Construction and Building Materials

Transcription

1 Construction and Building Materials 25 (2011) Contents lists available at ScienceDirect Construction and Building Materials journal omepage: Microstructure and mecanical properties of b-emiydrate produced gypsum: An insigt from its ydration process Q.L. Yu, H.J.H. Brouwers Department of Arcitecture, Building and Planning, Eindoven University of Tecnology, P.O. Box 513, 5600 MB Eindoven, Te Neterlands article info abstract Article istory: Received 10 December 2009 Received in revised form 19 November 2010 Accepted 7 December 2010 Available online 26 February 2011 Keywords: b-emiydrate Hydration Ultrasonic Workability Microstructure Mecanical properties Tis article addresses te microstructure and related mecanical properties of gypsum produced from b- emiydrate using information of its ydration process, from fres paste to ardened gypsum. Te water demand of te investigated b-emiydrate was determined applying te spread flow test. Te flowability was studied and te deformation coefficient of te b-emiydrate was derived accordingly. Te ydration process of te b-emiydrate was investigated applying an ultrasonic wave metod and te influence of te water content on ydration was analyzed. Te microstructure of te generated gypsum was studied by experiments and modeling. A model [1] was applied, wic was successfully validated by te present experiments. Furtermore, te mecanical properties of te produced gypsum were investigated, and a numerical relation between water amount and strengt was found. Ó 2010 Elsevier Ltd. All rigts reserved. 1. Introduction Corresponding autor. Tel.: +31 (0) : fax: +31 (0) address: q.yu@bwk.tue.nl (Q.L. Yu). Gypsum (CaSO 4 2H 2 O, also known as diydrate) plaster is one of te earliest building materials elaborated by mankind, and gypsum plasterboard is used widely as indoor building material because of its easy fabrication feature, environmental friendliness, fire resistance, aestetics, low price, etc. Te global production of gypsum in 2008 is over 250 million ton. Taking Europe as an example, tere are 160 quarries for gypsum production and te number of employees is over 85,000 [2]. Te CaSO 4 H 2 O system is composed of five solid pases: diydrate, emiydrate (CaSO 4 0.5H 2 O), anydrite I (CaSO 4 ), anydrite G (CaSO 4 ), and anydrite I (CaSO 4 ) [3]. Among tem emiydrate is usually used to produce gypsum plasterboard. Hemiydrate occurs in two forms, i.e. te a- and b-type, wereas b-emiydrate is mainly applied since te ydration product of te a-emiydrate is too brittle to be used as building material. Te properties of gypsum are influenced by te ydration process and te properties of te emiydrate. Te ydration of emiydrate suc as te ydration kinetics and setting as been studied intensively so far [4 11], but te empasis was always laid on a-emiydrate. Terefore topics suc as te ydration induced properties, and especially te beavior of te b-emiydrate upon ydration to produce gypsum (plasterboard) is still poorly understood. Tis article addresses te properties of gypsum (plasterboard) produced from a b-emiydrate from its ydration process. Te water demand of b-emiydrate, as well as te flowability of te ydrating system was investigated by te common spread flow test. Te ydration of b-emiydrate was investigated applying an ultrasonic metod and te influential factors on setting were studied. Te microstructure as well as induced mecanical properties of te gypsum during ydration was investigated using experimental data and modeling. 2. Material and experimental 2.1. Material Te applied material, i.e. b-emiydrate, in te present study was provided by Knauf Gips KG (Germany). Te b-emiydrate is produced from te flue gas desulfurization (FGD) gypsum. Te b-emiydrate was caracterized in te present study in order to evaluate its pysical properties suc as: mineralogy, element, particle size distribution, specific surface area, and microstructure. X-ray diffraction (XRD) patterns were obtained using a Rigaku diffractometer operating at room temperature in order to confirm te mineralogy and crystallinity. Te cemical analysis of te material was carried out using energy-dispersive X-ray spectroscopy (EDX). Te particle size distribution (PSD) was measured wit a Mastersizer Te specific surface area was analyzed from bot te measured PSD results and te BET metod (TriStar 3000, Micromeritics). Te microstructure of te ydrated system was measured wit scanning electron microscope (SEM) Water demand determination Flowability is widely used to describe te properties of building materials like concrete or gypsum in fres state and it is related to parameters suc as fluidity, mobility, and compactability. To assure te ydrating system is fluid, a tin layer of adsorbed water molecules around te particles and an extra amount water to fill /$ - see front matter Ó 2010 Elsevier Ltd. All rigts reserved. doi: /j.conbuildmat

