Acta Polytechnica Vol. 43 No. 2/2002 Measurement of Moisture Stora.ge Parameters of Buildirg Materifis M. Jiiidkov,i, R. i.rny, P. Rovnanfkovd The mouture rtorage paratneters of three dffirent butlding materials: calcium silicate, ceramic brick and autoclaaed aerated, c,ncrete, are detennined in the hygroscopic rgnge and oaerhygroscopii range. Measured sorption isothenns and, moisture retentxon cunLes are then combined'into moisture storagefunctions using the Keluin equation. A comparisoi of measuredresutts with global characteristics of:th, p*e space obtained b1 mercury intru.sion porosimetry shows a reasonable agreement; the ined,ian pore radii by ook p ory well within the lntiraat gium b1 the beginning and the end of the characteristic steep parts if the moisture retenlzon cun)es. Keluords: moisture retention curae, sorption isotherm, d,esorption isotherm, buitding mrtterials. I Introduction Application of mathematical and computational models to simulate transporr processes in building materials requires knowledge of the material parameters that are used as input parameters of the models. Ti.r'o basic types of such parameters can be recognized, i.e., transport parameters and storage parameters. Tiansport parameters are the phenomenological coeffi- cients that appear in the constitutive equations as proportion- ality factors between the generalized thermodynamic forces and fluxes. Storage parameters need to be defined because transport equations are generally formulated for particular mass and energy densities, and in their original form they do nor con- tain the basic state variables that appear (in the form of their gradients) in the constitutive equations. The main role of storage parameters is that they make it possible to calculate the partial derivatives of the particular mass or energy densi- ties with respect to the basic state variables. In modeling moisture rransporr, it is necessary to include in the transport equation at least one moisture storage pa- rameter, depending on the class of moisture transport model. Howeve4 it is physically more correct to define two parame- ters, one for the hygroscopic range, and the 0second for the overhygroscopic range. In the hygroscopic range, the sorp- tion (or desorption) isotherm is the main parameter used in most models, while in the overhygroscopic range, the mois- ture retention curve is employed. These nvo moisture storage parameters can be combined in a single relation, the moisture storage function, using the Kelvin equation (see, e.g., [] for more details). 'fhe sorption and desorption isotherms of porous build- ing materials are very frequently rneasured in building-phys- ics laboratories, and extensive catalogues are available fbr many materials (e.9., [2]). On the other hand, measur€menrs of moisture retention curves for building materials are still relatively rare (e.g., t3l, t4l), in contrast to the numerous measurements on soils (see, e.g., [5] for an overuiew). The main reason for this is that building physicists have until now seen more interested in knolving the moisture related properties in the hygroscopic rurnge tl.ran in the overhygro- scopic range. This attitude is also rcflected in most building physical standards all over the world, where liquid water related properties are not considered. The neglect of liquid water rransport and storage proper- ties in calculations related to building physics may lead in many cases to a significant departure from reality. Typical examples are situations where wind driven rain plays an important role, where there is rising damp in a building, or when capillary active or hydrophilic interior thermal insula- tion is used. Therefore, simulraneous utilization of both liquid water and water vapor related transport and storage proper- ties in computational codes is very desirable, and advanced codes already take rhem into accounr (e.S., t6]). In this paper; the moisture storage parameters of three diflerent building materials: ceramic brick, calcium silicate and autoclaved aerated concrete, are measured in the hygro- scopic and in the overhygroscopic range. 