silica [3]. A flexible material is more likely to perform well in accelerated ageing tests since it is less prone to crack when heated or cooled. Alumina has a higher refractive index in the visible wavelength range than silica, 1.6 compared to 1.4. Pure titania has a refractive index of 2.7. Thus refractive indices between 1.4 and 2.7 can be obtained by mixing silica and titania.

The materials structure of a thin coating is not completely permanent with time. Factors like high temperature, high air humidity, airborne pollutants and sun radiation can cause the coating to deteriorate and hence affect the optical selectivity of the surface [4]. High temperatures can speed up oxidation processes and high levels of humidity may create hydrolytic reactions i. e. electrochemical corrosion. Airborne pollutants might also accelerate electrochemical corrosion processes and solar radiation can initiate photochemical redox reactions. A combination of these processes can be devastating for a large number of materials, including solar selective coatings. The most accurate method to test the durability of a solar absorber is to assess it under normal working conditions. These so called in-situ tests are though very hard to carry through because of the great time length required to get
satisfying results. Instead of exposing the absorber surface to its natural working conditions for many years, inexpensive laboratory tests can be done in a climate chamber, where temperature and humidity can be controlled. The temperature and/or the humidity in the chamber can be elevated above normal levels in order to accelerate ageing processes.

Figs. 6-7 show polarisation curves of C/Al2O3/Al absorption surfaces and aluminium substrate at solutions pH 3.5(a), 5.5(a) and 5.5(b) (see Table 1). The conductivity of the solution pH 5.5(a) was too low for polarisation measurements without sulphate addition. The anodic current densities of the C/Al2O3/Al absorption surfaces and the aluminium substrate show passive behaviour in acid rainwater (pH 3.5a) in wide potential area up to -200 mV (Fig. 6a). The anodic current density of the C/Al2O3/Al surface is ca. decade lower than that of aluminium substrate probably indicating lower corrosion rate of the C/Al2O3/Al absorber surface. The current densities of the C/Al2O3/Al surface and aluminium substrate are higher in acidic rainwater at pH 3.5(a, b) than at neutral rainwater at pH 5.5(b). This indicates higher expected general corrosion rate of the C/Al2O3/Al surface and the aluminium substrate at pH 3.5 than at pH 5.5. The current densities of aluminium substrates are slightly higher than those of the C/Al2O3/Al surfaces in neutral rainwater at pH 5.5 and in acid rainwater at pH 3.5, thus indicating possibly better corrosion resistance of graphite coated aluminium compared to an aluminium substrate.

Aging time [h]

Fig. 3. Changes in solar absorptance for samples exposed to simulated acid rain immersion tests. Samples analyzed with EIS denoted as A and B.

0 20 40 60 80 100 120 140

Aging time [h]

Fig. 5. PC values for samples exposed to simulated acid rain immersion tests. Samples analyzed with EIS denoted as A and B.

Figs. 8 — 9 show the EIS impedance spectra of an C/Al2O3/Al absorption surface and an aluminium substrate at the beginning of the test, after 1 and 5 days at pH 3.5(a) and after 1 and 14 days at pH 5.5(a), respectively. The passive metal in a passive state is characterised by a wide capacitive area, i. e. linear plot in the Bode impedance diagram. The faradaic charge transfer reaction associated with corrosion gives rise to a finite resistive element at low frequencies. Therefore the impedance at low frequency area decreases as a result of the initiated corrosion. Indications of material degradation are also a decrease of the resistance element in the high frequency region and the change from a largely capacitive to a resistive behaviour. Generally, the constant impedance value in low frequency area i. e. the polarisation resistance is inversely proportional to the corrosion rate of the material.

In neutral rainwater (pH 5.5(a), Fig. 8) an C/Al2O3/Al absorber surface exhibited mostly capacitive behaviour during the 14 days of immersion. The corrosion resistance of an aluminium substrate seemed to be lower than that of the C/Al2O3/Al surface. Reason for this is unclear, but possible causes are minor failures in the nolan lacquer insulation of the samples. Generally, the corrosion probability of aluminium is lower at pH 5.5 than at pH

3.5 (Pourbaix, 1966).

At higher potentials the impedance is probably distorted due to the low conductivity of the solution (a) (Table 1). The C/Al2O3/Al samples were lightened after immersion. The impedance of an C/Al2O3/Al absorber surface between 1 and 1000 Hz (Fig. 9) decreased during the immersion test high frequency area indicating possible decrease in graphite coating quality. However, this sample did not exhibit typical optical degradation (Sample B in Figs. 3-5).