2 3150 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Nomenclature a specific surface area, mass based (cm 2 /g) a 0 parameter b 0 parameter d diameter (mm) d 0 te base diameter of te Hägermann cone (100 mm) (mm) E te deformation coefficient F rupt breaking load (N) F c maximum load at fracture (N) mass of te emiydrate (g) initial mass of te unreacted emiydrate (g) H eigt of prism (cm) K sp te solution product of gypsum L 0 te distance between te supports of te bending macine (cm) m mass (g) n ydration degree V volume (cm 3 ) W widt of prism (cm) w mass of te water (g) initial mass of te water (g) Greeks a activities of te free ions b te water/emiydrate percentage (vol.%) for U =0 U te relative slump of te gypsum slurry q density (g/cm 3 ) c te super saturation of gypsum u volume fraction d water layer tickness (cm) n sape factor r strengt (N/mm 2 ) r 0 strengt at u v = 0 (N/mm 2 ) x specific molar volume (cm 3 /mol) m specific volume (cm 3 /g) Subscripts Blaine Blaine metod c compressive strengt f flexural strengt g gypsum emiydrate n nonevaporable s srinkage spere spere w capillary water v void fraction te intergranular voids of te system is necessary [12]. Hence, te determination of water demand of fine powders is of vital importance. Wit te filled void fraction, te proportion of total specific surface area of te powders and te remaining water determines te workability [13]. Altoug many metods likeanglesflowboxtest, flowtabletest, orslumptestcan be used to determinete workability, te spread flow test was deployed in te present study because it is especially suitable to measure materials wic ave a collapsed slump. Te spread flow tests were carried out first by filling te test sample into te Hägermann cone and ten lifting it to allow a free flow of te sample to a dry, clean, orizontal and non sucking working surface (e.g. a glass plate was used ere). Te test was performed according to EN [14]. Ten mixes wit different / were measured in order to obtain a statistically reliable trend line for te regression analysis. A more detailed information about te test procedure can be found in [12] Hydration process measurement Eqs. (1) and (2) sow te ydration of emiydrate, wic takes place quickly after mixing it wit water via a troug-solution route [15], firstly te emiydrate dissolves (Eq. (1)) and ten te generated diydrate precipitates from te solution (Eq. (2)). Eq. (3) sows te integral ydration reaction as a combination of te foregoing two steps CaSO 4 0:5H 2O! Ca 2þ þ SO 2 4 þ 0:5H 2O ð1þ Ca 2þ þ SO 2 4 þ 2H 2 O! CaSO 4 2H 2 O ð2þ CaSO 4 0:5H 2O þ 1:5H 2O! CaSO 4 2H 2O ð3þ Te ultrasonic wave tecnology is widely used to detect te microstructure defects, variation and fracture of solids since te ultrasonic wave canges in different media [16]. Pani et al. [16] studied te strengt and elastic modulus of gypsum using te ultrasonic tecnology. In te present researc, te ultrasonic tecnology was deployed to measure te b-emiydrate ydration process, since te microstructure of te ydrating system experiences a continuous cange due to firstly te dissolution of te emiydrate and ten te generation and te precipitation of te diydrate [7] during te wole process. Te emiydrate ydration experiments wit te ultrasonic metod were performed at te University of Stuttgart, applying te FresCon system as sown in Fig. 1 [17]. More detailed information about te FresCon system and te measurement procedure was presented in [18] Microstructure and mecanical properties measurement Fig. 2 sows scematically te volume composition of te ydrating system. Before te ydration, te total volume is composed of only water and b-emiydrate as sown in Fig. 2a. During te ydration reaction as sown in Eq. (3), part of te water and b-emiydrate are consumed as reactants and gypsum is produced, so te total volume is composed of te cemical srinkage due to te ydration reaction, te water tat remains beyond te ydration reaction, te generated gypsum, and te remaining b-emiydrate, sown in Fig. 2b. Te void fraction of te ydrating system ere is defined as te fraction of void space in te ydrating system, were te void may contain air or water. Terefore te void fraction of te system upon te ydration reaction is described as te Fig. 1. Te FresCon system for ultrasonic velocity measurement [17].