2 Measuring methods 2.1 Adsorption and desorption isotherms The measurements of adsorption isotherms were per- formed in laboratory conditions, at 23+ I "C. The samples were placed in desiccators with di{Ierent solutions (see Fig. l) to simulate diflerent values of relative humidiry see Thble I [7]. The initial state for all rhe measurements was dry material. The experiment was performed in parallel in all desiccators in the same way. The mass of samples was mea- Fig. 1: Desiccator with measured samples 39 Acta Polytechnica Vol. 43 No. 2/2003 Table I: Relative humidity over the saturated solutions at 23± 1 oe Salt solution Relative humidity [%] LiCI 12 MgC12·6H20 33 NaN02 65 NH4CI 79.5 KNOs 94 ~Cr207 97 sured in specified time periods umil the steady state value of the mass was achieved. Then, the moisture coment by volume was calculated according to the equation m -ma w = s [m3/m3], V'Pw where ms is the mass of the wet sampie in steady state condi­ tions, mo is the initial mass of the sample (in dry state), Vis the volume of the sample, and Pw is the density of water at average temperature 20°C, Pw =998 kg/mj. The measurement of desorption isotherms was carried out in the same way as for the adsorption isotherms, except that the initial state was a capillary water saturated specimen. 2.2 Moisture retention curves The experiments were carried out using an ordinary pres­ sure plate device (see Fig. 2). It consists of a compressor inducing pressure above atmospheric pressure, a pressure panel with manometers and regulators, a pressure plate ex­ tractor, and ceramic plates (see e.g. [8]). Fig. 2: Experimental setup for moisture retention measurements The capillary saturated specimens were placed on an air­ proof and water saturated ceramic plate covered by a [tne kaolin layer and a fine meshed cloth in the pressure plate extractor (see Fig. 3). The extractor was closed, and a chosen pressure was applied. Water drained out of the outDow tube to the outf1ow burette to achieve equilibrium. When the out­ Dow of water desisted, the extractor was opened, and the mass of the specimens was determined by weighing. The experiment then continued at a new higher-pressure level. 40 Fig. 3: Placing specimens into the extractor or the pressure plate device After the measurements were finished the moisture content of the specimen was calculated at each pressure leve! and a mois­ ture retention curve was constructed. 3 Materials and samples The experiments were carried out on three building mate­ rials: calcium silicate plate, ceramic brick and autoc!aved aerated concrete (AAC). These materials were chosen be­ cause of the differences in pore structure, homogeneity and complexity. Calcium silicate plate is a low-density board product. In building structures it is mainly used as capillary active inside insulation, because of its very high capillary absorption coeffi­ cient and capillary moisture content. It is largely composed of synthetic mineral xonotlite, a complex calcium silicate hydrate (see [9]). Randomly orientated cellulose fibres are also present. Calcium silicate plate has mainly a fine pore structure, comprising the voids between matted acicular crys­ tals. The material can be identified as rather homogeneous and very porous. Compared to calcium silicate plate, ceramic brick has a tubiform pore structure (see [10]). The total open porosity of the material is much lower. Due to the nature of the material and its production process, higher variability of the material properties can be expected. The most typical feature of AAC is the presence of artifi­ cial air pores created during the production process. The strueture of AAC includes spherical air pores. The walls of these pores consist of grains of sand embedded in a cement­ -hrne matrix (see [lIJ). This matrix provides the cohesion of the material, and eonsists of fine plate-shaped crystals in ajumble. ln the measurements of adsorption and desorption iso­ therms, 10 test specimens of each material with overall dimensions of 30 x 30 x 10 mm; were prepared for each rela­ tive humidity. Experiments to determine the retention curves were car­ ried out with twe!ve capillary saturated speeimens of each material. The size of the specimens was 35-40 x 35-40 x 10-15 mm3 . Pressures ofO.16; 0.32; 1.06; 3.16 and 10.0 bar were applied. Acta Polytechnica Vol. 43 No. 2/2009 4 Results and discussion The results of adsorption and desorption experiments in the hygroscopic range for the particular marerials are summarized in