In acidic rainwater (Fig. 9) the impedance of the C/Al2O3/Al surface and the aluminium substrate in low frequency area are relatively low and therefore expected corrosion rate is higher than at pH 5.5(a). The increase in the impedance values after 1 days indicates some kind of passivation of the surfaces. However, after 5 days the impedance at low

SHAPE * MERGEFORMAT

al., 2003a) on the surface is oxidated through chemical reactions forming CO, CO2 and other compounds (Hihara and Latanision, 1994). The revealed A^O3 layer subsequently probably follows typical alumina-aluminium corrosion mechanisms. We used FTIR — spectroscopy for determining the hydration level of the absorber substrate, consisting of thicker than naturally formed heterogeneous alumina layer on 0.5 mm thick aluminium substrate of 99.5 % purity.

The combined results are shown in Figs. 3-5. All samples inside dotted boxes contain identified Al2O3 hydroxides i. e. pseudoboehmite and/or boehmite, possible other forms of hydroxides as well (e. g. bayerite and gibbsite). Detailed analyses of the results show that hydroxides are identified in all samples exposed to pH 5.5 except one sample. The most degraded samples (PC > 0.18) have absorption bands related to both pseudoboehmite and boehmite (see Fig. 5 in (Konttinen et al, 2004)), whereas the rest of the hydrated samples do not have the characteristic absorption band of pseudoboehmite.

Discussion

Total-immersion test results were compared to standard condensation test results (Konttinen and Lund, 2003) in order to determine correlation between the two methods. Samples degraded to PC=0.05 in total-immersion tests exhibit generally similar p*. as the samples degraded to PC=0.05 in standard condensation tests (for details, see (Konttinen et al, 2004)). Still, when comparing all the optical results (Figs. 3-5) we can conclude that the total-immersion method (Fig. 1) without controlled solution movement turned out not being optimal as the samples did not exhibit clearly detectable temperature and time dependencies. This method was chosen instead of e. g. rotating disk method (Magaino, 1997) because impedance spectroscopy (EIS) tests required the use of a liquid electrolyte in contact with the sample. We assumed that the gas feed and natural convection would rotate the solution inside the flask sufficiently enough for reproducible test results. It seems that this was not the case at least at pH 5.5. A controlled solution or sample movement combined with controlled temperature and gas feeding rate should be implemented in the future tests.

Semi-integrating device attached to the FTIR-spectrometer increases interference thus disturbing the exact determination of precise absorption peaks. Therefore we had to make rougher analyses of absorption bands.

Hihara and Latanision (Hihara and Latanision, 1994, pp. 251-252) noted in their study about corrosion of C/Al2O3/Al metal matrix composites (MMC) that proton reduction on graphite will polarise aluminium to noble potentials, explaining the negligible galvanic corrosion rates in de-aerated solutions for MMC. Their results are not directly comparable to ours as the MMC materials differ in structure and purpose of use from the C/Al2O3/Al absorber surfaces. Although our results for samples A-B do not generally match to those for MMC (Hihara and Latanision, 1994) with regard to aeration, it is possible that similar polarising phenomena occurs on the absorber surfaces as well under similar conditions, and may corrode the C/Al2O3/Al absorber surfaces during long-term natural exposure.

As a reference we exposed three C/Al2O3/Al absorber samples to non-aerated total — immersion at the room temperature, one at each pH. The samples exposed to pH 5.5 and pH 4.5 were hydrated approximately to the level of PC « 0.5-0.6 after 30 and 80 days, respectively. Sample exposed to pH 3.5 did not show any degradation after 160 days by visual inspection.

Conclusions

Tests reported in this paper have provided with a general picture on the degradation effect of acid rain (pH 3.5 or 4.5) and neutral rain (pH 5.5) rain on the rough C/Al2O3/Al solar absorber surfaces at 60, 80 and 99°C. In order to provide more accurate temperature — and time-dependent results the total-immersion test method used needs to be further
developed to include controlled movement of solution or sample. Similar to standard tests (draft proposal ISO/CD 12592,2) the main degradation mechanism has been found to be hydration of aluminium oxide (especially at pH 5.5). It is possible that pH 3.5 is too acidic for aluminium hydroxides to be formed, thus preventing further corrosion of the C/Al2O3/Al surface.

For aluminium/aluminium oxide containing samples FTIR-spectroscopy can be used for determining stages of hydration. In our tests we have observed similar degradation for another type commercial aluminium substrate based absorber surface as well.

According to the electrochemical measurements the corrosion rate of aluminium substrate is faster at pH 3.5 than at pH 5.5. At pH 3.5 the anodic current density of the C/Al2O3/Al surface is smaller compared to aluminium substrate. Electrochemical measurement results for aluminium substrate do not deviate significantly from the results for the C/Al2O3/Al surfaces, indicating that the electrochemical measurements measure corrosion characteristics of aluminium to a large extent. Optical degradation of C/Al2O3/Al surface is mainly due to hydration of aluminium oxide, and this phenomenon was not clearly detectable in these electrochemical measurements. With another type of EIS test system setup, it may be possible to obtain more coherent results.