3 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Fig. 2. Scematic diagram of te volume composition of te system (a: initial condition, m = 0 and b/c: upon ydration, m > 0). addition of te volume caused by te reaction srinkage and te remaining water as sown in Fig. 2c. Te microstructure of te ydrating system consequently influences te performance of te ydrated system suc as te mecanical properties. Te microstructure was investigated by bot modeling and experiments, including SEM analysis and mecanical testing. 3. Results and discussion 3.1. Material caracterization Fig. 3a sows te XRD result of te used b-emiydrate. Te peak sows tat te b-emiydrate almost only consists of CaSO 4 0.5H 2 O. Te cemical composition analysis measured by EDX metod is listed in Table 1. It sows tere are only very small impurities (Si, Mg and Al in total less tan 3% by mass) in te sample, and also te amount of Ca and S is in line wit te teoretical values, wic also confirms te XRD result. Te PSD of te b-emiydrate is sown in Fig. 4, wit te surface weigted mean (D [3,2]) of 6.07 lm. Te specific surface area was calculated from te measured PSD results based on an assumption tat all te particles are speres [19], yielding m 2 /g ( cm 2 /cm 3 ). A detailed calculation description can be found in [13,19]. Te measured BET surface area of te b-emiydrate is 7.50 m 2 /g, and results sow tere are no micropores inside Water demand and flowability analysis A concept of relative slump is used to analyze te water demand and workability in te present study. Te relative slump is calculated from C ¼ðd=d 0 Þ 2 1 Here d is calculated from te average value of te two perpendicular diameters measured from te spread of te omogeneously prepared sample. All te computed U based on Eq. (4) ten are plotted versus te respective water/emiydrate volume ratio (V w /V ) as sown in Fig. 5, te relation between V w /V and / is as follows: V w ¼ =q w ¼ m w V =q m A linear trend line is fitted troug te plotted values, reading V w V ¼ E C þ b From Eq. (6), te water demand (b ) of te b-emiydrate, wic represents tat in tis condition te slump flow equals to ð4þ ð5þ ð6þ zero, is derived. Tis indicates te minimum water demand to assure a fluid gypsum plaster. In te present study, a b value of was obtained, wic is in line wit [20]. Tis also confirms tat te spread flow test is suitable for te determination of te water demand of te emiydrate. Te deformation coefficient (E ) is also derived from Eq. (6) (te slope of tis function). Tis value indicates te sensitivity of te materials on te water demand for a specified workability, wic means tat te material wit a lower deformation coefficient sows a bigger cange in deformability to a certain cange in water amount [13]. A deformation coefficient of for cement (CEM I/B 42.5 N LH/HS) was given [21] wic is bigger tan tat of b-emiydrate (0.053) ere. Tis indicates tat water as a bigger influence on te workability of te ydrating system, wic is probably due to te smaller surface area of te b-emiydrate (9877 cm 2 /cm 3 ) tan tat of te mentioned cement (15,300 cm 2 / cm 3 ). As discussed above, a tin layer of adsorbed water molecules around te particles is necessary to assure te fluidity of te ydrating system. Brouwers and Radix [21] reported tat te tickness of tis layer (d) is related to te deformation coefficient and te surface area of te used material, wic was later confirmed by [13], reading E ¼ d Blaine a Blaine q ¼ d n a spere q wereby n is te sape factor and a spere te surface computed using te PSD and assuming speres. In te present study, using a a Blaine value of 3025 cm 2 /g [22] and a q value from Table 2, ad Blaine value of 66.8 nm is obtained, wic is in line wit [21] wo reported a d Blaine value of 44.6 nm for CEM I/B 42.5 N LH/HS. Furtermore tis is also confirmed by Marquardt [23] wo reported a d value of 45 nm wit a different test, wic was discussed in [12]. A linear relation was reported [13] between te specific surface area from Blaine metod and computed specific surface area from PSD metod, given by n a spere ¼ 1:7 a Blaine Substituting te surface areas yields n = Hydration process analysis Te measured ultrasonic velocity during te ydration wit different / is sown in Fig. 6 [17]. Te velocity curve of te ultrasonic wave as te following