Figs. 4-0, where the error ranges of the mea_ surcments are also indicated. Calcium silicate rias shown to be 11ery hygroscopic material, and its sorprion hysteresis (the difl'erence between the adsorption and desorption isotherms) was quite high. This is in accordance with the presumed fine pore structure of the material. On the other hand, ceramic brick appeared to be an almosr non-hygroscopic material. Therefore, its sorption hysteresis is nor so important for rnodeling the marerial behavior. AAC was found to be some- where in betrveen the above limits. Its hygroscopicity and also the sorption hysteresis was notable but not so high as in the case of calcium silicate. 3.5 c''{ 3.0 ctlll z.s I 2.0 8 r.s o 5 t.o o'6 0.5 0.0 0.12 ED J gt J o F 0.06 I o o .9 o = 0.00 E E o og o .9 = 0.8 ctt L I0.. o F 0.4 I o E o.z .2 o = 0.0 E c o o o a E 100 o 0.2 0.4 0.6 0.8 1 Relative humidity [-l Adsorption and desorption isotherms of calcium silicate 0.1 1 10 Capillary pressure [bar] Fig. 7: Moisture retention curve of calcium silicate 0.1 I 10 Capillary pressure [barl Fig. 8: Moisture retenrion cun,e of ceramic brick 5.0E-01 4.0E-01 3.0E-01 2.0E-01 1.0E-01 0.08+00 Fig. 4 l.0E+00 1.0E-01 t.0E-02 1.0E-03 1.0E-O4 Fig. 5: Adsorption and desorption isotherrns of ceramic brick Figs. 7-9 show the moisrure retenrion cun'es of rhe studied materials. The fastest decrease in moisture content with in- 0 0.2 0.4 0.6 0.8 Relative humidity [-] Fig. 6: Adsorption and desorption isotherms of AAC 0.1 1 10 Capillary pressure [barl Fig. 9: Moistule retention curve of AAC creasing pressure was exhibited by ceramic brick, where the characteristic edge on the moisture retention curve appeared at about 0.3 bar. This gives evidence of a substantial amounr of large capillary pores. C)n the other hand, the slowesr de- crease was obsen'ed for AAC, rvhere the edge was found at about 3 bar. This indicates that the relative amounr of caoil- larl pores in this material should be lowest among the studied materials. The results obtained for calcium silicare were in betleen. The edge on rhe moisture retenrion curve appeared at about I bar. Figs. l0-12 present the moisture srorage functions constructed from the desorption isotherms and moisture re- tention cur.res using recalculation of capillary pressure to relative humidity by the Kelvin equarion. Clearly, the hygro- 100 't.2E-01 c - 8.0E-02 Q' o E 4,0E-02 ,o = 0.0E+00 4l Acta Polytechnica Vol. 43 No. 2/2003 Relative humidity [-] Fig. 10: Moisture storage function of calcium silicate 0.2 -r--------------------, Material Vp Ap Tv Tsteep [cm3/g] [m2/g] [mm] [mm] Calcium silicate 3.22 75.00 0.27 0.14-1.44 Ceramic brick 0.10 3.12 2.70 1.44-4.80 I AAC 0.90 67.85 0.084 ?-0.48 the distribution curves, which can be negatively affected for instance by the presence of necks of larger pores (see, e.g., [12] for details). We therefore present only global characteris­ tics of the pore space of the particular materials. Table 2 gives the total intrusion volume Vp' the total pore area Ap' and the median pore radius by volume Tv' Table 2 Global characteristics of the pore space 0.80.60.40.2 0.0 -J---~=r_~--===;====:::::.:.-__..J O ­I::Cll-g 2.0 <.) ~ ~ 1.0 '0 :::E ~ 4.0,-------------------.., Cl....... Cl ~ 3.0 Fig. II: Moisture storage function of ceramic brick In order to compare the moisture retention data with the mercury porosimetry data, the capillary pressures corre­ sponding to the beginning and end of the steep parts of the moisture retention curves were recalculated to the respeClive pore radii. The results are shown in Table 2 in column T,«ep' where the first number corresponds to the end and the second number corresponds to the beginning of the steep part of the particular moisture retention curve. Clearly, the median pore radii determined by mercury intrusion porosi­ metry fall within the Iimits given by these two values, so that the agreement between the two methods can be considered as satisfactorily. 