All these results indicate that unglazed solar absorber surfaces based on aluminium substrate need to be well protected against rain diffusion onto the substrate in order to prevent degradation caused by hydration of aluminium oxide.

Acknowledgements

We wish to thank Mr. Mikko Mikkola for setup of the gas distribution system and Mr. Iwao Nitta for translating reference (Takahashi et al., 1987) into English.

References

ASTM, 1987. Standard reference method for making potentiostatic and potentiodynamic anodic polarisation measurements, G5. 1987, Annual book of ASTM Standards.

ASTM, 1989. Standard practice for calculation of corrosion rates and related information from electrochemical measurements, G102, 1989, Annual book of ASTM Standards. Brunold, S., Frei, U., Carlsson, B., Moller K., Kohl, M., 2000a. Accelerated life testing of solar absorber coatings: Testing procedure and results. Sol. Energy 68, 313 — 323. Carlsson, B., Frei, U., Kohl M., Moller, K., 1994. Accelerated Life Testing of Solar Energy Materials — Case study of some selective materials for DHW-systems. IEA SHCP Task X. Hihara, L. H., 1997. Corrosion of aluminium-matrix composites. Corros. Rev. 15, 361-386. Hihara, L. H., Latanision, R. M., 1994. Corrosion of metal matrix composites. Int. Mater. Rev. 39, 245-264.

Howells, G., 1990. Acid rain and acid waters. Ellis Horwood Limited, Chichester, p. 14. Konttinen, P., Kilpi, R. Lund, P. D., 2003a. Microstructural analysis of selective C/Al2O3/Al solar absorber surfaces. Thin Solid Films 425, 24-30.

Konttinen, P., Lund, P. D., 2003. Thermal stability and moisture resistance of C/Al2O3/Al solar absorber surfaces. Sol. Energy Mater. Sol. Cells, in press.

Konttinen, P., Lund, P. D., Kilpi, R. J, 2003b. Mechanically manufactured selective solar absorber surfaces. Sol. Energy Mater. Sol. Cells 79, 273-283.

Konttinen, P., Salo, T., Lund, P. D., 2004, Corrosion of unglazed rough graphite-aluminium solar absorber surfaces in simulated acid and neutral rain, submitted to Solar Energy Lorenz, W. and Mansfeld, F., 1981. Determination of corrosion rates by electrochemical DC and AC methods. Corrosion Science 21, 647 — 672.

Magaino, S., 1997. Corrosion rate of copper rotating-disk-electrode in simulated acid rain. Electrochim. Acta 42, 377-382.

Magaino, S., 1999. Zinc corrosion in simulated acid rain. Electrochim. Acta 44, 4307-4312.

Shreir, L. L., Jarman, R. A., Burstein, G. T., 1994. Corrosion. Butterworth Heinemann, Oxford, pp. 19:18-19:20.

Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon, New York, 1966.

Takahashi, H., Yamagami, M., Furuichi, R., Nagayama, M., 1987. FTIR Analysis of Hydroxide Films on Aluminum. J. Surface Sci. Soc. Japan 8, 279-281 (in Japanese). Tanemura, S., Yoshimura, K., Taga, K., Odaira, R., Ishida, M., Tsuboi, S., Yoshikawa, M., 1990. Accelerated aging testing of Ni pigmented anodized Al under heat and/or high humidity loads and degradation mechanism. In: Horigome, T., Kimura, K., Takakura, T., Nishino, T., Fujii, I., (Eds.), Proceedings of the 1989 Congress of the International Solar Energy Society, 4-8 September 1989, Kobe City, Japan. Pergamon Press.

Due to the integration of the air flow model into Type 56 the TRNSYS Solver can no longer be used for the iteration process between the two models. A into Type56 integrated solver is used to reach convergence. Depending on the boundary conditions the coupling of the two models forms a recursive flow of information with negative feedback. E. g. at cold outdoor air an increasing indoor temperature imposes an increasing exchange of air. In turn this counters another increase of the indoor temperature so that a balance will appear (Fig. 4). If the negative feed back gets too strong there will be the risk of getting an unstable system. A solver algorithm with a successive substitution method (TRNSYS solver 0) then also becomes unstable. Therefor the new solver integrated into Type 56 automatically dampens individually every recursive flow of information according to the iteration process so that an optimum of stability and convergence is achieved.

For TRNFLOW the functionality of PREBID has been enlarged. Beneath traditional data of the thermal model also the required input information of the air flow model can be entered into the new PREBID 5. This data is also stored in the BUI file and can be read in from it again. Analog to the Standard Type 56 files (TRN, BLD) the new PREBID 5 also creates a COMIS Input File (CIF) which is read by TrNfLOW. The CIF created this way is — with few additions — completely equivalent to the format described in the COMIS 3.1 User’s Guide [4] Unlike to former simulations with COMIS the COMIS Input File (CIF) is now checked to be consistent with the thermal model (bui-file).