caracteristics. Te first part is a dormant period, wic is distinguised by a constant low velocity value; and is ten followed wit te second part in wic te velocity increases rapidly until finally reaces a plateau, wic is caracterized also by a constant value. Te constant velocity value in te first period indicates te microstructure of te ydrating system remains stable or we can consider tis period as te induction time of te b-emiydrate in water. Eq. (9) gives te expression of te super saturation of te gypsum, were te K sp equals 4.62 [4]. Nucleation and growt of te gypsum take place wen c >1. c ¼ aðca 2þ ÞðSO 2 4 Þ=K sp ð9þ Te ultrasonic velocity canges quickly after te induction time due to te cange of te void fraction of te ydrating system and te pysical properties of te materials in te system [24,25], wic indicates te generation and precipitation of gypsum. Te measured values sow tat te ultrasonic velocity increases quickly in te precipitation period, but te increase rate gradually slows down until finally a plateau is reaced. Tis can also be explained by Eq. (9) tat te reduction of te free ions in ð7þ ð8þ

4 3152 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Fig. 3. Te X-ray diffraction patterns (a: b-emiydrate; b: gypsum produced wit / = 0.65; and c: gypsum produced wit / = 0.80). Table 1 Cemical composition analyzed by EDX. Element wt.% at.% K-ratio O Mg Al Si S Ca Total te solution due to te precipitation of te gypsum leads to te decrease of te ydration speed. Te plateau of te ultrasonic velocity indicates te ending of te setting. Te measured results indicate tat ultrasonic wave metod is suitable for te emiydrate ydration measurement. Te measured initial setting time of about 4.5 min under te condition of te / of 0.65 is in line wit [20], in wic te setting was measured wit te knife metods [14]. Te setting time measured using tis metod in tis study is muc sorter compared to te value from oter metod like electrical resistance metod [7]. Tis probably can be explained by te difference between te microstructure of te used materials, wic also indicates te used b- emiydrate as an obvious influence on ydration. Te ydrating system wit different / sows different workability as analyzed above. Tis finding indicates te distance between te molecules of te ydrating system is different under

5 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Fig. 4. Te particle size distribution of te b-emiydrate. Proportion V w /V V w /V = R² = Fig. 5. Results of te spread flow test, plot of relative slump flow versus water/ emiydrate volume ratio. different water content conditions. Te distance between te molecules becomes larger wen te water amount is iger wic in turn sould result in a longer reaction time. Tis is confirmed by te measured results wit ultrasonic wave metod sown in Fig. 6. Te results sow te induction time becomes longer wit an increase of te water amount. Te induction time is about 1.5 min wit te / of 0.65 wile around 7 min wen te / increases to Te longer induction time wit te iger / as sown in Fig. 7 can also be explained by Eq. (9), te generated gypsum needs longer time to be super saturated. Te relation between te water amount and induction time found in tis study is sown in Fig. 7. Te influence of te water amount on te period of precipitation is reflected from te period duration and te velocity variation. Wit te / of 0.65, te duration of tis period is only around 7 min and te final velocity reaces to around 2500 m/s, wile wit te / of 1.25, te duration of tis period is about 18 min and te final velocity reaces about 2000 m/s. Te duration of te precipitation period indicates evidently tat te water amount influences greatly te ydration process of te emiydrate. Te result of te ending time of te ydration is sown in Fig. 7 too. Results indicate tat water content at a lower / as a dominant influence on te final setting of te ydrating system. Tis finding is confirmed by Sciller [26] wo reported a similar trend of te relation between water ratio and setting time. But Table 2 Parameters of te paste model [1]. Substance M (g/mole) q (g/cm 3 ) x (cm 3 /mole) v /v w w n / v n /v w v n w n /v w v s /v Anydrate b-hemiydrate Diydrate a b c Fig. 6. Ultrasonic velocity during b-emiydrate ydration (a: / : 0.65; b: / : 0.80; and c: / : 1.25) [17].