0.4 0.6 0.8 Relative humidity [-j 0.2 OL-__~---=========..J O ]i Cl ~ 0.15 C .2!5 0.1 <.) e E 0.05 Vl '0 :::ii 0.8 -,-------------------, Fig. 12: Moisture storage funClion of AAC 5 Conclusions Solution of the increasingly complex problems of mois­ ture transport in building materiaJs requires the application of advanced computer codes involving simultaneous utiliza­ tion of both Iiquid water and water vapor related storage properties. However, in many building-physics laboratories only the hygroscopic parts of moisture storage functions - the sorption isotherms - are commonly measured. In this paper, the moisture storage functions of three different building materials were determined in the whole moisture range. The parts of the curves corresponding to the hygroscopic and overhygroscopic ranges correspondent together, and the curves were also in good agreement with other basic require­ ments for this type of function. A comparison of the liquid water related parts of the moisture storage functions with the results of mercury intrusion porosimetry exhibited reasonable agreement for all three materials. References Acknowledgements This research has been supported by the Ministry of Education of the Czech Republic, under contract No. MSM:21000000S. [I] Černý, R., Rovnaníková, P.: Transport Processes in Con­ cTele. London: Spon Press, 2002. 0.4 0.6 0.8 Relative humidity [-j 0.2 O .jL----,--------.,------,-----..--I O Cl... Cl ~ 0.6­I:: Cll 'EO 0.4 <.) ~ ~ 0.2 '0 :E scopic and overhygroscopic parts of the curves are mono­ tonie. They correspond for aH materials, and no ambiguities appeal'. The results of the moisture retention measurements should approximately agree with the porosimetric measure­ ments, as already indicated in the explanation of the shape of the curves. Therefore, additional mercury porosimetry measurements were carried out to verifY this agreement. In these experiments, a Micromeritics PORESIZER 9310 mercury intrusion porosimeter with a maximum working pressure of 200 MPa and pore distribution in the range SOO !lm to 0.006 !lm was employed. It is well known that the results of mercury intrusion porosimetry have to be interpreted with care, particularly 42 Acta Polytechnica Vol. 43 No. 2/2002 I2l Hansen, K. K.: Sorption Isothzrms. Technical Report 163/86. Lyngby; Technical Universiry of Denrnark, 1986. t3l Brocken, H.J. P.: Mo,isture Transport,in Brirk Masonry: thz GrE Area Behtem Brbks. phD Thesis, Eindhoven: TU Eindhoven, 1998. I4l Roels, S.: Mofuling Unsaturated Moisture Transport, in Hetcrogmcow Limcstone. phD Thesis, Leuven: Iktholieke Universiteit Leuven, 2000. t5l Hillel, D.: Fun"dnmentak of SoiI physics. New york: Aca_ demic Press. 1980. t6l Grunewald,J.: DELPHIN 4.1 - Docmnentation.Theoreti- cal Fundamentals. Dresden: TU Dresden. 2000. l7l Arai, C., Hosaka, S., Murase, K., Sano, y.: Measuremmts of thz Relntiae Humulity of Saturated Aqueuos Salt Solutions, J. Chem. Eng.Jap., Vol.9, 1926, p.323-330. t8] SBI Report 295: Retmtion Cur-ues Meas.ured [Jsing pressure Plnte aruJ Pressure Mentbrane Apparatw. Horshoim: Dan- ish Building Research Institute, 1998. tgl Hamilton, A., Hall, C.: Composition and Add,itiorurl Charac- lerisalion of Calcium Silicate Material: Reaision with neu Data. 51n Framework EU HAMS-|AD - project, working document, Edinburgh: University of Edinburgh, 2003. [0] Carmeliet, J., Descamps, F., Houvenaghel, G.: A Mutti- scaLe Netuorh Moful for Simulnting Moisture Transfer Properties of Porous Media. Transporr in porous Media, Vol.35, 1999, p.67-88. I I I ] Roels, S., Sermijn, J., Carmeliet, J. : MofutLing IJnsaturated Moislure Transport in Autockned Aerated Concrete: a Micro- structural Approach. In: Building physics in the Nordic Countries. Proceedings of the 6th symposium, Trond- heim, June 17-19, 2002. Eds. Gustavsen A. and Thue J, V., Trondheim: Norwegian Building Research Insti_ tute and Nonvegian University of Science and Technol- ogy, 2002. p2l Diamond, S.: Mercury poros,imctry: An hwppropriare Method for the Measuremmt of pore Sizz Distributions in Cement-based, Maferilrk. Cement and Concrete Research. Vol. 30, 2000, p. l5l7-t52b. Ing. MilenaJiiidkovii phone: +420 224 354 498 e-mail: jiricko@fsv.cvur.cz Prof. Ing. Robert eernf, DrSc. phone: +420 224 354 429 e-mail: cernyr@fsv.cvut.cz Department of Structural Mechanics Czech Technical University in Prague Faculty of Civil Engineering Thiikurova 7 166 29 Prague 6, Czech Republic Doc. RNDr. Pavla Rovnanikovd. CSc. phone: +420 541 147 633 e-mail: chrov@fce.vutbr.cz Institute of Chemistry Faculty of Civil Engineering Technical University of Brno ZiZkova 17 662 37 Brno, Czech Republic 43 Scan39 Scan40 Scan41 Scan42 Scan43