6 3154 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) tis finding is in contrast to Solberg and Hansen [9] wo reported tat ydration of te gypsum is almost independent of water amount witin te / range of Fig. 3b and c sows te XRD pattern of te ydrated gypsum wit / of 0.65 and 0.80, respectively. Te results sow obviously tat te ydrated system only include CaSO 4 2H 2 O, wic indicates tat te b-emiydrate is fully ydrated and te water content does not influence te final composition of te ydrated system. Te difference between te final velocities indicates te final ydrated system as different microstructures due to te different initial water amount, wic will be discussed in next section. Tis also can be explained from te so called void fraction of te system, wic will be discussed in te next section as well Microstructure of te ydrating system Te cemical srinkage of te ydrating system is computed from te difference between te reactant water and te cemical combined water following from Eq. (3). In tis study a model proposed by Brouwers [1] is applied to describe te volume fraction of cemical srinkage as follows: n 1 v vn u w s ¼ m m w þ w n ð10þ Table 2 lists te values of m n /m w, w n /, m /m w for CaSO 4 systems [1]. Here te void fraction of te cemical srinkage upon te ydration is defined as te proportion of te reaction srinkage to te total volume tat is constant during te wole reaction process, wic is in line wit Sciller [6] wo also studied te void fraction of gypsum produced from a-emiydrate based on te same assumption. Te specific volume (cm 3 /g) of te cemical combined water (m n ) used ere is computed from te difference of te specific volume between te generated gypsum and te b- emiydrate. As discussed above, due to te requirement of te fluidity, an extra amount of water is needed. During te ydration reaction, te extra water works as transport liquid, and ten it is evacuated during te curing period, wic in turn leads to an open space of te generated gypsum. Te void fraction caused by tis unydrated water reads u w ¼ 0 n wn m m w þ ð11þ Terefore te void fraction during te ydration follows from te addition of te two parts, reading Fig. 7. Analysis of te ydration process. u v ¼ u w þ u s ¼ n m mn w m m w þ w n ð12þ Substituting te values of te b-emiydrate for te m n /v w, x n /, m / v w from Table 2, we obtain u v ¼ u w þ u s ¼ 0 0:13n 0:38 þ ð13þ It is evident from Eq. (13) tat te void fraction is related to te water content and te ydration degree (n) of te ydrating system. Unlike cement, b-emiydrate ydration occurs very quickly, for instance in te present study wit te / of 0.65 te full ydration ends witin 10 min (see Fig. 7). Terefore te influence of water on void fraction in tis study is mainly focused on te case tat te ydration degree is unity, i.e. te system experiences te full ydration, yielding u v ¼ 0 0:13 0:38 þ ð14þ Sciller [6] proposed a void faction model for gypsum, wic is in accordance wit te model applied ere (Eq. (13)), valid for te special case of full ydration and a-emiydrate only. Te applied model is verified by experiments in te case of b- emiydrate and full ydration. Te measured void fraction of te gypsum is plotted versus /, togeter wit te computed values from te applied model (Eq. (14)) sown as Fig. 8. Te good agreement between predicted value and experimental data reveals te validity of te applied model. Te microstructure of te generated gypsum is sown in Fig. 9. Te gypsum in Fig. 9a and b was prepared wit a / of 0.65 and 0.80, respectively. It is clear tat te crystals in Fig. 9a are muc tigter and omogeneous tan tat in Fig. 9b. Tis sows tat te gypsum produced wit a lower water content as a better bonding between te gypsum crystals wic sould lead better mecanical properties. Gypsum in Fig. 9a also sows a smaller void fraction compared to te gypsum in Fig. 9b, wic is also in line wit te applied model above Mecanical properties of gypsum As typical for many building materials, gypsum beaves elastically under a load before reacing its elastic limit. However, it is obvious tat te ardened gypsum is not a compact solid due to its ig void fraction. Lewry and Williamson [27] reported tat te gypsum strengt develops during setting via a tree-stage process: first a development of an interlocking matrix of diydrate needles, ten te relief of internal stress due to te build-up of pressure as needles, and final a strengt increase during te removal of te excess water. From te strengt development process, one can see tat te final strengt of te gypsum is related to te formation of te produced gypsum crystals and te bond between tem. From ere we can assume tat te water amount influences te strengt of gypsum by influencing te void fraction since it influences te formation of te gypsum crystals and te bonds, wic is confirmed by te SEM pictures as sown in Fig. 9a and b. Te strengt tests were performed using a Zwick test device according to [14]. In tis study te size of te samples was controlled strictly according to EN [14] to promise a representative value since Coquard et al. [28] reported tat te sample size as an obvious influence on te strengt of gypsum. Te flexural strengt was measured wit te tree-point bending metod, and te flexural strengt is calculated from

7 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Fig. 8. Te void faction versus te water-emiydrate ratio ( / ). a r f ¼ 3F ruptl 0 2WH 2 ð15þ Te compressive strengt was determined by applying a load to te broken parts of te original specimen (40 mm 40 mm 160 mm) used for te flexural strengt test, and te compressive strengt is calculated from F r c c ¼ ð16þ 1600 mm 2 Te influence of te water content on strengt of te gypsum was investigated and te measured flexural and compressive strengt values are plotted versus void fraction as sown in Fig. 10. It is evident tat te strengt decreases wit te increase of te void fraction, wic can be explained by te bond between te crystals. Wen te water amount increases, te void fraction of te generated gypsum increases as analyzed above wic in turn leads to a weaker bond between te gypsum crystals, wic finally results in a decreased strengt. Several models [16,28] were proposed to describe te relation between te void fraction and te strengt of gypsum, suc as r ¼ r 0 ð1 b 0 u v Þ r ¼ r 0 ð1 b 0 u v Þ a 0 ð17þ ð18þ Fig. 10. Te strengt of te specimen versus te void fraction (a: flexural strengt and b: compressive strengt). b Fig. 9. Te SEM picture of te gypsum (a: produced wit / = 0.65 and b: produced wit / = 0.80).

8 3156 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) Table 3 Comparison of te four applied models. Model Flexural strengt (r f, N/mm 2 ) Compressive strengt (r c, N/mm 2 ) r 2 a 0 b 0 r 0 r 2 a 0 b 0 r 0 Lineal (Eq. (17)) Pani (Eq. (18)) Sciller (Eq. (19)) Exponential (Eq. (20)) r ¼ a 0 ln b 0 u v r ¼ r 0 expð b 0 u v Þ ð19þ ð20þ In tis study tese models were applied and compared. Te obtained coefficient of determination (r 2 ) and relevant parameters (a 0, b 0, and r 0 ) are listed in Table 3. One point sould be mentioned ere is tat bot models from Sciller [6] and Pani et al. [16] were proposed based on a-emiydrate produced gypsum. Actually te linear model (Eq. (17)) is one special case of te power model (Eq. (18)), i.e. wen te exponent is unity. Te power model gives best predictions in te present study. A critical void fraction value of of te gypsum was obtained ere (1/b 0 in Eqs. (17) (19)). Tis value indicates tat te produced gypsum will lose its strengt at tis void fraction, wic is in line wit Sciller [29] wo reported a critical void fraction value of 0.79 for gypsum. Tis finding also indicates te strengt is independent on te type of te used emiydrate. An extreme strengt (r 0, flexural strengt and compressive strengt), i.e. te strengt of gypsum at zero void fraction, is also derived from te models, as listed in Table 3. We can see tat te Sciller model (Eq. (19)) as a limitation since it only can predict te critical void fraction oter tan te extreme strengt, wile te Exponential model (Eq. (20)) only gives an extreme strengt oter tan te critical void fraction. 4. Conclusions As an environmental friendly and sustainable building material, gypsum is widely used in te construction field. Tis study aims to contribute to a better understanding of te microstructure and related mecanical properties of gypsum produced from b-emiydrate. Te following conclusions can be drawn: 1. Water demand of b-emiydrate was determined using spread flow test, and results sow tat spread flow test is a suitable test metod. Te workability was investigated and te influence of water content on te flowability was analyzed. 2. Te deformation coefficient of te b-emiydrate was derived, te tickness of a water layer around te particles to assure te fluidity of te ydrating system was discussed, and te sape factor of te investigated b-emiydrate was derived. 3. Te ultrasonic wave metod is suitable for b-emiydrate ydration measurement. A relation between te induction time and ending time of b-emiydrate ydration and water content was found. 4. A model was applied to describe te microstructure of te ydrating system, wic was validated by experiments. Subsequently a relation between water amount and te strengt of generated gypsum was derived. Acknowledgements Te autors wis to express teir appreciations to Prof. Dr. H.U. Hummel and Mrs. K. Engelardt from Knauf Gips KG (Germany) for materials supply; Prof. Dr. C.U. Grosse and Mr. F. Lemann from University of Stuttgart (Germany) for executing te ultrasonic measurements. Tey furtermore express teir gratitude to te European Commission (I-SSB Project, Proposal No ) for funding tis researc, as well as to te following sponsors: Bouwdienst Rijkswaterstaat, Graniet-Import Benelux, Kijlstra Betonmortel, Struyk Verwo, Insulinde, Enci, Provincie Overijssel, Rijkswaterstaat Directie Zeeland, A&G maasvlakte, BTE, Alvon Bouwsystemen, V.d. Bosc Beton, Selor, Twee R Recycling, GMB, Scenk Concrete Consultancy, De Mobiele Fabriek, Creative Matc, Intron, Geocem Researc, Icopal, and BN International (cronological order of joining). Tanks are also given to teir colleagues: A.C.J. de Korte, M.M. Ballari, P. Spiesz, M.V.A. Marinescu, A. Lazaro Garcia, and G. Quercia Bianci for teir elp. References [1] Brouwers HJH. A ydration model for Portland cement using te work of Powers and Brownyard, Skokie, Illinois, USA: Portland Cement Association; in press. [2] Eurogypsum, <ttp:// [3] Wirscing F. Calcium sulfate. Weineim: Wiley-VCH Verlag GmbH & Co. KGaA; [4] Amatieu L, Boistelle R. Crystallization kinetics of gypsum from dense suspension of emiydrate water. J Cryst Growt 1988;88: [5] Ridge MJ, Surkevicius H. Hydration of calcium sulpate emiydrate I. Kinetics of te reaction. J Appl Cem 1962;12: [6] Sciller KK. Porosity and strengt of brittle solids (wit particular reference to gypsum). In: Mecanical properties of non-brittle materials. London (UK): Butterworts Scientific Publications; p [7] Lewry AJ, Williamson J. Te Setting of gypsum plaster. Part 1. Te ydration of calcium sulpate emiydrate. J Mater Sci 1994;29: [8] Lancia A, Musmarra D, Prisciandaro M. Measuring induction period for calcium sulpate diydrate precipitation. AICHE J 1999;45(2): [9] Solberg C, Hansen S. Dissolution of CaSO 4 1/2H 2 O and precipitation of CaSO 4 2H 2 O: A kinetic study by syncrotron X-ray powder diffraction. Cem Concr Res 2001;31(4): [10] Gmou A, Eve S, Samdi A, Moussa R, Trica L, Aazzab B, et al. Development and validation of a dimensional variation measurement set-application to te plaster setting. Mater Sci Eng A Struct 2004;372: [11] Sattler H, Bruckner HP. Canges in volume and density during te ydration of gypsum binders as a function of te quantity of water available. ZKG Int 2001;9: [12] Hunger M, Brouwers HJH. Flow analysis of water powder mixtures: application to specific surface area and sape factor. Cem Concr Compos 2009;31: [13] Hunger M. An integral design concept for ecological self-compacting concrete. PD tesis. Eindoven, Te Neterlands: Eindoven University of Tecnology; [14] EN , European Standard. Gypsum binders and gypsum plasters. CEN; 2004 [15] de Korte ACJ, Brouwers HJH. Hydration of modeling of calcium sulpates. In: Al-Mattarne H, Mustapa KN, Nuruddin MP, editors. Proceedings of international conference on construction and building tecnology. Malaysia: Kuala Lumpur; p [16] Pani KK, Niyogi SK, Maitra AK, Roycaudury M. Strengt and elastic modulus of a porous brittle solid: an acousto-ultrasonic study. J Mater Sci 1986;21: [17] Grosse CU, Lemann F. Ultrasound measurements of ydrating of emiydrates. Universitat Stuttgart; personal communication, [18] Reinardt HW, Grosse CU. Continuous monitoring of setting and ardening of mortar and concrete. Constr Build Mater 2004;18(3): [19] Mastersizer 2000 user manual, Malvern Instruments Ltd; p [20] Yu QL, Brouwers HJH, de Korte ACJ. Gypsum ydration: a teoretical and experimental study. In: Fiscer HB, editor. Proceedings of 17t international conference in building materials (Internationale Baustofftagung). Germany: Weimar; p

9 Q.L. Yu, H.J.H. Brouwers / Construction and Building Materials 25 (2011) [21] Brouwers HJH, Radix HJ. Self-compacting concrete: teoretical and experimental study. Cem Concr Res 2005;35: [22] Engelardt K. Pers Commun [23] Marquardt I. Determination of te composition of self-compacting concretes on te basis of te water requirements of te constituent materials presentation of a new mix concept. Betonwerk + Fertigteil tecnik BFT 2002;11: [24] Harker AH, Temple JAG. Velocity and attenuation of ultrasound in suspensions of particles in fluids. J Pys D Appl Pys 1988;21(11): [25] Robeyst N, Gruyaert E, Grosse CU, De Belie N. Monitoring te setting of concrete containing blast-furnace slag by measuring te ultrasonic p-wave velocity. Cem Concr Res 2008;38: [26] Sciller KK. Te course of ydration: its practical importance and teoretical interpretation. J Appl Cem Biotecnol 1974;24: [27] Lewry AJ, Williamson J. Te Setting of gypsum plaster part 2 te development of microstructure and strengt. J Mater Sci 1994;29: [28] Coquard P, Boistelle R, Amatieu L, Barriac P. Hardness, elasticity modulus and flexion strengt of dry set plaster. J Mater Sci 1994;29: [29] Sciller KK. Strengt of igly porous brittle materials. Nature 1957;180:862 3.